[Emc-users] Carousel Component

[Rick Lair]

Good Morning Guys,

Andy,

Had a question on your carousel component, I have skimmed over it, and 
it looks pretty appetizing, but this is set up for a binary or gray-code 
input in relation to the carousel location. I only have two limit 
switches on the carousel, Home Position, and Once Per Tool. Do I need to 
look at integrating something in order to have a better positioning 
signal coming from the carousel in order to use your component?



-- 

Thanks


Rick Lair
Superior Roll & Turning LLC
399 East Center Street
Petersburg MI, 49270
PH: 734-279-1831
FAX: 734-279-1166
www.superiorroll.com


--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

[Emc-users] Carousel Component

[rick]

Good Morning Guys,

Have a question for Andy Pugh,

Finally got to the point of integrating the carousel component into my
machine, and was wondering, how do I connect a signal name to the
parameter pins in the component, so I can use the home related pins in
my ladder logic?

I have the rest of it programmed, I just need to see the home state of
the component, and I was going to use the compare function in
classicladder.

Thanks

Rick

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[Andy Pugh]

> On 10 Jun 2015, at 13:18, Rick Lair  wrote:
> 
> only have two limit 
> switches on the carousel, Home Position, and Once Per Tool. Do I need to 
> look at integrating something in order to have a better positioning 
> signal coming from the carousel in order to use your component?

JT has the same arrangement. It seems common enough that the component should 
grow the ability to handle it. 
I will probably call it "index" feedback. 
--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[Rick Lair]

In trying to keep off of Chris Kelley's thread about his newly running 
machine, I think Pete should look at this component, I think it will 
work straight off for your Cinci, I believe it is binary coded for the 
carousel position, we have a 1996 Sabre-750 in house, with the original 
control on it, that has the same tool carousel on as yours, I can't 
remember what machine you have though, one of the weird names cinci used 
for there machining centers.

Look at this Pete,

Quoting Andy,

I have written a hal module specifically for carousel control.
I hope to put it into mainline LinuxCNC eventually, but at the moment
you can try it out in a simulator config here:
(Full Vismach model with spindle orient and carousel motion)
http://www.linuxcnc.org/index.php/english/forum/24-hal-components/29153-vmc-related-hal-questions?limitstart=0#58732



Andy, have you had a chance to look at that "index" style feedback you 
commented on, regarding our machine?  I really like the ease of use you 
put into that component, and would love to run it on our machine.

Thanks

Rick


On 6/10/2015 9:01 AM, Andy Pugh wrote:
>
>> On 10 Jun 2015, at 13:18, Rick Lair  wrote:
>>
>> only have two limit
>> switches on the carousel, Home Position, and Once Per Tool. Do I need to
>> look at integrating something in order to have a better positioning
>> signal coming from the carousel in order to use your component?
> JT has the same arrangement. It seems common enough that the component should 
> grow the ability to handle it.
> I will probably call it "index" feedback.
> --
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users

-- 

Thanks


Rick Lair
Superior Roll & Turning LLC
399 East Center Street
Petersburg MI, 49270
PH: 734-279-1831
FAX: 734-279-1166
www.superiorroll.com


--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[andy pugh]

On 15 June 2015 at 16:51, Rick Lair  wrote:
> Andy, have you had a chance to look at that "index" style feedback you
> commented on, regarding our machine?

Not yet, I have been busy building a new shed to put the spare bikes
in so that I can actually get at the new lathe to work on that.

-- 
atp
If you can't fix it, you don't own it.
http://www.ifixit.com/Manifesto

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[Rick Lair]

No problem Andy,

Good things come to those who wait :)

I have my hands full again anyway re-tuning this thing.


Rick


On 6/15/2015 12:13 PM, andy pugh wrote:
> On 15 June 2015 at 16:51, Rick Lair  wrote:
>> Andy, have you had a chance to look at that "index" style feedback you
>> commented on, regarding our machine?
> Not yet, I have been busy building a new shed to put the spare bikes
> in so that I can actually get at the new lathe to work on that.
>

-- 

Thanks


Rick Lair
Superior Roll & Turning LLC
399 East Center Street
Petersburg MI, 49270
PH: 734-279-1831
FAX: 734-279-1166
www.superiorroll.com


--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[Rick Lair]

Hey Andy,

Just wondering if you made any headway on the carousel component?

I am still working on a bunch of other stuff on the machine, but was 
just wondering if you had a chance to take a look at it.

Thanks

Rick

On 6/15/2015 12:40 PM, Rick Lair wrote:
> No problem Andy,
>
> Good things come to those who wait :)
>
> I have my hands full again anyway re-tuning this thing.
>
>
> Rick
>
>
> On 6/15/2015 12:13 PM, andy pugh wrote:
>> On 15 June 2015 at 16:51, Rick Lair  wrote:
>>> Andy, have you had a chance to look at that "index" style feedback you
>>> commented on, regarding our machine?
>> Not yet, I have been busy building a new shed to put the spare bikes
>> in so that I can actually get at the new lathe to work on that.
>>
>

-- 

Thanks


Rick Lair
Superior Roll & Turning LLC
399 East Center Street
Petersburg MI, 49270
PH: 734-279-1831
FAX: 734-279-1166
www.superiorroll.com


--
Don't Limit Your Business. Reach for the Cloud.
GigeNET's Cloud Solutions provide you with the tools and support that
you need to offload your IT needs and focus on growing your business.
Configured For All Businesses. Start Your Cloud Today.
https://www.gigenetcloud.com/
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[andy pugh]

On 8 July 2015 at 19:50, Rick Lair  wrote:

> Just wondering if you made any headway on the carousel component?
>

None at all, I have been distracted by not-working motorcycles, not-working
parental legs (putting in a downstairs bathroom) and not-working CNC mills.
Currently I am down to only one working LinuxCNC box, but I think I can do
the carousel stuff in a VM, and tonight is actually quite a good time to
look at it as I am waiting for a motherboard to arrive to fix the mill
controller.


-- 
atp
If you can't fix it, you don't own it.
http://www.ifixit.com/Manifesto
--
Don't Limit Your Business. Reach for the Cloud.
GigeNET's Cloud Solutions provide you with the tools and support that
you need to offload your IT needs and focus on growing your business.
Configured For All Businesses. Start Your Cloud Today.
https://www.gigenetcloud.com/
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[andy pugh]

On 8 July 2015 at 22:17, andy pugh  wrote:

> tonight is actually quite a good time to look at it


Not-working Mac problems now. I am starting to feel unlucky.
No progress today, but I am away from home for the next two weeks working
in Granada, Spain, so might have nothing else to do but software.

-- 
atp
If you can't fix it, you don't own it.
http://www.ifixit.com/Manifesto
--
Don't Limit Your Business. Reach for the Cloud.
GigeNET's Cloud Solutions provide you with the tools and support that
you need to offload your IT needs and focus on growing your business.
Configured For All Businesses. Start Your Cloud Today.
https://www.gigenetcloud.com/
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[andy pugh]

On 30 October 2015 at 13:49,   wrote:

> Finally got to the point of integrating the carousel component into my
> machine, and was wondering, how do I connect a signal name to the
> parameter pins in the component, so I can use the home related pins in
> my ladder logic?

You can't. I am afraid I didn't consider that anyone would want to.

You will need to edit the carousel.comp defnition file to make the
home state into a pin rather than a parameter.

ie change
param r  signed   state = 0 "Current component state";

to
pin out signed   state = 0 "Current component state";

Then sudo halcompile --install carousel.comp  to install the new version.

(Why do we have both pins and parameters? Pins use up shared memory,
parameters use the much less limited ordinary memory).

-- 
atp
If you can't fix it, you don't own it.
http://www.ifixit.com/Manifesto

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[Rick Lair]

Is there any way to compare against that parameter, with out compiling? 
I am using the 2.7.1 package, not building from the command line, or 
will it still work the way I am using linuxcnc, modifying the comp file?




On 10/30/2015 10:25 AM, andy pugh wrote:
> On 30 October 2015 at 13:49,   wrote:
>
>> Finally got to the point of integrating the carousel component into my
>> machine, and was wondering, how do I connect a signal name to the
>> parameter pins in the component, so I can use the home related pins in
>> my ladder logic?
> You can't. I am afraid I didn't consider that anyone would want to.
>
> You will need to edit the carousel.comp defnition file to make the
> home state into a pin rather than a parameter.
>
> ie change
> param r  signed   state = 0 "Current component state";
>
> to
> pin out signed   state = 0 "Current component state";
>
> Then sudo halcompile --install carousel.comp  to install the new version.
>
> (Why do we have both pins and parameters? Pins use up shared memory,
> parameters use the much less limited ordinary memory).
>

-- 

Thanks


Rick Lair
Superior Roll & Turning LLC
399 East Center Street
Petersburg MI, 49270
PH: 734-279-1831
FAX: 734-279-1166
www.superiorroll.com


--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[andy pugh]

On 30 October 2015 at 14:36, Rick Lair  wrote:
> Is there any way to compare against that parameter, with out compiling?

No tidy way. You can shell out to halcmd and parse the returned text

> I am using the 2.7.1 package, not building from the command line, or
> will it still work the way I am using linuxcnc, modifying the comp file?

You can create comps and use halcompile on any LinuxCNC system. You
might need the LinuxCNC-dev package from the package manager, but you
don't need git / sourcecode etc.


-- 
atp
If you can't fix it, you don't own it.
http://www.ifixit.com/Manifesto

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[John Kasunich]

On Fri, Oct 30, 2015, at 10:25 AM, andy pugh wrote:
> 
> (Why do we have both pins and parameters? Pins use up shared memory,
> parameters use the much less limited ordinary memory).

Actually pins and parameters both use shared memory for their metadata
(which is much bigger than the actual pin data, 50-60 bytes vs 4-8 bytes).

We have both because I made a bad decision a long time ago.  I used a
hardware analogy when designing HAL.  Pins are like the inputs and 
outputs of a servo amp.  Parameters are like the adjustment pots on that
amp (gain, etc).  I made a false distinction when I decided that some things
would need to be interconnected and other things wouldn't.  People always
find new ways to do things, and they want to interconnect stuff that I thought
they wouldn't interconnect.  In hindsight, everything should have been a pin.

-- 
  John Kasunich
  jmkasun...@fastmail.fm

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[rick]

Where is the .comp file at?

I can't find the file, debian doesn't have the handy search utility on
the tool bar like ubuntu did, and I have been looking around in the file
system, and don't see anything.


Rick



>  Original Message ----
> Subject: Re: [Emc-users] Carousel Component
> From: andy pugh 
> Date: Fri, October 30, 2015 10:38 am
> To: "Enhanced Machine Controller (EMC)"
> 
> 
> 
> On 30 October 2015 at 14:36, Rick Lair  wrote:
> > Is there any way to compare against that parameter, with out compiling?
> 
> No tidy way. You can shell out to halcmd and parse the returned text
> 
> > I am using the 2.7.1 package, not building from the command line, or
> > will it still work the way I am using linuxcnc, modifying the comp file?
> 
> You can create comps and use halcompile on any LinuxCNC system. You
> might need the LinuxCNC-dev package from the package manager, but you
> don't need git / sourcecode etc.
> 
> 
> -- 
> atp
> If you can't fix it, you don't own it.
> http://www.ifixit.com/Manifesto
> 
> --
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[Mark Wendt]

Command line search:

#>  cd /
#> find . -name  -print

Long before GUI's became so popular, command line utilities were the way to go.

Mark

On Fri, Oct 30, 2015 at 11:14 AM,   wrote:
> Where is the .comp file at?
>
> I can't find the file, debian doesn't have the handy search utility on
> the tool bar like ubuntu did, and I have been looking around in the file
> system, and don't see anything.
>
>
> Rick
>

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[rick]

Thank You Mark,

I found the file with locate, then saw your email,

I changed the lines in the comp file, but am running into a problem on
the compile, I am trying 

"greenmill@greenmill:~/linuxcnc-dev$ sudo halcompile --install
carousel.comp
[sudo] password for greenmill: 
sudo: halcompile: command not found
greenmill@greenmill:~/linuxcnc-dev$"

And you can see the result, any thoughts?

Rick



>  Original Message 
> Subject: Re: [Emc-users] Carousel Component
> From: Mark Wendt 
> Date: Fri, October 30, 2015 11:20 am
> To: "Enhanced Machine Controller (EMC)"
> 
> 
> 
> Command line search:
> 
> #>  cd /
> #> find . -name  -print
> 
> Long before GUI's became so popular, command line utilities were the way to 
> go.
> 
> Mark
> 
> On Fri, Oct 30, 2015 at 11:14 AM,   wrote:
> > Where is the .comp file at?
> >
> > I can't find the file, debian doesn't have the handy search utility on
> > the tool bar like ubuntu did, and I have been looking around in the file
> > system, and don't see anything.
> >
> >
> > Rick
> >
> 
> --
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[Mark Wendt]

Probably means it's not on your path.  You can confirm:

#>  which halcompile

Should show if in your path or not.

If not, use the find command again.

Then, either use the full path, or cd to the directory it's in and
preface the command with a dot slash:

#> sudo ./halcompile yada yada.

On Fri, Oct 30, 2015 at 11:29 AM,   wrote:
> Thank You Mark,
>
> I found the file with locate, then saw your email,
>
> I changed the lines in the comp file, but am running into a problem on
> the compile, I am trying
>
> "greenmill@greenmill:~/linuxcnc-dev$ sudo halcompile --install
> carousel.comp
> [sudo] password for greenmill:
> sudo: halcompile: command not found
> greenmill@greenmill:~/linuxcnc-dev$"
>
> And you can see the result, any thoughts?
>
> Rick
>
>
>
>>  Original Message 
>> Subject: Re: [Emc-users] Carousel Component
>> From: Mark Wendt 
>> Date: Fri, October 30, 2015 11:20 am
>> To: "Enhanced Machine Controller (EMC)"
>> 
>>
>>
>> Command line search:
>>
>> #>  cd /
>> #> find . -name  -print
>>
>> Long before GUI's became so popular, command line utilities were the way to 
>> go.
>>
>> Mark
>>
>> On Fri, Oct 30, 2015 at 11:14 AM,   wrote:
>> > Where is the .comp file at?
>> >
>> > I can't find the file, debian doesn't have the handy search utility on
>> > the tool bar like ubuntu did, and I have been looking around in the file
>> > system, and don't see anything.
>> >
>> >
>> > Rick
>> >
>>
>> --
>> ___
>> Emc-users mailing list
>> Emc-users@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/emc-users
>
> --
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[rick]

greenmill@greenmill:~/linuxcnc-dev/src/hal/components$ locate
carousel.comp
/home/greenmill/.local/share/Trash/files/linuxcnc-dev/src/hal/components/carousel.comp
/home/greenmill/linuxcnc-dev/src/hal/components/carousel.comp
greenmill@greenmill:~/linuxcnc-dev/src/hal/components$ sudo ./halcompile
--install carousel.comp
sudo: ./halcompile: command not found
greenmill@greenmill:~/linuxcnc-dev/src/hal/components$ 




>  Original Message 
> Subject: Re: [Emc-users] Carousel Component
> From: Mark Wendt 
> Date: Fri, October 30, 2015 11:34 am
> To: "Enhanced Machine Controller (EMC)"
> 
> 
> 
> Probably means it's not on your path.  You can confirm:
> 
> #>  which halcompile
> 
> Should show if in your path or not.
> 
> If not, use the find command again.
> 
> Then, either use the full path, or cd to the directory it's in and
> preface the command with a dot slash:
> 
> #> sudo ./halcompile yada yada.
> 
> On Fri, Oct 30, 2015 at 11:29 AM,   wrote:
> > Thank You Mark,
> >
> > I found the file with locate, then saw your email,
> >
> > I changed the lines in the comp file, but am running into a problem on
> > the compile, I am trying
> >
> > "greenmill@greenmill:~/linuxcnc-dev$ sudo halcompile --install
> > carousel.comp
> > [sudo] password for greenmill:
> > sudo: halcompile: command not found
> > greenmill@greenmill:~/linuxcnc-dev$"
> >
> > And you can see the result, any thoughts?
> >
> > Rick
> >
> >
> >
> >>  Original Message 
> >> Subject: Re: [Emc-users] Carousel Component
> >> From: Mark Wendt 
> >> Date: Fri, October 30, 2015 11:20 am
> >> To: "Enhanced Machine Controller (EMC)"
> >> 
> >>
> >>
> >> Command line search:
> >>
> >> #>  cd /
> >> #> find . -name  -print
> >>
> >> Long before GUI's became so popular, command line utilities were the way 
> >> to go.
> >>
> >> Mark
> >>
> >> On Fri, Oct 30, 2015 at 11:14 AM,   wrote:
> >> > Where is the .comp file at?
> >> >
> >> > I can't find the file, debian doesn't have the handy search utility on
> >> > the tool bar like ubuntu did, and I have been looking around in the file
> >> > system, and don't see anything.
> >> >
> >> >
> >> > Rick
> >> >
> >>
> >> --
> >> ___
> >> Emc-users mailing list
> >> Emc-users@lists.sourceforge.net
> >> https://lists.sourceforge.net/lists/listinfo/emc-users
> >
> > --
> > ___
> > Emc-users mailing list
> > Emc-users@lists.sourceforge.net
> > https://lists.sourceforge.net/lists/listinfo/emc-users
> 
> --
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[Mark Wendt]

You have to use locate or find to find out where "halcompile" is.  Or
you may have to install the dev package if it's not on your machine.

Mark

On Fri, Oct 30, 2015 at 11:47 AM,   wrote:
>
>
>
> greenmill@greenmill:~/linuxcnc-dev/src/hal/components$ locate
> carousel.comp
> /home/greenmill/.local/share/Trash/files/linuxcnc-dev/src/hal/components/carousel.comp
> /home/greenmill/linuxcnc-dev/src/hal/components/carousel.comp
> greenmill@greenmill:~/linuxcnc-dev/src/hal/components$ sudo ./halcompile
> --install carousel.comp
> sudo: ./halcompile: command not found
> greenmill@greenmill:~/linuxcnc-dev/src/hal/components$
>
>
>
>
>> ---- Original Message 
>> Subject: Re: [Emc-users] Carousel Component
>> From: Mark Wendt 
>> Date: Fri, October 30, 2015 11:34 am
>> To: "Enhanced Machine Controller (EMC)"
>> 
>>
>>
>> Probably means it's not on your path.  You can confirm:
>>
>> #>  which halcompile
>>
>> Should show if in your path or not.
>>
>> If not, use the find command again.
>>
>> Then, either use the full path, or cd to the directory it's in and
>> preface the command with a dot slash:
>>
>> #> sudo ./halcompile yada yada.
>>
>> On Fri, Oct 30, 2015 at 11:29 AM,   wrote:
>> > Thank You Mark,
>> >
>> > I found the file with locate, then saw your email,
>> >
>> > I changed the lines in the comp file, but am running into a problem on
>> > the compile, I am trying
>> >
>> > "greenmill@greenmill:~/linuxcnc-dev$ sudo halcompile --install
>> > carousel.comp
>> > [sudo] password for greenmill:
>> > sudo: halcompile: command not found
>> > greenmill@greenmill:~/linuxcnc-dev$"
>> >
>> > And you can see the result, any thoughts?
>> >
>> > Rick
>> >
>> >
>> >
>> >>  Original Message 
>> >> Subject: Re: [Emc-users] Carousel Component
>> >> From: Mark Wendt 
>> >> Date: Fri, October 30, 2015 11:20 am
>> >> To: "Enhanced Machine Controller (EMC)"
>> >> 
>> >>
>> >>
>> >> Command line search:
>> >>
>> >> #>  cd /
>> >> #> find . -name  -print
>> >>
>> >> Long before GUI's became so popular, command line utilities were the way 
>> >> to go.
>> >>
>> >> Mark
>> >>
>> >> On Fri, Oct 30, 2015 at 11:14 AM,   wrote:
>> >> > Where is the .comp file at?
>> >> >
>> >> > I can't find the file, debian doesn't have the handy search utility on
>> >> > the tool bar like ubuntu did, and I have been looking around in the file
>> >> > system, and don't see anything.
>> >> >
>> >> >
>> >> > Rick
>> >> >
>> >>
>> >> --
>> >> ___
>> >> Emc-users mailing list
>> >> Emc-users@lists.sourceforge.net
>> >> https://lists.sourceforge.net/lists/listinfo/emc-users
>> >
>> > --
>> > ___
>> > Emc-users mailing list
>> > Emc-users@lists.sourceforge.net
>> > https://lists.sourceforge.net/lists/listinfo/emc-users
>>
>> --
>> ___
>> Emc-users mailing list
>> Emc-users@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/emc-users
>
> --
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[andy pugh]

On 30 October 2015 at 15:29,   wrote:
> sudo: halcompile: command not found

Do you have the linuxcnc-dev package installed?

halcompile used to be called comp, but I think that your version is
new enough for it to be halcompile.

I am slightly surprised that you have the carousel.comp file,
actually. And that you appear to have a linuxcnc-dev directory.

-- 
atp
If you can't fix it, you don't own it.
http://www.ifixit.com/Manifesto

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[Rick Lair]

I don't have the dev package installed, I just downloaded it, to get to 
the remap stuff.

On 10/30/2015 12:01 PM, andy pugh wrote:
> On 30 October 2015 at 15:29,   wrote:
>> sudo: halcompile: command not found
> Do you have the linuxcnc-dev package installed?
>
> halcompile used to be called comp, but I think that your version is
> new enough for it to be halcompile.
>
> I am slightly surprised that you have the carousel.comp file,
> actually. And that you appear to have a linuxcnc-dev directory.
>

-- 

Thanks


Rick Lair
Superior Roll & Turning LLC
399 East Center Street
Petersburg MI, 49270
PH: 734-279-1831
FAX: 734-279-1166
www.superiorroll.com


--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[rick]

This what locate returns for halcompile,


greenmill@greenmill:~$ locate halcompile
/home/greenmill/.local/share/Trash/files/linuxcnc-dev/docs/man/man1/halcompile.1
/home/greenmill/.local/share/Trash/files/linuxcnc-dev/src/hal/utils/halcompile.g
/home/greenmill/linuxcnc-dev/docs/man/man1/halcompile.1
/home/greenmill/linuxcnc-dev/src/hal/utils/halcompile.g


>  Original Message 
> Subject: Re: [Emc-users] Carousel Component
> From: andy pugh 
> Date: Fri, October 30, 2015 12:01 pm
> To: "Enhanced Machine Controller (EMC)"
> 
> 
> 
> On 30 October 2015 at 15:29,   wrote:
> > sudo: halcompile: command not found
> 
> Do you have the linuxcnc-dev package installed?
> 
> halcompile used to be called comp, but I think that your version is
> new enough for it to be halcompile.
> 
> I am slightly surprised that you have the carousel.comp file,
> actually. And that you appear to have a linuxcnc-dev directory.
> 
> -- 
> atp
> If you can't fix it, you don't own it.
> http://www.ifixit.com/Manifesto
> 
> --
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[andy pugh]

On 30 October 2015 at 16:10, Rick Lair  wrote:
> I don't have the dev package installed, I just downloaded it, to get to
> the remap stuff.

There is scope for confusion here. linuxcnc-dev is a package that can
be downloaded from the package manager, either through the synaptic
gui or with
sudo apt-get install linuxcnc-dev

However, if you download the LinuxCNC source code, that generally ends
up on a folder also called linuxcnc-dev. But you can have either one
without the other.

-- 
atp
If you can't fix it, you don't own it.
http://www.ifixit.com/Manifesto

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[Mark Wendt]

>From the looks of things, and from Andy mentioned, it doesn't look
like you have the linuxcnc-dev package installed.  Installing that
should fix this problem.

Mark

On Fri, Oct 30, 2015 at 1:03 PM,   wrote:
> This what locate returns for halcompile,
>
>
> greenmill@greenmill:~$ locate halcompile
> /home/greenmill/.local/share/Trash/files/linuxcnc-dev/docs/man/man1/halcompile.1
> /home/greenmill/.local/share/Trash/files/linuxcnc-dev/src/hal/utils/halcompile.g
> /home/greenmill/linuxcnc-dev/docs/man/man1/halcompile.1
> /home/greenmill/linuxcnc-dev/src/hal/utils/halcompile.g
>
>
>>  Original Message ----
>> Subject: Re: [Emc-users] Carousel Component
>> From: andy pugh 
>> Date: Fri, October 30, 2015 12:01 pm
>> To: "Enhanced Machine Controller (EMC)"
>> 
>>
>>
>> On 30 October 2015 at 15:29,   wrote:
>> > sudo: halcompile: command not found
>>
>> Do you have the linuxcnc-dev package installed?
>>
>> halcompile used to be called comp, but I think that your version is
>> new enough for it to be halcompile.
>>
>> I am slightly surprised that you have the carousel.comp file,
>> actually. And that you appear to have a linuxcnc-dev directory.
>>
>> --
>> atp
>> If you can't fix it, you don't own it.
>> http://www.ifixit.com/Manifesto
>>
>> --
>> ___
>> Emc-users mailing list
>> Emc-users@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/emc-users
>
> --
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[rick]

Hey guys,  

I will have to double check on Monday, but I am going to have to say that it is 
not installed, but just a folder in my home directory. 

As for the component in general, I went ahead with some test and tune, looks 
awesome for sure. Only question I have though, in regards to the homing 
sequence, if the Carousel is not homed, and I set "enable" true, it should stop 
rotating when it sees the sense.0 pin "home position" go true, then report back 
out on the carousel.0.current-position as being pocket 1 correct? And this 
should be done with out telling it go to pocket 1 on the 
carousel.0.pocket-number input during the homing routine?  

Reason I ask, the only way I could get it to stop on the home position switch " 
pocket 1" during homing, was to use an assignment block in my ladder logic that 
I set to "1", at the same time I start the carousel homing sequence. Then it 
works as intended I believe. 

Rick 

 Original Message ----
Subject: Re: [Emc-users] Carousel Component
From: "andy pugh" 
Date: Fri, October 30, 2015 1:22 pm
To: "Enhanced Machine Controller (EMC)" 

On 30 October 2015 at 16:10, Rick Lair  wrote:
> I don't have the dev package installed, I just downloaded it, to get to
> the remap stuff.

There is scope for confusion here. linuxcnc-dev is a package that can
be downloaded from the package manager, either through the synaptic
gui or with
sudo apt-get install linuxcnc-dev

However, if you download the LinuxCNC source code, that generally ends
up on a folder also called linuxcnc-dev. But you can have either one
without the other.

-- 
atp
If you can't fix it, you don't own it.
http://www.ifixit.com/Manifesto

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users


--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[andy pugh]

On 31 October 2015 at 02:59,   wrote:
>if the Carousel is not homed, and I set "enable" true, it should stop rotating 
>when it sees the sense.0 pin "home position" go true, then report back out on 
>the carousel.0.current-position as being pocket 1 correct?

It needs to see both the index and the pocket pin, ie it will consider
itself homed when it sees sense0 and sense1 go true.

> And this should be done with out telling it go to pocket 1 on the 
> carousel.0.pocket-number input during the homing routine?

No, it will "stick" in state 1 until the pocket-number pin is >=1 and
<= num-pockets.

> Reason I ask, the only way I could get it to stop on the home position switch 
> " pocket 1" during homing, was to use an assignment block in my ladder logic 
> that I set to "1", at the same time I start the carousel homing sequence. 
> Then it works as intended I believe.

You need to give it a pocket number to go to before it will get to the
"finished" state. There is probably a bit of an error there, as
setting enable in the either the homed or un-homed state without
providing a pocket number will leave it "stuck" in state 1, and it
will stay there even if enable goes false.

So, if you want to home the carousel as part of the homing sequence
you currently have to supply a pocket number to go to during that
sequence. I should probably change things slightly so that enable
going to false always takes the component back to state 0.

-- 
atp
If you can't fix it, you don't own it.
http://www.ifixit.com/Manifesto

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[Gene Heskett]

On Saturday 31 October 2015 20:56:53 Gene Heskett wrote:

> On Saturday 31 October 2015 06:42:25 andy pugh wrote:
> > On 31 October 2015 at 02:59,   wrote:
> > >if the Carousel is not homed, and I set "enable" true, it should
> > > stop rotating when it sees the sense.0 pin "home position" go
> > > true, then report back out on the carousel.0.current-position as
> > > being pocket 1 correct?
> >
> > It needs to see both the index and the pocket pin, ie it will
> > consider itself homed when it sees sense0 and sense1 go true.
> >
> > > And this should be done with out telling it go to pocket 1 on the
> > > carousel.0.pocket-number input during the homing routine?
> >
> > No, it will "stick" in state 1 until the pocket-number pin is >=1
> > and <= num-pockets.
> >
> > > Reason I ask, the only way I could get it to stop on the home
> > > position switch " pocket 1" during homing, was to use an
> > > assignment block in my ladder logic that I set to "1", at the same
> > > time I start the carousel homing sequence. Then it works as
> > > intended I believe.
> >
> > You need to give it a pocket number to go to before it will get to
> > the "finished" state. There is probably a bit of an error there, as
> > setting enable in the either the homed or un-homed state without
> > providing a pocket number will leave it "stuck" in state 1, and it
> > will stay there even if enable goes false.
> >
> > So, if you want to home the carousel as part of the homing sequence
> > you currently have to supply a pocket number to go to during that
> > sequence. I should probably change things slightly so that enable
> > going to false always takes the component back to state 0.
>
> Where can I get the source Andy?
>
> What I have in mind will take quite some interaction between the
> machine and this component.
>
> What I have in mind would follow this general outline of action when
> an M6 Tx is encountered.
>
> 1.The first thing is to run the head to the top of the post.
>
> 2. When the head has arrived, trigger a motor to swing the changer
> from its parking position to the right of the post, bringing its wheel
> into position under the spindle. This pocket would belong to the
> loaded tool, which if its in the spindle, would be empty.
>
> 3. At this point, start the spindle at about 20 revs in fwd.  At the
> same time, run another motor attached to the gearshift, while the
> wrench is looking to grab the drawbar.
>
> 4. I'll have a switch there to tell LCNC or the carousel that the
> wrench is engaged. Give it about half a turn of slack.  When the motor
> is stalled against the wrench, reverse the motor to fwd and raise the
> speed to 120% for perhaps 1/2 second, then back to 20 rpms still in
> fwd for perhaps another second, which should drop the TTS adaptor into
> the empty pocket.
>
> 5.  Raise the head about 2.25" for turning clearance, then tell
> carousel what tool pocket is needed next & turn the wheel until that
> pocket and tool is under the spindle.  Lower the head, running in
> reverse until its pushing gently on the TTS big ring or has engaged
> the drawbar, possibly clearing the bottom of the tool pocket by a
> couple thou.
>
> 6. If this condition can be detected, (switch on the wrench?) hit the
> motor with about 40% fwd power for a very short (.1 seconds?) time to
> back it away from the stop, then bang it with 30% until its stalled,
> giving it a bit of a run before it hits the stop, which will snug the
> R8 up on the adapter quite nicely.
>
> 7. Run the head up enough to lift the adapter out of its pocket, and
> swing the whole thing back out of the way to the right of the post.
>
> 8. Carousel will obviously need help since it will also be controlling
> Z movements in addition to whailing away on the spindle motor, AND be
> running 2 other motors at the same time.  All in the proper sequence.

Correction, 4 motors or? because moving the carousel will need two and 
the drawbar wrench is 3, and the head into low gear is 4. All 2 way 
motions.

> Homing it should consist of finding pocket 1, and then recording any
> empty pockets.
>
> I have a handfull of 52oz/in steppers, one of which could run the
> swing mechanism, and a handfull of geared steppers that could do away
> with the pocket motions by just issueing the correct # of steps to
> take it to the next pocket.  Ditto for the swing power, one of these
> geared down steppers could be enabled to run until it hits the switch,
> 2, one in the parked position & one to stop it under the spindle. 
> With my 500hz charge pump as stepper drive, I could probably hack that 
up in the .hal file.
>
> But that will still need another 4 or 6 bit pins out of carousel to 
steer & sequence all the .hal stuffs.
>
> A scan of the wooden wheel I made today is attached.  The smaller,
> fuzzy inside hole just clears the TTS nut.  Carved in 1/2" alu, it
> should work well once the motors & driver electronics are sorted.
>
> Cheers, Gene Heskett


Cheers, Gene Heskett
-- 
"There 

Re: [Emc-users] Carousel Component

[Gene Heskett]

On Saturday 31 October 2015 21:06:03 Gene Heskett wrote:

> On Saturday 31 October 2015 20:56:53 Gene Heskett wrote:
> > On Saturday 31 October 2015 06:42:25 andy pugh wrote:
> > > On 31 October 2015 at 02:59,   wrote:
> > > >if the Carousel is not homed, and I set "enable" true, it should
> > > > stop rotating when it sees the sense.0 pin "home position" go
> > > > true, then report back out on the carousel.0.current-position as
> > > > being pocket 1 correct?
> > >
> > > It needs to see both the index and the pocket pin, ie it will
> > > consider itself homed when it sees sense0 and sense1 go true.
> > >
> > > > And this should be done with out telling it go to pocket 1 on
> > > > the carousel.0.pocket-number input during the homing routine?
> > >
> > > No, it will "stick" in state 1 until the pocket-number pin is >=1
> > > and <= num-pockets.
> > >
> > > > Reason I ask, the only way I could get it to stop on the home
> > > > position switch " pocket 1" during homing, was to use an
> > > > assignment block in my ladder logic that I set to "1", at the
> > > > same time I start the carousel homing sequence. Then it works as
> > > > intended I believe.
> > >
> > > You need to give it a pocket number to go to before it will get to
> > > the "finished" state. There is probably a bit of an error there,
> > > as setting enable in the either the homed or un-homed state
> > > without providing a pocket number will leave it "stuck" in state
> > > 1, and it will stay there even if enable goes false.
> > >
> > > So, if you want to home the carousel as part of the homing
> > > sequence you currently have to supply a pocket number to go to
> > > during that sequence. I should probably change things slightly so
> > > that enable going to false always takes the component back to
> > > state 0.
> >
> > Where can I get the source Andy?
> >
> > What I have in mind will take quite some interaction between the
> > machine and this component.
> >
> > What I have in mind would follow this general outline of action when
> > an M6 Tx is encountered.
> >
> > 1.The first thing is to run the head to the top of the post.
> >
> > 2. When the head has arrived, trigger a motor to swing the changer
> > from its parking position to the right of the post, bringing its
> > wheel into position under the spindle. This pocket would belong to
> > the loaded tool, which if its in the spindle, would be empty.
> >
> > 3. At this point, start the spindle at about 20 revs in fwd.  At the
> > same time, run another motor attached to the gearshift, while the
> > wrench is looking to grab the drawbar.
> >
> > 4. I'll have a switch there to tell LCNC or the carousel that the
> > wrench is engaged. Give it about half a turn of slack.  When the
> > motor is stalled against the wrench, reverse the motor to fwd and
> > raise the speed to 120% for perhaps 1/2 second, then back to 20 rpms
> > still in fwd for perhaps another second, which should drop the TTS
> > adaptor into the empty pocket.
> >
> > 5.  Raise the head about 2.25" for turning clearance, then tell
> > carousel what tool pocket is needed next & turn the wheel until that
> > pocket and tool is under the spindle.  Lower the head, running in
> > reverse until its pushing gently on the TTS big ring or has engaged
> > the drawbar, possibly clearing the bottom of the tool pocket by a
> > couple thou.
> >
> > 6. If this condition can be detected, (switch on the wrench?) hit
> > the motor with about 40% fwd power for a very short (.1 seconds?)
> > time to back it away from the stop, then bang it with 30% until its
> > stalled, giving it a bit of a run before it hits the stop, which
> > will snug the R8 up on the adapter quite nicely.
> >
> > 7. Run the head up enough to lift the adapter out of its pocket, and
> > swing the whole thing back out of the way to the right of the post.
> >
> > 8. Carousel will obviously need help since it will also be
> > controlling Z movements in addition to whailing away on the spindle
> > motor, AND be running 2 other motors at the same time.  All in the
> > proper sequence.
>
> Correction, 4 motors or? because moving the carousel will need two and
> the drawbar wrench is 3, and the head into low gear is 4. All 2 way
> motions.
>
> > Homing it should consist of finding pocket 1, and then recording any
> > empty pockets.
> >
> > I have a handfull of 52oz/in steppers, one of which could run the
> > swing mechanism, and a handfull of geared steppers that could do
> > away with the pocket motions by just issueing the correct # of steps
> > to take it to the next pocket.  Ditto for the swing power, one of
> > these geared down steppers could be enabled to run until it hits the
> > switch, 2, one in the parked position & one to stop it under the
> > spindle. With my 500hz charge pump as stepper drive, I could
> > probably hack that
>
> up in the .hal file.
>
> > But that will still need another 4 or 6 bit pins out of carousel to
>
> steer & sequence all the .hal stuffs.
>
> >

Re: [Emc-users] Carousel Component

[Jon Elson]

On 10/31/2015 10:15 PM, Gene Heskett wrote:
>
> I have 3, 8 wire motors, I presume intended for direct 24 volt operation
> since the coil ohms is 145, and the current/phase matched that at 166
> mills per coil.
Ugh, these motors will have HORRIBLE performance.  Maybe 
they could work for a tool carousel, but even there I think 
you might be disappointed.  Just take a guess at the coil 
inductance, and you can see that it would take a long time 
for the current to ramp up.

My guess is these are 120 V synchronous motors, to be run 
off a 60 Hz 120 V AC line with a phase shift capacitor.  
They were developed by Superior Electric in the 1960's to 
run powerstats (Variacs) remotely from a lighting control 
board.  They were called Slo-Syn motors.  People found they 
made passable stepper motors for positioning use, and pretty 
soon a whole new industry popped up.

Jon


--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[Gene Heskett]

On Saturday 31 October 2015 23:40:41 Jon Elson wrote:

> On 10/31/2015 10:15 PM, Gene Heskett wrote:
> > I have 3, 8 wire motors, I presume intended for direct 24 volt
> > operation since the coil ohms is 145, and the current/phase matched
> > that at 166 mills per coil.
>
> Ugh, these motors will have HORRIBLE performance.  Maybe
> they could work for a tool carousel, but even there I think
> you might be disappointed.  Just take a guess at the coil
> inductance, and you can see that it would take a long time
> for the current to ramp up.
>
> My guess is these are 120 V synchronous motors, to be run
> off a 60 Hz 120 V AC line with a phase shift capacitor.
> They were developed by Superior Electric in the 1960's to
> run powerstats (Variacs) remotely from a lighting control
> board.  They were called Slo-Syn motors.  People found they
> made passable stepper motors for positioning use, and pretty
> soon a whole new industry popped up.
>
> Jon

I am familiar with slow syn's, we used them in the pan & tilt mechanisms 
that were on the Trieste when it went down into the mohole with Lt Walsh 
and Jacques Cousteau in it, in Feb 1960.  They worked well.  These are 
not slowsyn's, they are an early double stack stepper, 15 degrees a 
step, but with a substantial 15/1 gearbox attached, so they move 1 
degree per full step. The 500 hz charge pump would likely need a /8, 
giving a 62.5 hz effective drive rate.  Thats still 62.5 degrees a 
second at the output shaft and actually a fairly speedy rotation if its 
driving the wheel I posted a pix of which is 72 degrees per pocket.  
That would get from tool pocket to tool pocket in 1.25 seconds or all 
the way around in nominally 6 seconds.  A further /4 to a 2nd copy 
should move the whole thing in and out of position as thats going to be 
a couple lbs to move a foot or so & might need a PID module to profile 
the accels/decels.

A 3rd one would have plenty of power to run the heads gearshift shaft ( 
took the detent ball assembly out, turns easily & smoothly now, but 
would also be hanging out if a direct drive. That leaves a solenoid to 
engage the drawbar holding wrench as the final problem. I see that as a 
socket with a disk, with a sturdy flag sticking out, to turn with the 
spindle full time, but with a solenoid driven pin to block the rotation 
when its desired to hold the drawbar to unlock the tool, or lock in the 
new tool. With a substantial flag on the disk/wrench the solenoid could 
stick up a pin to block & lock the socket, and yet give most of a turn 
so that when it hits the pin, the drawbar will be loosened or tightened 
depending on the direction its turning when it hits the pin.

Pocket to pocket moves would simply be issuing the correct number of 
steps to get to the next pocket, in this case 72.  If thats not the 
desired pocket, do it again until it is.

The devil of course is in the details.  Film at 11 of course if & when 
its working.  I haven't even checked to see if I'd have to offset the 
tool pocket rotation motor in order to get clearance to the head when 
its in the active position. I fully expect I'll have to unless I put 
that motor on the bottom. That would also solve the potential 
interference between the gearshift motor and the tool pocket motor too. 
Since its not designed yet, its just a design detail.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page 

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[andy pugh]

On 1 November 2015 at 00:56, Gene Heskett  wrote:
> Where can I get the source Andy?

The source is in the usual place, git.linuxcnc.org
http://git.linuxcnc.org/gitweb?p=linuxcnc.git;a=blob_plain;f=src/hal/components/carousel.comp;hb=HEAD

> What I have in mind will take quite some interaction between the machine
> and this component.

Take a look at the carousel demo sim config. That builds a conplete
sequence with G-code and the carousel component. The demo also
includes spindle orientation and a Vismach model.

sim->axis->Vismach->vmc_toolchange

-- 
atp
If you can't fix it, you don't own it.
http://www.ifixit.com/Manifesto

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[Gene Heskett]

On Sunday 01 November 2015 18:41:30 andy pugh wrote:

> On 1 November 2015 at 00:56, Gene Heskett  wrote:
> > Where can I get the source Andy?
>
> The source is in the usual place, git.linuxcnc.org
> http://git.linuxcnc.org/gitweb?p=linuxcnc.git;a=blob_plain;f=src/hal/c
>omponents/carousel.comp;hb=HEAD
>
> > What I have in mind will take quite some interaction between the
> > machine and this component.
>
> Take a look at the carousel demo sim config. That builds a conplete
> sequence with G-code and the carousel component. The demo also
> includes spindle orientation and a Vismach model.
>
> sim->axis->Vismach->vmc_toolchange

I can see sim>axis>vismach> but vmc_toolchange doesn't appear to be a 
legit startup option.

So how do I run that demo?  Keeping in mind that I have shuffled some 
stuffs around in my sim axis setup here in order to simulate probe 
contacts for the G38.2 command.

I also tried it on the machine that will need it, with the same results, 
nothing in the vismach tree is executable.

I am no doubt missing the point. :(

Thanks Andy

Thanks Andy.
Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page 

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[andy pugh]

On 2 November 2015 at 00:22, Gene Heskett  wrote:
> I can see sim>axis>vismach> but vmc_toolchange doesn't appear to be a
> legit startup option.

There are two sub-configs in there.

-- 
atp
If you can't fix it, you don't own it.
http://www.ifixit.com/Manifesto

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[Gene Heskett]

On Sunday 01 November 2015 19:28:20 andy pugh wrote:

> On 2 November 2015 at 00:22, Gene Heskett  wrote:
> > I can see sim>axis>vismach> but vmc_toolchange doesn't appear to be
> > a legit startup option.
>
> There are two sub-configs in there.

Found it!  That is a quite simplified version that assumes some unseen
mechanism to run the drawbar, and a straight out the side of the wheel
exit strategy.  Mine is not going to be that simple.  The drawbar will
be operated, but will use spindle power to loosen or tighten the drawbolt
using spindle power while the drawbolt is held stationary.

So the carousel module will need an extra pin to run the gearshift to
put it in low so the torque will be sufficient, another extra pin to
run the drawbar lock, a port with a programmable by setp means, to issue
probably at least 2, and as high as 4, signed values to be fed to the
spindle motors pwmgen in order for it to function as the air wrench some
use.

It will also need a signed value pin for running the z axis, programmable
by setp statements in the hal file. All at the correct time of course.

It might be better done in ladder logic.  So far its all a "what if".

I did some quick measurements this afternoon, and I don't think there
would be adequate room under the tool wheel for the motor, so it will
have to be on top, probably high enough the head will never hit it.

And offset away from the side of the head with a small set of spur gears
at the wheel.  Or the wheel will need to grow some diameter just to get
the offset from the shaft coming down the side of the head, and still get
the center of the pocket at the center of the spindle, in which case it
may as well grow more tool pockets.  A minimum practical size IOW.

So I need to get that measurement, and add a 1/2" shaft to get the circle
diameter for the pockets.

The head is 4.75" wide, and I figure a 1/2" shaft, so the pocket circle
is then 5.25/2 or 2.625" radius. That gives me room in that circle for
10 pockets. I am also going to need a longer Y bolt unless I cut the
outside off at a radius that still retains a TTS adapter. I think I can.

But the outer, beyond the circle you can see in this code, will exceed
the length of my Y screw by about 1.5" as I have only 5.4719" between
the limits.  Some of that loss I could fix as I added a bellows in front
of the table to keep swarf out of the Y screw as it is exposed when the
table is all the way to the rear.  I can gain a bit of travel there, but
nowhere near enough for this.

The code:
(; carve a wheel to hold TTS tool adapters )

G20 G17 G90
G0 z0.1
g0 x0 y0 ( starter locations, move to suit )
( touch off at safe centers ) 
(Useful subroutine)
o100 sub
( make wherever it is at, home for duration of subroutine )
G92 x0.0 y0.000
G0 Z0.25000 (clear workpiece for tool comp moves )
G41
G1 F60 X.3000
G1 X0
G0 Z 0.02500 (clear workpiece )
( drill tool counterbore )
#<_z_depth> = 0. (init it for this subs duration)
( drill halfway thru tool holder plate at tool OD of 1.501" )
G1 F40 Z#<_z_depth>
G1 F40 X#<_pocket_rad>
G3 X#<_pocket_rad> Y-0. I-#<_pocket_rad> J-0. Z-0.20 P10
G3 X#<_pocket_rad> Y-0. I-#<_pocket_rad> J0.000 (for flat bottomed 
recesss )

#<_z_depth> = 0.02500 ( otherwise first cut on center is brutal )
( now drill the rest of the way thru the plate )
#<_z_depth> = [#<_z_depth> -.025000]
G1 F40 Z#<_z_depth>
G1 X#<_grip_rad>
G3 X#<_grip_rad> Y0. I-#<_grip_rad> J-0.0 Z-0.4700 P 20
G3 X#<_grip_rad> Y0. I-#<_grip_rad> J-0.0 (leave flat bottom )
G1 X0 Y0.3 Z0.1
G40 (turn off tool comp)
G0 Y0
G92.2

o100 endsub

( setup to 10 identical circle carvings at 36 degree intervals )
#<_pocket_rad>  = 0.75050
#<_grip_rad>= 0.66900  (clears the hex collet nut)
#<_circle_rad>  = 2.62500  (radius of pocket pattern)
#<_angle_inc>   = 36.
#<_this_angle>  = 0.

o200 DO
( position machine to center of pocket )
G1 F40 Z0.1000
G1 F40 X[sin[#<_this_angle>] * #<_circle_rad>] Y[cos[#<_this_angle>] * 
#<_circle_rad>]
s2500
M6 T1
M3 (start spindle)
M8 (start vacuum)
o100 CALL

G1 F40 Z0.05000 ( clear workpiece )
( calculate new angle )
#<_this_angle>  = [#<_this_angle> + #<_angle_inc>]
o200 WHILE [#<_this_angle> lt 360.]

G0 Z0.01 ( clear workpiece )
G40
G0 X0.0 Y0.0 (make sure centered on original center )
( Now, needs a 1/2" hole for the axle )
G1 X0.12500
( note 0.470 is a piece of plywood, real alu will be 0.5000" thick)
G3 X0.12500 Y0. I-0.12500 J0. Z-0.4700 P20
G3 X0.12500 Y0. I-0.12500 J0. (last pass for flat bottom! )
G0 Z0.1
G0 X0. Y0.

( now, need ratchet catching slots? maybe not )
(#<_ratchet_rad>=   2.)
(#<_this_angle> =   36) (space these between tool pockets )

(o300 DO)
(G1 F40 X[SIN[#<_this_angle>] * #<_ratchet_rad>] Y[COS[#<_this_angle>] * 
#<_ratchet_rad>])
(G1 F10 Z-0.47)
(G0 Z0.0500)

Re: [Emc-users] Carousel Component

[andy pugh]

On 2 November 2015 at 02:26, Gene Heskett  wrote:
> Found it!  That is a quite simplified version that assumes some unseen
> mechanism to run the drawbar, and a straight out the side of the wheel
> exit strategy.

Yes, that is how the demo is set up, but the actual sequence is set by
a G-code subroutine, so is reasonably easy to change.

All the carousel module does is interpret carousel position feeback
and  issue motor movement commands for a tool carousel. Everything
else is handled elsewhere. In the examples it is by G-code, but it
would interface with Ladder just as easily.

As you need to move the head up and down and issue spindle speed
commands I think that you will find G-code more convenient, moving the
head from ladder is more difficult.

Also, that example includes spindle orient. You might want to consider
a spindle orient commands rather than a spindle start command.
The docs say that M19 is limited to R-values of 0-360 but you can
force movement in one direction or the other. I think that a string of
180 degree moves could be used to unscrew the drawbar by a requested
amount. It also has a built-in timeout in case things stick. The
advantage would be that you could ensure that the drabar thread
remained engaged.

My milling machine has a sturdy cast cap over a clllar the drawbar on
the horizontal spindle. If you unscrew far enough it actively ejects
the taper.

-- 
atp
If you can't fix it, you don't own it.
http://www.ifixit.com/Manifesto

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[rick]

>  Original Message 
> Subject: Re: [Emc-users] Carousel Component
> From: andy pugh 
> Date: Fri, October 30, 2015 1:22 pm
> To: "Enhanced Machine Controller (EMC)"
> 
> 
> 
> On 30 October 2015 at 16:10, Rick Lair  wrote:
> > I don't have the dev package installed, I just downloaded it, to get to
> > the remap stuff.
> 
> There is scope for confusion here. linuxcnc-dev is a package that can
> be downloaded from the package manager, either through the synaptic
> gui or with
> sudo apt-get install linuxcnc-dev
> 

I am getting an error trying this ^^,

Here is the error,

greenmill@greenmill:~$ sudo apt-get install linuxcnc-dev
[sudo] password for greenmill: 
Reading package lists... Done
Building dependency tree   
Reading state information... Done
Package linuxcnc-dev is not available, but is referred to by another
package.
This may mean that the package is missing, has been obsoleted, or
is only available from another source

E: Package 'linuxcnc-dev' has no installation candidate
greenmill@greenmill:~$ 


Is this because of running uspace?



Rick 




> However, if you download the LinuxCNC source code, that generally ends
> up on a folder also called linuxcnc-dev. But you can have either one
> without the other.
> 
> -- 
> atp
> If you can't fix it, you don't own it.
> http://www.ifixit.com/Manifesto
> 
> --
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[andy pugh]

On 2 November 2015 at 13:16,   wrote:

> What does the carousel.0.state number define, what does it mean?

It shows where the system is in the alignment sequence. It will only
really make sense looking at the code:
http://git.linuxcnc.org/gitweb?p=linuxcnc.git;a=blob;f=src/hal/components/carousel.comp;h=c38343c24b9b21e743e1e7dd7a4d57e7cfec74be;hb=HEAD#l150
Starting at line 150.

The comment on each state says what it is doing in that state, and
then it moves from one state to another on the basis of comparison in
that state.

Hint, in C read "!" as "not"

case 0: // waiting at start
case 1: // choose direction
case 2: // moving
case 3: //waiting for enable to go false
case 10: // start of homing
case 11: // waiting for index & pulse

> Also, after the homing sequence does the carousel.0.pocket-number need
> to be reset to 0, and reset after every completed toolchange?

No, you can leave it as it is, and if you set enable with it already
in the right position the compoinent will fall through the states
without setting any outputs other than "ready"

>  it is sitting on pocket 1, with the following conditions
> of all pins,

It is waiting for the "enable" pin to go false. The pin needs to go
from 0 to 1 to start a new move. If you connect it to
iocontrol.0.tool-change then that should happen automatically.


-- 
atp
If you can't fix it, you don't own it.
http://www.ifixit.com/Manifesto

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[Gene Heskett]

On Monday 02 November 2015 05:12:32 andy pugh wrote:

> On 2 November 2015 at 02:26, Gene Heskett  wrote:
> > Found it!  That is a quite simplified version that assumes some
> > unseen mechanism to run the drawbar, and a straight out the side of
> > the wheel exit strategy.
>
> Yes, that is how the demo is set up, but the actual sequence is set by
> a G-code subroutine, so is reasonably easy to change.
>
> All the carousel module does is interpret carousel position feeback
> and  issue motor movement commands for a tool carousel. Everything
> else is handled elsewhere. In the examples it is by G-code, but it
> would interface with Ladder just as easily.
>
> As you need to move the head up and down and issue spindle speed
> commands I think that you will find G-code more convenient, moving the
> head from ladder is more difficult.
>
> Also, that example includes spindle orient. You might want to consider
> a spindle orient commands rather than a spindle start command.

I've no clue how a velocity servo could do that neatly because of the 
inevitable variations in bearing friction.  This is the cause of some 
spindle heating in this mill, and I've played with the tension a wee 
bit, but haven't explored their replacement with higher quality 
bearings. I am well aware that this thing is NOT a Hardinge.

> The docs say that M19 is limited to R-values of 0-360 but you can
> force movement in one direction or the other. I think that a string of
> 180 degree moves could be used to unscrew the drawbar by a requested
> amount. It also has a built-in timeout in case things stick.

Precluding that is one of the reasons for the head gearshift.  In low 
gear, I don't think stickage would ever be a problem.  The torque is 
there, all I have to do is give it some "good gulf".  Referring to Gulf 
Oil's best refinery output.  "Tipping the can a little higher" etc etc.
Jon's pwm-servo amp can do 20 amps at 160 volts, this motor originally 
could do 2250 revs on the OEM traic (or scr) controller, which for some 
reason unk to me was restricted to about 1100 revs in reverse.  I have 
about 107 volts of DC available, at 11 amps CCS, but can do 40 till the 
wall breaker trips.  Jon's servo has a 16 amp current setting limit, so 
I can make it do things with authority.  It can do about 2850 revs on 
this controller, but only for 3 minutes or so before an internal thermal 
limit shuts it down for a 1 minute (approximately) cooldown.  However, 
leaving it at 2750 for half an hour produces only a 10F thermal rise 
looking at the motor.  In that same half hour, the bottom of the spindle 
will be close to 200F!  Piss poor bearings IMNSHO.

But, the size of a 10 pocket wheel precludes making it on this machine, 
not enough Y travel, unless I can A: Borrow a 6 to 8" rotary table, or 
B: figure out how to cut off the back 4 pockets, making a 6 pocket wheel 
(more than enough said stupid)  G-Code wise that is not impossible, and 
is likely the path I will follow here since the only rotary table I have 
is both a piece of junk I have motorized, and a 4" whose braking clamps 
are a brass thumbscrew and in-capable of holding it against the cutting 
forces involved here, it WILL move to its backlash limits.  That was 
nearly a show stopper when I made those sprockets for the bandsaw fence 
lashup a couple years ago with your code help. That backlash made the 
tooth too wide, and I had to file them about 3 thou wider before they 
would enter a piece of #25 chain.  Still a little sticky, but it works, 
which was the point of that project.  Not your fault.  If I were to 
remake them again, I'd program in backlash takeup moves at the peaks and 
gully's of the tooth.

I do need to get a bigger, more precise table and motorize it, but $500+ 
(plus another sky hook in the ceiling for a rope hoist to handle it 
with, I use a deer hoist, 3 sheeves wide, small nylon rope for that sort 
of thing) to get that going isn't $omething I want to $pend ATM.

It will happen eventually unless I fall over first though.

So a truncated wheel, carrying 5 tools seems like the best way fwd right 
now. Raw material may dictate that as I have either the whole deck panel 
from an Ampex VR1200, ( liberally drilled here and there for hole to 
mount stuff when it really was a VR-1200) or better yet, a few pieces of 
its 4" wide stiffening skirt thats not cut up yet.  If I put the axle 
hole on the edge & make a cap to clamp it to the rotating shaft with, 
the rest of it should be wide enough for a 5 pocket half wheel.

That, if I can whack away the code, I can get going yet today on the 
small mill as it can do about 5" of Y motion. That way I won't have to 
move the jig on the 704's table.

Needs more coffee though, I'm not quite up to speed for the day yet.

So attack the code next. I can see where arcbuddy.py is going to get a 
workout. :)

> The 
> advantage would be that you could ensure that the drabar thread
> remained engaged.
>
> My milling machine has a sturdy c

Re: [Emc-users] Carousel Component

[andy pugh]

On 2 November 2015 at 15:08, Gene Heskett  wrote:
>> Also, that example includes spindle orient. You might want to consider
>> a spindle orient commands rather than a spindle start command.
>
> I've no clue how a velocity servo could do that neatly because of the
> inevitable variations in bearing friction.

PID.


-- 
atp
If you can't fix it, you don't own it.
http://www.ifixit.com/Manifesto

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[Gene Heskett]

On Monday 02 November 2015 10:28:13 andy pugh wrote:

> On 2 November 2015 at 15:08, Gene Heskett  wrote:
> >> Also, that example includes spindle orient. You might want to
> >> consider a spindle orient commands rather than a spindle start
> >> command.
> >
> > I've no clue how a velocity servo could do that neatly because of
> > the inevitable variations in bearing friction.
>
> PID.

Already in there, setup for velocity. That, its quite stiff at doing.
[SPINDLE_9]
P = 2.0
I = 0.03
D = 0.00
FF0 = 1.00
FF1 = 0.020
FF2 = 0.00
BIAS = 0.00
FBIAS = 0.60
RBIAS = -0.60
ACCELL = 6.7
LIMIT_S_ACCEL = 750
DEADBAND = 0.00
ENCODER_SCALE = 268.0
PWMGEN_S_SCALE =46.00
MAXACCEL = 1500
PWMGEN_S_FREQ = 3
NEAR_TOL = 5
NEAR_SCALE = 1.1
ENCODER_SAMPLE = 50

Suggested changes to make position work?

The accuracy, at that encoder scale is no better than a couple degrees.  
It would have to interpolate, and hunt, or use a deadband to stop the 
hunting.

My thinking has been one of working around that limit since I do not 
intend to ever drop the 3/4" R8 as long as I have TTS adapters left, 
unless I have to disable the tool changer and do a special setup by 
hand.  Working around that limit also gives turning room so there is 
some inertial energy in the spindle when the drawbar wrench hits its 
stop pin. That should help to loosen or tighten the drawbar.

Playing with the code, using that 4" as a limit, I can put 6 pockets in 
it, and won't have to make the shaft clamp cap, there's room for the 
hole and a setscrew or 2 to fix it to the 1/2" shaft.

Got that worked out, and about to be started on the toy mill.

News at 11.

Thanks Andy.
Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page 

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[andy pugh]

On 2 November 2015 at 17:30, Gene Heskett  wrote:
> Suggested changes to make position work?

Pretty picture here:
http://wiki.linuxcnc.org/cgi-bin/wiki.pl?SpindleOrient

You might well be able to use spindle.hal from that vmc_toolchange
config directly (which is a HAL implementation of that logic) , except
that it looks under [SPINDLE] rather than [SPINDLE_9] I think.

-- 
atp
If you can't fix it, you don't own it.
http://www.ifixit.com/Manifesto

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[Gene Heskett]

On Monday 02 November 2015 13:13:34 andy pugh wrote:

> On 2 November 2015 at 17:30, Gene Heskett  wrote:
> > Suggested changes to make position work?
>
> Pretty picture here:
> http://wiki.linuxcnc.org/cgi-bin/wiki.pl?SpindleOrient
>
> You might well be able to use spindle.hal from that vmc_toolchange
> config directly (which is a HAL implementation of that logic)
In which case it IS editable to fix the below comment.  But would there 
be name clashes between the two setups?  Perhaps a separate [SPINDLE] 
grouping in the .ini might be on purpose...

> , except 
> that it looks under [SPINDLE] rather than [SPINDLE_9] I think.

Msg marked for when I am ready to tackle that.

Thank you very much, Andy.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page 

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[Rick Lair]

 > sudo apt-get install linuxcnc-dev
 >

I am getting an error trying this ^^

Here is the error,

greenmill@greenmill:~$ sudo apt-get install linuxcnc-dev
[sudo] password for greenmill:
Reading package lists... Done
Building dependency tree
Reading state information... Done
Package linuxcnc-dev is not available, but is referred to by another
package.
This may mean that the package is missing, has been obsoleted, or
is only available from another source

E: Package 'linuxcnc-dev' has no installation candidate
greenmill@greenmill:~$


Is this because of running uspace?



Rick





-- 

Thanks


Rick Lair
Superior Roll & Turning LLC
399 East Center Street
Petersburg MI, 49270
PH: 734-279-1831
FAX: 734-279-1166
www.superiorroll.com


--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[andy pugh]

On 3 November 2015 at 15:41, Rick Lair  wrote:
> E: Package 'linuxcnc-dev' has no installation candidate
> greenmill@greenmill:~$
>
>
> Is this because of running uspace?

I don't know why it is. Can you see it in the graphical package manager?

It seems that both Wheezy and Ubuntu use Synaptic, do you have the
LinuxCNC repository in your third-party sources?
https://help.ubuntu.com/community/Repositories/Ubuntu#Third-Party_Software_Tab


-- 
atp
If you can't fix it, you don't own it.
http://www.ifixit.com/Manifesto

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel Component

[andy pugh]

On 3 November 2015 at 16:27,   wrote:
> I don't see linuxcnc-dev in the list, and I am going to have to say
> after looking at the source list, thats probably why.

If you are using uspace, that all looks right.

I need to experiment on my own system to check a few things, and that
is at home.

-- 
atp
If you can't fix it, you don't own it.
http://www.ifixit.com/Manifesto

--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

[Emc-users] Carousel component questions

[Jan Bos]

Hello everybody,

I am working on an ATC for my mill and intend to use the carousel
component. My ATC has only one sensor that detects the pockets, it is a
Geneva style ATC and the sensor is on the drive wheel and senses each
rotation i.e. each step the wheel makes. There is no sensor to tell the
system which pocket is facing the spindle. If the wheel rotates the sensor
pulses for each pocket only.

Question now is how do I tell LinuxCNC which pocket is facing the spindle
and how to home this thing?

Do I need to add a 'home' sensor or a 'pocket 1' sensor?

I am not quite clear on how this is supposed to work, could anybody shine
some light on how the Carousel component works with sensor inputs like this?

Thanks,

JB
--
Developer Access Program for Intel Xeon Phi Processors
Access to Intel Xeon Phi processor-based developer platforms.
With one year of Intel Parallel Studio XE.
Training and support from Colfax.
Order your platform today. http://sdm.link/xeonphi
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

[Emc-users] carousel component. calling andy

[andrew beck]

hey guys just have a question about the carousel component

I have a 80 tool carousel on my mazak horizontal 5axis

and i need to control it.

i know the carousel component can be used with a stepper motor and it just
needs one home switch and then calculates step and direction signals for
any movement from there

I have a couple of options here to make this work
i could pull the current motor out which is a yaskawa inverter rated motor
with a 5v differential encoder on the end(ready to plug into a mesa card).
And then just replace it with a servo motor and use step and direction to
control it and away we go.  the downside of this is there is a spline on
the original motor i don't really want to have to touch..  and the cost of
a servo and driver  approx 350usd

the 2nd option is a expensive vfd that has 0 +-10v i am hoping that i can
control that as a servo and the carousel component tells it what to rotate
to for each tool.  cost 300usd

the 3rd option and the one i am hoping i can use is i buy a cheap vfd that
has a encoder card for closed loop control (approx80usd)  and just plug it
into motor and wire encoder in to vfd and then back to mesa card.  the
problem is the cheap vfds only have 0-10v not +-10v

so i would have to have a direction pin plus 0-10v.  Andy would that be
able to work ok?  it means that i can't run a PID in linuxcnc on it i
think.  but it might work ok still.  I know that i can get full servo-like
control down to about 10rpm still and it works great.  i use them on my
other cnc mills for the spindle and they are great and provide pretty good
spindle orientation definitely good enough to do toolchanges with no issue
in about 1 toolchanges so far.

anyway let me know what you guys think

just looking at options currently though want to chose some sort of option
in the next week as we need this cnc mill going.

i have jobs for it to do soon:)

___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component questions

[Chris Radek]

On Fri, Nov 04, 2016 at 09:23:00AM +0800, Jan Bos wrote:
> 
> Do I need to add a 'home' sensor or a 'pocket 1' sensor?

Yes!  You will never regret fixing hardware so it's not stupid.

I did this on my vmc's tool changer - I moved a prox to give a home
and quadrature signal instead of the stupid original configuration
that would let the original control load the wrong tool if the
contactor stuck for a smidgen of a second - it has worked perfectly
since.

When you are retrofitting do not be ashamed to make the machine work
better than it did originally :-)


--
Developer Access Program for Intel Xeon Phi Processors
Access to Intel Xeon Phi processor-based developer platforms.
With one year of Intel Parallel Studio XE.
Training and support from Colfax.
Order your platform today. http://sdm.link/xeonphi
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component questions

[andy pugh]

On 4 November 2016 at 01:23, Jan Bos  wrote:
> My ATC has only one sensor that detects the pockets, it is a
> Geneva style ATC and the sensor is on the drive wheel and senses each
> rotation i.e. each step the wheel makes. There is no sensor to tell the
> system which pocket is facing the spindle.

How does it work then? Does it rely on the operator jogging to tool 1
at the start of the day?

If you add an extra pocket-1 sensor then the carousel component ought
to work. Without that I don't see how anything could ever work.

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1916

--
Developer Access Program for Intel Xeon Phi Processors
Access to Intel Xeon Phi processor-based developer platforms.
With one year of Intel Parallel Studio XE.
Training and support from Colfax.
Order your platform today. http://sdm.link/xeonphi
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component questions

[Kirk Wallace]

On 11/03/2016 06:23 PM, Jan Bos wrote:
...
> Do I need to add a 'home' sensor or a 'pocket 1' sensor?
>
> I am not quite clear on how this is supposed to work, could anybody shine
> some light on how the Carousel component works with sensor inputs like this?
...

For reference, I don't use the carousel component but this is how my 
mill was set up:
http://www.wallacecompany.com/tmp/Carousel_Position.png

The chain has a foot mounted at the home position to activate the home 
switch. The pocket switch sensed when a pocket was in parked position. I 
seem to recall that the g-code called for X number of pockets forward or 
back, and did not keep track of pocket ID.

(The Geneva has been replaced with direct drive and there is a position 
encoder on the carousel.)

-- 
Kirk Wallace
http://www.wallacecompany.com/machine_shop/
http://www.wallacecompany.com/E45/

--
Developer Access Program for Intel Xeon Phi Processors
Access to Intel Xeon Phi processor-based developer platforms.
With one year of Intel Parallel Studio XE.
Training and support from Colfax.
Order your platform today. http://sdm.link/xeonphi
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component questions

[Jan Bos]

Thank you all for your feedback. As I expected, I will add the sensor and
give it a try.

I am following the 'VMV Hal related questions' thread on the forum and I
have the virtual machine working. This virtual set-up is very close to what
I want to achieve. I am hooking up the sensors now for the ATC and the
relays for the motor and then I will try to make stuff move. I do not need
spindle orientation since I have this in the VFD.

I am confident that I will be back for further help but for now, I have
work to do.

Thanks again!

JB

On Fri, Nov 4, 2016 at 11:50 PM, Kirk Wallace 
wrote:

> On 11/03/2016 06:23 PM, Jan Bos wrote:
> ...
> > Do I need to add a 'home' sensor or a 'pocket 1' sensor?
> >
> > I am not quite clear on how this is supposed to work, could anybody shine
> > some light on how the Carousel component works with sensor inputs like
> this?
> ...
>
> For reference, I don't use the carousel component but this is how my
> mill was set up:
> http://www.wallacecompany.com/tmp/Carousel_Position.png
>
> The chain has a foot mounted at the home position to activate the home
> switch. The pocket switch sensed when a pocket was in parked position. I
> seem to recall that the g-code called for X number of pockets forward or
> back, and did not keep track of pocket ID.
>
> (The Geneva has been replaced with direct drive and there is a position
> encoder on the carousel.)
>
> --
> Kirk Wallace
> http://www.wallacecompany.com/machine_shop/
> http://www.wallacecompany.com/E45/
>
> 
> --
> Developer Access Program for Intel Xeon Phi Processors
> Access to Intel Xeon Phi processor-based developer platforms.
> With one year of Intel Parallel Studio XE.
> Training and support from Colfax.
> Order your platform today. http://sdm.link/xeonphi
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>
--
Developer Access Program for Intel Xeon Phi Processors
Access to Intel Xeon Phi processor-based developer platforms.
With one year of Intel Parallel Studio XE.
Training and support from Colfax.
Order your platform today. http://sdm.link/xeonphi
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component questions

[Marius Alksnys]

Hello, Jan,

this is my carousel HAL component version. I made it more reliable, was 
tested a lot and is working on a real machine. Sense-pocket is low when 
carousel is fixed and high between positions. Sense-index occurs between 
positions AFAIR.
Might require changes because of hard-coded sensor states it expects. I 
used it with M6 re-mapping, thus used analog-out for socket number.
Here is the code:

component mycarousel "Orient a tool drum custom component";
// partly derived from andy-pugh carousel HAL component

pin in  float pocket-number-f "The pocket to move to when the .enable 
pin goes high. Float to be compatible with motion.analog-out";
pin in  bitenable "Set this pin high to start movement. Setting it 
low will stop movement";
pin in  bitunhome "Unhomes the component and forces to find index 
again";
pin out bitactive "indicates that the component is active";
pin out bitready "This pin goes high when the drum is in-position";
pin in  bitsense-index "Drum index position feedback pin";
pin in  bitsense-pocket = true "Drum pocket position feedback pin";
pin out bitmotor-fwd "Indicates the motor should run forwards 
(bigger numbers)";
pin out bitmotor-rev "Indicates the motor should run reverse";
pin out bitis-ok = true "No fault output";
pin out signed pos = 0 "This pin indicates the current position feedback";
pin out bithomed = false "Shows that homing is complete.";

param r  signed state = 0 "Current component state";
param rw signed home-pocket = 19 "Pocket number at index.";
param rw signed pockets = 21 "Pocket count in drum";
param rw signed timeout-one = 60 "Maximum time to wait for next pocket 
in thread periods";
param rw signed timeout-all = 1600 "Maximum time to rotate to any pocket 
in thread periods";

license "GPL";
author "andy pugh, Marius Alksnys";

variable hal_bit_t old_sense = false;
variable hal_bit_t old_index = false;
variable hal_s32_t timeone = 3;
variable hal_s32_t timeall = 3;
variable hal_s32_t pocket_number = 0;

function _ fp;

option singleton yes;

;;

#include 

#define ST_WAITING 0
#define ST_HOMING_INDEX 10
#define ST_FINISH_HOMING 20
#define ST_CHOOSE_DIR 30
#define ST_MOVING 40
#define ST_STOP_AT_POCKET 50
#define ST_DONE 60
#define ST_HALT 96
#define ST_HALTED 99

#define MSG_SOURCE "Tool carousel: "

FUNCTION(_) {
 if (unhome) homed = false;
 if (sense_pocket && !old_sense) { // Pocket sensor signal just rised!
 timeone = timeout_one;
 if (homed) {
 if (motor_fwd) {
 pos += 1;
 if (pos > pockets) pos -= pockets;
 }
 if (motor_rev) {
 pos -= 1;
 if (pos < 1) pos += pockets;
 }
 if (!motor_fwd && !motor_rev) {
 homed = false;
 state = ST_WAITING;
 rtapi_print_msg(RTAPI_MSG_ERR, MSG_SOURCE "got socket 
sensor signal, while no command to turn. Will re-home.");
 }
 }
 }
 old_sense = sense_pocket;

 if (sense_index && !old_index) { // Index sensor signal just rised!
 if (homed) {
 if ((motor_fwd && (pos != home_pocket)) || (motor_rev && 
(pos != home_pocket + 1))) {
 state = ST_HALT;
 rtapi_print_msg(RTAPI_MSG_ERR, MSG_SOURCE "home sensor 
signal at wrong time (very big risk to brake equipment), component will 
halt!!!");
 }
 }
 }
 old_index = sense_index;

 if (is_ok) {
 if (sense_pocket && sense_index) {
 state = ST_HALT;
 rtapi_print_msg(RTAPI_MSG_ERR, MSG_SOURCE "both sensors 
active (very big risk to brake equipment), component will halt!!!");
 }
 if (active) { // Checking for timeouts
 if (timeone-- < 0) {
 motor_fwd = false;
 motor_rev = false;
 active = false;
 is_ok = false;
 state = ST_DONE;
 rtapi_print_msg(RTAPI_MSG_ERR, MSG_SOURCE "time-out 
waiting for one position!");
 }
 if (timeall-- < 0) {
 is_ok = false;
 state = ST_STOP_AT_POCKET;
 rtapi_print_msg(RTAPI_MSG_ERR, MSG_SOURCE "requested 
position time-out!");
 }
 }
 }

 switch (state) {
 case ST_WAITING: // waiting at start
 if (! enable) return;
 timeall = timeout_all;
 if (! sense_pocket) {
 homed = false;
 rtapi_print_msg(RTAPI_MSG_ERR, MSG_SOURCE "carousel was not 
fixed before rotating. Will re-home.");
 }
 active = true;
 pocket_number = round(pocket_number_f);
 if (!homed) {
 motor_fwd = false;
 motor_rev = true;
 timeone = timeout_one;
 state = ST_HOMING_INDEX;
 break;
 }
 state = ST_CHOOSE_DIR;
 ready = false;
 case ST

Re: [Emc-users] Carousel component questions

[Marius Alksnys]

11/07/2016 09:37 AM, Marius Alksnys rašė:

Hello, Jan,

this is my carousel HAL component version. I made it more reliable, was
tested a lot and is working on a real machine. Sense-pocket is low when
carousel is fixed and high between positions. Sense-index occurs once per 
revolution between
positions AFAIR.
Might require changes because of hard-coded sensor states it expects. I
used it with M6 re-mapping, thus used analog-out for socket number.


The component source file is attached.
component mycarousel "Orient a tool drum custom component";
// partly derived from andy-pugh carousel HAL component

pin in  float pocket-number-f "The pocket to move to when the .enable pin goes 
high. Float to be compatible with motion.analog-out";
pin in  bitenable "Set this pin high to start movement. Setting it low will 
stop movement";
pin in  bitunhome "Unhomes the component and forces to find index again";
pin out bitactive "indicates that the component is active";
pin out bitready "This pin goes high when the drum is in-position";
pin in  bitsense-index "Drum index position feedback pin";
pin in  bitsense-pocket = true "Drum pocket position feedback pin";
pin out bitmotor-fwd "Indicates the motor should run forwards (bigger 
numbers)";
pin out bitmotor-rev "Indicates the motor should run reverse";
pin out bitis-ok = true "No fault output";
pin out signed pos = 0 "This pin indicates the current position feedback";
pin out bithomed = false "Shows that homing is complete.";

param r  signed state = 0 "Current component state";
param rw signed home-pocket = 19 "Pocket number at index.";
param rw signed pockets = 21 "Pocket count in drum";
param rw signed timeout-one = 60 "Maximum time to wait for next pocket in 
thread periods";
param rw signed timeout-all = 1600 "Maximum time to rotate to any pocket in 
thread periods";

license "GPL";
author "andy pugh, Marius Alksnys";

variable hal_bit_t old_sense = false;
variable hal_bit_t old_index = false;
variable hal_s32_t timeone = 3;
variable hal_s32_t timeall = 3;
variable hal_s32_t pocket_number = 0;

function _ fp;

option singleton yes;

;;

#include 

#define ST_WAITING 0
#define ST_HOMING_INDEX 10
#define ST_FINISH_HOMING 20
#define ST_CHOOSE_DIR 30
#define ST_MOVING 40
#define ST_STOP_AT_POCKET 50
#define ST_DONE 60
#define ST_HALT 96
#define ST_HALTED 99

#define MSG_SOURCE "Tool carousel: "

FUNCTION(_) {
if (unhome) homed = false;
if (sense_pocket && !old_sense) { // Pocket sensor signal just rised!
timeone = timeout_one;
if (homed) {
if (motor_fwd) {
pos += 1;
if (pos > pockets) pos -= pockets;
}
if (motor_rev) {
pos -= 1;
if (pos < 1) pos += pockets;
}
if (!motor_fwd && !motor_rev) {
homed = false;
state = ST_WAITING;
rtapi_print_msg(RTAPI_MSG_ERR, MSG_SOURCE "got socket sensor 
signal, while no command to turn. Will re-home.");
}
}
}
old_sense = sense_pocket;

if (sense_index && !old_index) { // Index sensor signal just rised!
if (homed) {
if ((motor_fwd && (pos != home_pocket)) || (motor_rev && (pos != 
home_pocket + 1))) {
state = ST_HALT;
rtapi_print_msg(RTAPI_MSG_ERR, MSG_SOURCE "home sensor signal 
at wrong time (very big risk to brake equipment), component will halt!!!");
}
}
}
old_index = sense_index;

if (is_ok) {
if (sense_pocket && sense_index) {
state = ST_HALT;
rtapi_print_msg(RTAPI_MSG_ERR, MSG_SOURCE "both sensors active 
(very big risk to brake equipment), component will halt!!!");
}
if (active) { // Checking for timeouts
if (timeone-- < 0) {
motor_fwd = false;
motor_rev = false;
active = false;
is_ok = false;
state = ST_DONE;
rtapi_print_msg(RTAPI_MSG_ERR, MSG_SOURCE "time-out waiting for 
one position!");
}
if (timeall-- < 0) {
is_ok = false;
state = ST_STOP_AT_POCKET;
rtapi_print_msg(RTAPI_MSG_ERR, MSG_SOURCE "requested position 
time-out!");
}
}
}

switch (state) {
case ST_WAITING: // waiting at start
if (! enable) return;
timeall = timeout_all;
if (! sense_pocket) {
homed = false;
rtapi_print_msg(RTAPI_MSG_ERR, MSG_SOURCE "carousel was not fixed 
before rotating. Will re-home.");
}
active = true;
pocket_number = round(pocket_number_f);
if (!homed) {
motor_fwd = false;
motor_rev = true;
timeone = timeout_one;
state = ST_HOMING_INDEX;
break;
}
state = ST_CHOOSE_DIR;
r

Re: [Emc-users] Carousel component questions

[Jan Bos]

Hello Marius,

Thanks a lot for your detailed explanation and your comp file. I printed it
and will go through it to see what I need to do to adapt it.

I see you made quite some changes to it compared to the original carousel
component, am I missing the shortcomings of the original carousel component
or is yours just 'different'? What I do see are added state verification
routines and state error routines.

Thanks again for your support, much appreciated!

JB

On Mon, Nov 7, 2016 at 3:51 PM, Marius Alksnys  wrote:

>
>
> 11/07/2016 09:37 AM, Marius Alksnys rašė:
>
>> Hello, Jan,
>>
>> this is my carousel HAL component version. I made it more reliable, was
>> tested a lot and is working on a real machine. Sense-pocket is low when
>> carousel is fixed and high between positions. Sense-index occurs once per
>> revolution between
>> positions AFAIR.
>> Might require changes because of hard-coded sensor states it expects. I
>> used it with M6 re-mapping, thus used analog-out for socket number.
>>
>
> The component source file is attached.
>
> 
> --
> Developer Access Program for Intel Xeon Phi Processors
> Access to Intel Xeon Phi processor-based developer platforms.
> With one year of Intel Parallel Studio XE.
> Training and support from Colfax.
> Order your platform today. http://sdm.link/xeonphi
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>
>
--
Developer Access Program for Intel Xeon Phi Processors
Access to Intel Xeon Phi processor-based developer platforms.
With one year of Intel Parallel Studio XE.
Training and support from Colfax.
Order your platform today. http://sdm.link/xeonphi
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

[Emc-users] Carousel component indexing/trigger problem

[Andreas Linnarsson]

Hi!

I've almost completed my Emco PC Turn 50 conversion but my headache now is
the tool turret. Using the carousel component I've almost made it work.

The problem I'm facing is that the turret rotates forward until it's
triggered by the strobe/pulse (pocket sensor), before rotating back into
locking position. Since the reversal de-triggers the pocket sensor that
effectively means that the "next" pocket sensor trigger is the same as the
one before. After having performed the homing sequence the turret thus just
moves back and forth within the same pocket.

Is it possible to have the carousel component ignore the first pocket
sensor trigger in order to proceed to the next slot or, can i be run in
some kind of timer mode so this doesn't happen? Any inputs would be greatly
appreciated!

Kind regards,

Andreas

___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

[Emc-users] carousel component question for andy

[andrew beck]

Hey Andy I see your component supports step direction now.

2 questions.

Does it still do shortest path calculations to speed up tool changes?

And 2nd question does it happen to support analog servo drives with a built
in pid in linuxcnc?

Regards


Andrew

___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] carousel component. calling andy

[Stuart Stevenson]

There is not currently a carousel motor controller on the machine?



On Fri, Jul 26, 2024 at 6:07 AM andrew beck 
wrote:

> hey guys just have a question about the carousel component
>
> I have a 80 tool carousel on my mazak horizontal 5axis
>
> and i need to control it.
>
> i know the carousel component can be used with a stepper motor and it just
> needs one home switch and then calculates step and direction signals for
> any movement from there
>
> I have a couple of options here to make this work
> i could pull the current motor out which is a yaskawa inverter rated motor
> with a 5v differential encoder on the end(ready to plug into a mesa card).
> And then just replace it with a servo motor and use step and direction to
> control it and away we go.  the downside of this is there is a spline on
> the original motor i don't really want to have to touch..  and the cost of
> a servo and driver  approx 350usd
>
> the 2nd option is a expensive vfd that has 0 +-10v i am hoping that i can
> control that as a servo and the carousel component tells it what to rotate
> to for each tool.  cost 300usd
>
> the 3rd option and the one i am hoping i can use is i buy a cheap vfd that
> has a encoder card for closed loop control (approx80usd)  and just plug it
> into motor and wire encoder in to vfd and then back to mesa card.  the
> problem is the cheap vfds only have 0-10v not +-10v
>
> so i would have to have a direction pin plus 0-10v.  Andy would that be
> able to work ok?  it means that i can't run a PID in linuxcnc on it i
> think.  but it might work ok still.  I know that i can get full servo-like
> control down to about 10rpm still and it works great.  i use them on my
> other cnc mills for the spindle and they are great and provide pretty good
> spindle orientation definitely good enough to do toolchanges with no issue
> in about 1 toolchanges so far.
>
> anyway let me know what you guys think
>
> just looking at options currently though want to chose some sort of option
> in the next week as we need this cnc mill going.
>
> i have jobs for it to do soon:)
>
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>


-- 
Addressee is the intended audience.
If you are not the addressee then my consent is not given for you to read
this email furthermore it is my wish you would close this without saving or
reading, and cease and desist from saving or opening my private
correspondence.
Thank you for honoring my wish.

___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] carousel component. calling andy

[andrew beck]

hey stuart nice to hear from you

this machine might be like some of the big ones you did

yes good question

its a yaskawa inverter motor internal positioner

yes i could control it.  but they are worth lots of money on ebay like 3k.
and i would prefer to handle it all inside linuxcnc than have a magic black
box i don't understand.  then i can interlink classic ladder and estops and
everything etc





On Fri, Jul 26, 2024 at 11:25 PM Stuart Stevenson  wrote:

> There is not currently a carousel motor controller on the machine?
>
>
>
> On Fri, Jul 26, 2024 at 6:07 AM andrew beck 
> wrote:
>
> > hey guys just have a question about the carousel component
> >
> > I have a 80 tool carousel on my mazak horizontal 5axis
> >
> > and i need to control it.
> >
> > i know the carousel component can be used with a stepper motor and it
> just
> > needs one home switch and then calculates step and direction signals for
> > any movement from there
> >
> > I have a couple of options here to make this work
> > i could pull the current motor out which is a yaskawa inverter rated
> motor
> > with a 5v differential encoder on the end(ready to plug into a mesa
> card).
> > And then just replace it with a servo motor and use step and direction to
> > control it and away we go.  the downside of this is there is a spline on
> > the original motor i don't really want to have to touch..  and the cost
> of
> > a servo and driver  approx 350usd
> >
> > the 2nd option is a expensive vfd that has 0 +-10v i am hoping that i can
> > control that as a servo and the carousel component tells it what to
> rotate
> > to for each tool.  cost 300usd
> >
> > the 3rd option and the one i am hoping i can use is i buy a cheap vfd
> that
> > has a encoder card for closed loop control (approx80usd)  and just plug
> it
> > into motor and wire encoder in to vfd and then back to mesa card.  the
> > problem is the cheap vfds only have 0-10v not +-10v
> >
> > so i would have to have a direction pin plus 0-10v.  Andy would that be
> > able to work ok?  it means that i can't run a PID in linuxcnc on it i
> > think.  but it might work ok still.  I know that i can get full
> servo-like
> > control down to about 10rpm still and it works great.  i use them on my
> > other cnc mills for the spindle and they are great and provide pretty
> good
> > spindle orientation definitely good enough to do toolchanges with no
> issue
> > in about 1 toolchanges so far.
> >
> > anyway let me know what you guys think
> >
> > just looking at options currently though want to chose some sort of
> option
> > in the next week as we need this cnc mill going.
> >
> > i have jobs for it to do soon:)
> >
> > ___
> > Emc-users mailing list
> > Emc-users@lists.sourceforge.net
> > https://lists.sourceforge.net/lists/listinfo/emc-users
> >
>
>
> --
> Addressee is the intended audience.
> If you are not the addressee then my consent is not given for you to read
> this email furthermore it is my wish you would close this without saving or
> reading, and cease and desist from saving or opening my private
> correspondence.
> Thank you for honoring my wish.
>
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>

___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] carousel component. calling andy

[Stuart Stevenson]

What is the mazak model so I can visualize what you are working on?

3k is not a lot of money if the needed machining is delayed.

It seems to me all your proposals will work.

Option 3 is just a repeat of what you already know.
Would you not get 0v-10v by swapping the leads/

I like your preference of 'doing it in LinuxCNC' rather than buying a
solution just to simplify.
Doing it in LinuxCNC enhances LinuxCNC for a lot of people and projects.
I realize the main goal is 'making money'. This allows for some of the
'preferences' to be satisfied.

have fun
Stuart

My 'shop' is growing.
I now have:
A 5 speed bench drill press from Harbor Freight.
A Makerbot Replicator Mini
and recently acquired Ender-3 S1 Plus - I had to order a motherboard
yesterday (25 Jul 2024) rather than a Raspberry Pi because a project
requires the machine running ASAP. The motherboard from Creality is
supposed to be a quiet stepper controller. There is no feedback from the
stepper so I have my doubts as to the quietness.


On Fri, Jul 26, 2024 at 6:34 AM andrew beck 
wrote:

> hey stuart nice to hear from you
>
> this machine might be like some of the big ones you did
>
> yes good question
>
> its a yaskawa inverter motor internal positioner
>
> yes i could control it.  but they are worth lots of money on ebay like 3k.
> and i would prefer to handle it all inside linuxcnc than have a magic black
> box i don't understand.  then i can interlink classic ladder and estops and
> everything etc
>
>
>
>
>
> On Fri, Jul 26, 2024 at 11:25 PM Stuart Stevenson 
> wrote:
>
> > There is not currently a carousel motor controller on the machine?
> >
> >
> >
> > On Fri, Jul 26, 2024 at 6:07 AM andrew beck 
> > wrote:
> >
> > > hey guys just have a question about the carousel component
> > >
> > > I have a 80 tool carousel on my mazak horizontal 5axis
> > >
> > > and i need to control it.
> > >
> > > i know the carousel component can be used with a stepper motor and it
> > just
> > > needs one home switch and then calculates step and direction signals
> for
> > > any movement from there
> > >
> > > I have a couple of options here to make this work
> > > i could pull the current motor out which is a yaskawa inverter rated
> > motor
> > > with a 5v differential encoder on the end(ready to plug into a mesa
> > card).
> > > And then just replace it with a servo motor and use step and direction
> to
> > > control it and away we go.  the downside of this is there is a spline
> on
> > > the original motor i don't really want to have to touch..  and the cost
> > of
> > > a servo and driver  approx 350usd
> > >
> > > the 2nd option is a expensive vfd that has 0 +-10v i am hoping that i
> can
> > > control that as a servo and the carousel component tells it what to
> > rotate
> > > to for each tool.  cost 300usd
> > >
> > > the 3rd option and the one i am hoping i can use is i buy a cheap vfd
> > that
> > > has a encoder card for closed loop control (approx80usd)  and just plug
> > it
> > > into motor and wire encoder in to vfd and then back to mesa card.  the
> > > problem is the cheap vfds only have 0-10v not +-10v
> > >
> > > so i would have to have a direction pin plus 0-10v.  Andy would that be
> > > able to work ok?  it means that i can't run a PID in linuxcnc on it i
> > > think.  but it might work ok still.  I know that i can get full
> > servo-like
> > > control down to about 10rpm still and it works great.  i use them on my
> > > other cnc mills for the spindle and they are great and provide pretty
> > good
> > > spindle orientation definitely good enough to do toolchanges with no
> > issue
> > > in about 1 toolchanges so far.
> > >
> > > anyway let me know what you guys think
> > >
> > > just looking at options currently though want to chose some sort of
> > option
> > > in the next week as we need this cnc mill going.
> > >
> > > i have jobs for it to do soon:)
> > >
> > > ___
> > > Emc-users mailing list
> > > Emc-users@lists.sourceforge.net
> > > https://lists.sourceforge.net/lists/listinfo/emc-users
> > >
> >
> >
> > --
> > Addressee is the intended audience.
> > If you are not the addressee then my consent is not given for you to read
> > this email furthermore it is my wish you would close this without saving
> or
> > reading, and cease and desist from saving or opening my private
> > correspondence.
> > Thank you for honoring my wish.
> >
> > ___
> > Emc-users mailing list
> > Emc-users@lists.sourceforge.net
> > https://lists.sourceforge.net/lists/listinfo/emc-users
> >
>
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>


-- 
Addressee is the intended audience.
If you are not the addressee then my consent is not given for you to read
this email furthermore it is my wish you would close this without saving or
reading, and cease and desist 

Re: [Emc-users] carousel component. calling andy

[andy pugh]

On Fri, 26 Jul 2024 at 12:07, andrew beck  wrote:

> i know the carousel component can be used with a stepper motor and it just
> needs one home switch and then calculates step and direction signals for
> any movement from there

Yes, but it can also work with many other forms of  feedback, almost
certainly including what the toolchanger currently offers.

And it also supports just turning a fixed-speed motor on and off.


-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1912


___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] carousel component. calling andy

[Dmitry Yurtaev]

On Fri, Jul 26, 2024 at 2:34 PM andrew beck 
wrote:


> its a yaskawa inverter motor internal positioner
>
> yes i could control it.  but they are worth lots of money on ebay like 3k.
> and i would prefer to handle it all inside linuxcnc than have a magic black
> box i don't understand.  then i can interlink classic ladder and estops and
> everything etc
>

what is the inverter? yaskawa justpoint? i used a CIMR-xxJP3 in a mazak
vtc20 linuxcnc retrofit lately - pretty straight-forward...
i doubt it will sell fast for 3k ;)

/dmitry

___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] carousel component. calling andy

[andrew beck]

Existing inverters are

Same one on both magazine and pallet changer

Yaskawa drive is Cimr-22jp 3B00M


Dimitry I'll message you on telegram.


Andy reason I ask about carousel component is I want a fast toolchanger
that accelerates up to speed and knows when it is getting close to correct
pocket from the encoder.

I have used the carousel component on another machine and it kept over
shooting as I needed to know before I hit the pocket sensors that it was
time to stop.  As it took time to stop.


I think when your component is in step direction mode that it does that.


Open to all ideas though.


Not overly fond of reusing old drives as they have already been repaired
once..  but I have spares from 2nd machine I guess so that's not a issue
really.  I have 2 spares to use...

On Sat, 27 Jul 2024, 00:51 Dmitry Yurtaev,  wrote:

> On Fri, Jul 26, 2024 at 2:34 PM andrew beck 
> wrote:
>
>
> > its a yaskawa inverter motor internal positioner
> >
> > yes i could control it.  but they are worth lots of money on ebay like
> 3k.
> > and i would prefer to handle it all inside linuxcnc than have a magic
> black
> > box i don't understand.  then i can interlink classic ladder and estops
> and
> > everything etc
> >
>
> what is the inverter? yaskawa justpoint? i used a CIMR-xxJP3 in a mazak
> vtc20 linuxcnc retrofit lately - pretty straight-forward...
> i doubt it will sell fast for 3k ;)
>
> /dmitry
>
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>

___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] carousel component. calling andy

[andy pugh]

On Fri, 26 Jul 2024 at 21:05, andrew beck  wrote:

> I have used the carousel component on another machine and it kept over
> shooting as I needed to know before I hit the pocket sensors that it was
> time to stop.  As it took time to stop.

The latest version has the capability of stopping, overshooting, and
then backing up to the index. (a bit like the classic LinuxCNC axis
homing sequence)

If you halcompile the version in master you could try that.

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1912


___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[andy pugh]

On Fri, 8 Jan 2021 at 07:26, Andreas Linnarsson
 wrote:

> Is it possible to have the carousel component ignore the first pocket
> sensor trigger in order to proceed to the next slot or, can i be run in
> some kind of timer mode so this doesn't happen? Any inputs would be greatly
> appreciated!

Can you tell us your loadrt string for carousel? So that I can work
out how the component is behaving in your case?

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1912


___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[Andreas Linnarsson]

Hi,

I'm using "loadrt carousel pockets=6 encoding=index dir=1" and have mapped
the home/index signal to "carousel.0.sense-0" and pocket/strobe to
"carousel.0.sense-1".

During homing the turret behaves as expected since it ignores the strobe -
I suppose. After that it's stuck within the same pocket due to the
reversing motion, thus backing off the pocket sensor.

The electrical behaviour of the stock control system can be seen in the
attached picture. When examining the STROBE (pocket) signal I can
understand why I'm experiencing what I do now. The TOOL CMD is the 24V
signal to the DC motor turning the turret and SYNC is the index/home signal.

Regards,

Andreas

[image: Turn_50_Turrret.PNG]



Den fre 8 jan. 2021 kl 13:07 skrev andy pugh :

> On Fri, 8 Jan 2021 at 07:26, Andreas Linnarsson
>  wrote:
>
> > Is it possible to have the carousel component ignore the first pocket
> > sensor trigger in order to proceed to the next slot or, can i be run in
> > some kind of timer mode so this doesn't happen? Any inputs would be
> greatly
> > appreciated!
>
> Can you tell us your loadrt string for carousel? So that I can work
> out how the component is behaving in your case?
>
> --
> atp
> "A motorcycle is a bicycle with a pandemonium attachment and is
> designed for the especial use of mechanical geniuses, daredevils and
> lunatics."
> — George Fitch, Atlanta Constitution Newspaper, 1912
>
>
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>


toolchange_index.hal
Description: Binary data
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[andy pugh]

On Fri, 8 Jan 2021 at 12:31, Andreas Linnarsson
 wrote:

> The electrical behaviour of the stock control system can be seen in the
> attached picture. When examining the STROBE (pocket) signal I can
> understand why I'm experiencing what I do now. The TOOL CMD is the 24V
> signal to the DC motor turning the turret and SYNC is the index/home signal.

I think that a quick-fix in your case might be to insert:

current_position -= 1

After line 317 of carousel.comp
https://github.com/LinuxCNC/linuxcnc/blob/master/src/hal/components/carousel.comp#L311

It would need to be a bit cleverer to be a _general_ solution.

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1912


___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[Andreas Linnarsson]

Even with the addition of that line the behaviour is the same. But since
I'm a novice when it comes to LinuxCNC I just want to make sure I
implemented the change correctly:

case 3:
timer -= fperiod;
if (timer > 0) break;
state = 4;
motor_fwd = 0;
motor_rev = 0;
motor_vel = hold_dc;
current_position -= 1;   //Line added
active = 0;
if (enable) ready = 1;

Then:
sudo halcompile --install carousel.comp

Did I miss(do) anything?



Den fre 8 jan. 2021 kl 13:44 skrev andy pugh :

> On Fri, 8 Jan 2021 at 12:31, Andreas Linnarsson
>  wrote:
>
> > The electrical behaviour of the stock control system can be seen in the
> > attached picture. When examining the STROBE (pocket) signal I can
> > understand why I'm experiencing what I do now. The TOOL CMD is the 24V
> > signal to the DC motor turning the turret and SYNC is the index/home
> signal.
>
> I think that a quick-fix in your case might be to insert:
>
> current_position -= 1
>
> After line 317 of carousel.comp
>
> https://github.com/LinuxCNC/linuxcnc/blob/master/src/hal/components/carousel.comp#L311
>
> It would need to be a bit cleverer to be a _general_ solution.
>
> --
> atp
> "A motorcycle is a bicycle with a pandemonium attachment and is
> designed for the especial use of mechanical geniuses, daredevils and
> lunatics."
> — George Fitch, Atlanta Constitution Newspaper, 1912
>
>
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>

___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[andy pugh]

On Fri, 8 Jan 2021 at 13:27, Andreas Linnarsson
 wrote:

> Did I miss(do) anything?

I don't think so.

Can you look at the current-position HAL pin to see if it decreases by
1 at the end of the reverse pulse?

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1912


___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[Andreas Linnarsson]

It will increase by "1" after every (unsuccessful) completion of a fwd jog.
After position 6 it will wrap around to 1, even though the index position
isn't sensed. I understand if index isn't used after homing though.

Do you know anyone who has used the carousel comp with this kind tool
turret? Or am I the first? If they've made it work I'm quite curious about
what their encoder wheel looked like or how they did it. I've 3D-printed a
variety of designs in search for a pattern that the carousel comp would be
compatible with but I cannot really get my head around if another design
would be the solution. I haven't found anything that works so far and
logically I can't see how it would work to stop on the first pocket
trigger. Not since that will always be the same as the last one due to the
reverse locking.

Regards,

Andreas

Den fre 8 jan. 2021 kl 15:02 skrev andy pugh :

> On Fri, 8 Jan 2021 at 13:27, Andreas Linnarsson
>  wrote:
>
> > Did I miss(do) anything?
>
> I don't think so.
>
> Can you look at the current-position HAL pin to see if it decreases by
> 1 at the end of the reverse pulse?
>
> --
> atp
> "A motorcycle is a bicycle with a pandemonium attachment and is
> designed for the especial use of mechanical geniuses, daredevils and
> lunatics."
> — George Fitch, Atlanta Constitution Newspaper, 1912
>
>
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>

___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[andy pugh]

On Fri, 8 Jan 2021 at 16:08, Andreas Linnarsson
 wrote:

> Do you know anyone who has used the carousel comp with this kind tool
> turret? Or am I the first?

You may be the first with an Emco. The addition of the reverse pulse
etc was made at a LinuxCNC fest where I made it work with a Boxford.

When I suggested adding the extra line I didn't look at the jog
behaviour. Does the Tn command also not work still?

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1912


___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[Gene Heskett]

On Friday 08 January 2021 11:06:00 Andreas Linnarsson wrote:

> It will increase by "1" after every (unsuccessful) completion of a fwd
> jog. After position 6 it will wrap around to 1, even though the index
> position isn't sensed. I understand if index isn't used after homing
> though.
>
> Do you know anyone who has used the carousel comp with this kind tool
> turret? Or am I the first? If they've made it work I'm quite curious
> about what their encoder wheel looked like or how they did it. I've
> 3D-printed a variety of designs in search for a pattern that the
> carousel comp would be compatible with but I cannot really get my head
> around if another design would be the solution. I haven't found
> anything that works so far and logically I can't see how it would work
> to stop on the first pocket trigger. Not since that will always be the
> same as the last one due to the reverse locking.
>
> Regards,
>
> Andreas

Can you describe your signal generation as it turns?

It seems to me that your next pulse s/b derived not from the carousel 
position, but from the locking pawl as it falls off a tooth and it then 
backs up against the new tooth it just fell off of to initiate the lock. 

And that the falling pawl generates a pulse to initiate the reversal 
itself. So you get one and only one edge pulse as it falls. And one 
might detect the locked rotor amps of the motor being stopped by the 
pawl to both turn off the motor, and tell linuxcnc that a lock in the 
new position has been achieved.

I would also if switching the motor with a relay, would use 2, wired so 
the when off, the motor armature is shorted, so the sequence would be, 
see the pawl fall, turn off the motor fwd relay so its back side 
contacts short the motor, stopping it fairly rapidly but gently, then 
enable the reverse relay maybe 100 milliseconds later. Lots easier on 
things too. Particularly the psu running the motor. There is also, for 
about 7 dollars US a BTS 7960 half bridge based board with 2 of those 
with a 27 volt rating that can switch 43 amps that does essentially the 
same thing as the relays and a bunch cheaper. I am controlling a 100 
watt 24 volt motor to drive a BS-1 indexing head from linuxcnc.  No 
heating even when my pid is miss set and its oscillating violently. I've 
a 450 watt 24 volt supply its crowbarred and turned off for a 3 minute 
cooldown, but no detectable heat from this board. Phenomenal little 2" 
square board for 7 bucks. Driving it both directions with a pwmgen in 
mode 2.

But I don't have such a tool changer critter myself, so I'm just 
guessing. My holdup is figuring out how to bold the spindle and drop a 
ratcheting impact tool onto the drawbar and run it, its an R8 spindle. 
And the end wrench supplied does not exert enough force to adequately 
hold the tool. So the ratcheting screwdriver is needed.

> Den fre 8 jan. 2021 kl 15:02 skrev andy pugh :
> > On Fri, 8 Jan 2021 at 13:27, Andreas Linnarsson
> >
> >  wrote:
> > > Did I miss(do) anything?
> >
> > I don't think so.
> >
> > Can you look at the current-position HAL pin to see if it decreases
> > by 1 at the end of the reverse pulse?
> >
> > --
> > atp
> > "A motorcycle is a bicycle with a pandemonium attachment and is
> > designed for the especial use of mechanical geniuses, daredevils and
> > lunatics."
> > — George Fitch, Atlanta Constitution Newspaper, 1912
> >
> >
> > ___
> > Emc-users mailing list
> > Emc-users@lists.sourceforge.net
> > https://lists.sourceforge.net/lists/listinfo/emc-users
>
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users


Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
If we desire respect for the law, we must first make the law respectable.
 - Louis D. Brandeis
Genes Web page 


___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[Andreas Linnarsson]

Well, my M6 remap is an admittedly rudimentary "best shot" at it so the
problem might in the code there. But unfortunately it doesn't work. It
doesn't cast any errors but nothing happens when i run "M6 Tn". It "runs"
for a while and then pauses.

Does the Boxford use a gray encoding scheme or something similar? I that
case I understand why it works. But I'm wondering if anyone has been
successful with the "Index" encoding in combination with a reverse lock.

Den fre 8 jan. 2021 kl 17:39 skrev andy pugh :

> On Fri, 8 Jan 2021 at 16:08, Andreas Linnarsson
>  wrote:
>
> > Do you know anyone who has used the carousel comp with this kind tool
> > turret? Or am I the first?
>
> You may be the first with an Emco. The addition of the reverse pulse
> etc was made at a LinuxCNC fest where I made it work with a Boxford.
>
> When I suggested adding the extra line I didn't look at the jog
> behaviour. Does the Tn command also not work still?
>
> --
> atp
> "A motorcycle is a bicycle with a pandemonium attachment and is
> designed for the especial use of mechanical geniuses, daredevils and
> lunatics."
> — George Fitch, Atlanta Constitution Newspaper, 1912
>
>
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>


toolchange_index.hal
Description: Binary data


toolchange.ngc
Description: Binary data
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[andy pugh]

On Fri, 8 Jan 2021 at 17:34, Andreas Linnarsson
 wrote:

> Well, my M6 remap is an admittedly rudimentary "best shot" at it so the
> problem might in the code there. But unfortunately it doesn't work. It
> doesn't cast any errors but nothing happens when i run "M6 Tn". It "runs"
> for a while and then pauses.

That sounds like it thinks that it is already where it needs to be.
Does it do that even when the tool requested is a large number of
steps away?
I think that it needs investigation with Halmeter and/or Halscope
looking at the current-position and target-position.

> Does the Boxford use a gray encoding scheme or something similar?

No, it has the most rudimentary of target discs, 4 segments, 2 black,
2 silver, (As far as I remember)

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1912


___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[Andreas Linnarsson]

I don't know if you've seen my attached picture before but it shows the
signals during a complete turn of the turret (slightly more than a turn to
be honest). The TOOL CMD shows the period of when the 24V are applied to
the DC motor that drives the turret forward, about 3.3 seconds in time. The
total tool change time is about 4.9 seconds including locking. The STROBE
is the pocket sensor which sort of indicates the border between the
pockets. To shift to the next tool pocket it's therefore not sufficient to
back off when that signal is reached the first time. As can be seen from
the recording using the stock hardware one must continue past the first
trigger, continuing until the next one before backing off. I might switch
to better hardware eventually but spontaneously it feels like this can be
solved in software - hopefully. :-)

Regarding your tool changer I've understood that a ratcheting screwdriver
is a good solution for that. If it works it doesn't matter it if makes some
noise.

[image: Turn_50_Turrret.PNG]


Den fre 8 jan. 2021 kl 18:32 skrev Gene Heskett :

> On Friday 08 January 2021 11:06:00 Andreas Linnarsson wrote:
>
> > It will increase by "1" after every (unsuccessful) completion of a fwd
> > jog. After position 6 it will wrap around to 1, even though the index
> > position isn't sensed. I understand if index isn't used after homing
> > though.
> >
> > Do you know anyone who has used the carousel comp with this kind tool
> > turret? Or am I the first? If they've made it work I'm quite curious
> > about what their encoder wheel looked like or how they did it. I've
> > 3D-printed a variety of designs in search for a pattern that the
> > carousel comp would be compatible with but I cannot really get my head
> > around if another design would be the solution. I haven't found
> > anything that works so far and logically I can't see how it would work
> > to stop on the first pocket trigger. Not since that will always be the
> > same as the last one due to the reverse locking.
> >
> > Regards,
> >
> > Andreas
>
> Can you describe your signal generation as it turns?
>
> It seems to me that your next pulse s/b derived not from the carousel
> position, but from the locking pawl as it falls off a tooth and it then
> backs up against the new tooth it just fell off of to initiate the lock.
>
> And that the falling pawl generates a pulse to initiate the reversal
> itself. So you get one and only one edge pulse as it falls. And one
> might detect the locked rotor amps of the motor being stopped by the
> pawl to both turn off the motor, and tell linuxcnc that a lock in the
> new position has been achieved.
>
> I would also if switching the motor with a relay, would use 2, wired so
> the when off, the motor armature is shorted, so the sequence would be,
> see the pawl fall, turn off the motor fwd relay so its back side
> contacts short the motor, stopping it fairly rapidly but gently, then
> enable the reverse relay maybe 100 milliseconds later. Lots easier on
> things too. Particularly the psu running the motor. There is also, for
> about 7 dollars US a BTS 7960 half bridge based board with 2 of those
> with a 27 volt rating that can switch 43 amps that does essentially the
> same thing as the relays and a bunch cheaper. I am controlling a 100
> watt 24 volt motor to drive a BS-1 indexing head from linuxcnc.  No
> heating even when my pid is miss set and its oscillating violently. I've
> a 450 watt 24 volt supply its crowbarred and turned off for a 3 minute
> cooldown, but no detectable heat from this board. Phenomenal little 2"
> square board for 7 bucks. Driving it both directions with a pwmgen in
> mode 2.
>
> But I don't have such a tool changer critter myself, so I'm just
> guessing. My holdup is figuring out how to bold the spindle and drop a
> ratcheting impact tool onto the drawbar and run it, its an R8 spindle.
> And the end wrench supplied does not exert enough force to adequately
> hold the tool. So the ratcheting screwdriver is needed.
>
> > Den fre 8 jan. 2021 kl 15:02 skrev andy pugh :
> > > On Fri, 8 Jan 2021 at 13:27, Andreas Linnarsson
> > >
> > >  wrote:
> > > > Did I miss(do) anything?
> > >
> > > I don't think so.
> > >
> > > Can you look at the current-position HAL pin to see if it decreases
> > > by 1 at the end of the reverse pulse?
> > >
> > > --
> > > atp
> > > "A motorcycle is a bicycle with a pandemonium attachment and is
> > > designed for the especial use of mechanical geniuses, daredevils and
> > > lunatics."
> > > — George Fitch, Atlanta Constitution Newspaper, 1912
> > >
> > >
> > > ___
> > > Emc-users mailing list
> > > Emc-users@lists.sourceforge.net
> > > https://lists.sourceforge.net/lists/listinfo/emc-users
> >
> > ___
> > Emc-users mailing list
> > Emc-users@lists.sourceforge.net
> > https://lists.sourceforge.net/lists/listinfo/emc-users
>
>
> Cheers, Gene Heskett
> --
> "There are f

Re: [Emc-users] Carousel component indexing/trigger problem

[andy pugh]

On Fri, 8 Jan 2021 at 18:03, Andreas Linnarsson
 wrote:
>
> I don't know if you've seen my attached picture before but it shows the
> signals during a complete turn of the turret (slightly more than a turn to
> be honest).

Yes, (but it isn't making much sense to me).
But I want to see what carousel is doing that doesn't match what I expect.

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1912


___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[Andreas Linnarsson]

I could record a movie clip if it helps but I've commented the picture
hoping that it would clarify what is happening during a tool change. Can I
send movie clips to the mailing list? It didn't accept zip-files the last
time I tried...

Regards,

Andreas

Den fre 8 jan. 2021 kl 19:28 skrev andy pugh :

> On Fri, 8 Jan 2021 at 18:03, Andreas Linnarsson
>  wrote:
> >
> > I don't know if you've seen my attached picture before but it shows the
> > signals during a complete turn of the turret (slightly more than a turn
> to
> > be honest).
>
> Yes, (but it isn't making much sense to me).
> But I want to see what carousel is doing that doesn't match what I expect.
>
> --
> atp
> "A motorcycle is a bicycle with a pandemonium attachment and is
> designed for the especial use of mechanical geniuses, daredevils and
> lunatics."
> — George Fitch, Atlanta Constitution Newspaper, 1912
>
>
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>


PC Turn 50 Turret.pdf
Description: Adobe PDF document
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[andy pugh]

On Fri, 8 Jan 2021 at 19:53, Andreas Linnarsson
 wrote:
>
> I could record a movie clip if it helps but I've commented the picture
> hoping that it would clarify what is happening during a tool change.

Are you aware of Halscope? A halscope trace of the input and output
pins of the carousel component would be very useful.

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1912


___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[Andreas Linnarsson]

Yes, I'm aware of the HAL scope but I'm not very experienced using it. But
I'll try to record a sample of a turret jog.

Den fre 8 jan. 2021 kl 21:59 skrev andy pugh :

> On Fri, 8 Jan 2021 at 19:53, Andreas Linnarsson
>  wrote:
> >
> > I could record a movie clip if it helps but I've commented the picture
> > hoping that it would clarify what is happening during a tool change.
>
> Are you aware of Halscope? A halscope trace of the input and output
> pins of the carousel component would be very useful.
>
> --
> atp
> "A motorcycle is a bicycle with a pandemonium attachment and is
> designed for the especial use of mechanical geniuses, daredevils and
> lunatics."
> — George Fitch, Atlanta Constitution Newspaper, 1912
>
>
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>

___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[Gregg Eshelman via Emc-users]

 Look for emco cnc lathe compact 5 turret on YouTube and you'll see several 
videos of it operating, stock and modified. Stock, they are pretty at changing 
tools.This guy used EdingCNC and wrote a subroutine to operate the upgraded 
stepper motor he put on the turret. Quite speedy changes. The original steppers 
on the Compact 5 were just 75 steps rather than the typical 180.

"The Emco turret has a ratchet wheel and a pawl to keep the turretplate in 
position.. i wrote the subroutine for the toolchange so that it rotates 4 
degrees extra to release the pawl, and then 10 degrees back to put the ratchet 
wheel firmly to the pawl.. with this ratchet wheel.. the turret can only turn 
in one direction.. if you look closely.. you can see the turret turning back a 
bit after a toolchange." 
https://www.youtube.com/watch?v=bfIzuWAEZiw
Here's a demo of a stock standalone Compact 5 CNC lathe turret. Note it rotates 
quite a bit farther past the tool position before backing up to lock, and the 
rotation speed is slower. Crazy thing was programmed to be able to advance 6 
positions, which runs it around 360 degrees back to the same tool. Why not have 
it ignore commands for a 6 position change?
https://www.youtube.com/watch?v=-kNYLt6ne_w
'Tis a shame the Welturn and Welmill upgrades for the Emco standalone lathe and 
mill are long out of production and nobody has bothered to clone or make lower 
complexity functional equivalents with 21st century components.

On Friday, January 8, 2021, 5:07:36 AM MST, andy pugh  
wrote:  
 On Fri, 8 Jan 2021 at 07:26, Andreas Linnarsson
 wrote:

> Is it possible to have the carousel component ignore the first pocket
> sensor trigger in order to proceed to the next slot or, can i be run in
> some kind of timer mode so this doesn't happen? Any inputs would be greatly
> appreciated!

Can you tell us your loadrt string for carousel? So that I can work
out how the component is behaving in your case?  
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[Les Newell]

Many moons ago I converted a Compact5 to Mach3. I used a larger DC motor 
on the tool changer with a cam operated switch on the back end of the 
turret spindle to count tool positions. Tool changes took less than a 
second. I have a video on YouTube, filmed with a mouldy potato 

You can just about make out the aluminium cam on the back of the turret 
spindle. The Z axis stepper was replaced with a servo but the X axis 
stepper was retained.


Les

On 09/01/2021 02:30, Gregg Eshelman via Emc-users wrote:

  Look for emco cnc lathe compact 5 turret on YouTube and you'll see several 
videos of it operating, stock and modified. Stock, they are pretty at changing 
tools.This guy used EdingCNC and wrote a subroutine to operate the upgraded 
stepper motor he put on the turret. Quite speedy changes. The original steppers 
on the Compact 5 were just 75 steps rather than the typical 180.

"The Emco turret has a ratchet wheel and a pawl to keep the turretplate in 
position.. i wrote the subroutine for the toolchange so that it rotates 4 degrees extra 
to release the pawl, and then 10 degrees back to put the ratchet wheel firmly to the 
pawl.. with this ratchet wheel.. the turret can only turn in one direction.. if you look 
closely.. you can see the turret turning back a bit after a toolchange."
https://www.youtube.com/watch?v=bfIzuWAEZiw
Here's a demo of a stock standalone Compact 5 CNC lathe turret. Note it rotates 
quite a bit farther past the tool position before backing up to lock, and the 
rotation speed is slower. Crazy thing was programmed to be able to advance 6 
positions, which runs it around 360 degrees back to the same tool. Why not have 
it ignore commands for a 6 position change?
https://www.youtube.com/watch?v=-kNYLt6ne_w
'Tis a shame the Welturn and Welmill upgrades for the Emco standalone lathe and 
mill are long out of production and nobody has bothered to clone or make lower 
complexity functional equivalents with 21st century components.

 On Friday, January 8, 2021, 5:07:36 AM MST, andy pugh  
wrote:
  On Fri, 8 Jan 2021 at 07:26, Andreas Linnarsson
 wrote:





___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[andy pugh]

On Sat, 9 Jan 2021 at 02:32, Gregg Eshelman via Emc-users
 wrote:
>
>  Look for emco cnc lathe compact 5 turret on YouTube and you'll see several 
> videos of it operating, stock and modified. Stock, they are pretty at 
> changing tools.This guy used EdingCNC and wrote a subroutine to operate the 
> upgraded stepper motor he put on the turret.

Yes, but I think that it is a much better idea to fix the existing
LinuxCNC module that is intended to work with these turrets but seems
to not work properly.

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1912


___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[Gene Heskett]

On Saturday 09 January 2021 06:48:33 andy pugh wrote:

> On Sat, 9 Jan 2021 at 02:32, Gregg Eshelman via Emc-users
>
>  wrote:
> >  Look for emco cnc lathe compact 5 turret on YouTube and you'll see
> > several videos of it operating, stock and modified. Stock, they are
> > pretty at changing tools.This guy used EdingCNC and wrote a
> > subroutine to operate the upgraded stepper motor he put on the
> > turret.
>
> Yes, but I think that it is a much better idea to fix the existing
> LinuxCNC module that is intended to work with these turrets but seems
> to not work properly.

And that module is called? 

Thanks Andy.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
If we desire respect for the law, we must first make the law respectable.
 - Louis D. Brandeis
Genes Web page 


___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[andrew beck]

Just link a YouTube video here

On Sat, Jan 9, 2021, 8:54 AM Andreas Linnarsson <
andreas.linnars...@gmail.com> wrote:

> I could record a movie clip if it helps but I've commented the picture
> hoping that it would clarify what is happening during a tool change. Can I
> send movie clips to the mailing list? It didn't accept zip-files the last
> time I tried...
>
> Regards,
>
> Andreas
>
> Den fre 8 jan. 2021 kl 19:28 skrev andy pugh :
>
> > On Fri, 8 Jan 2021 at 18:03, Andreas Linnarsson
> >  wrote:
> > >
> > > I don't know if you've seen my attached picture before but it shows the
> > > signals during a complete turn of the turret (slightly more than a turn
> > to
> > > be honest).
> >
> > Yes, (but it isn't making much sense to me).
> > But I want to see what carousel is doing that doesn't match what I
> expect.
> >
> > --
> > atp
> > "A motorcycle is a bicycle with a pandemonium attachment and is
> > designed for the especial use of mechanical geniuses, daredevils and
> > lunatics."
> > — George Fitch, Atlanta Constitution Newspaper, 1912
> >
> >
> > ___
> > Emc-users mailing list
> > Emc-users@lists.sourceforge.net
> > https://lists.sourceforge.net/lists/listinfo/emc-users
> >
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>

___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[Andreas Linnarsson]

Not that elegant of a solution but I did some small changes to
component.comp that made it work at least physically. I changed line 235 to:
target = current_position + jog_fwd - jog_rev + 1; // Added "+ 1"

The only problem I face is that the current tool number increase by two for
every fwd jog...

Den lör 9 jan. 2021 kl 21:50 skrev Andreas Linnarsson <
andreas.linnars...@gmail.com>:

> Andy: Here's my first shot at using the HAL scope. As you can se the motor
> is stopped as soon as the pocket sensor that it backed away from triggers
> again. Thus the current-position increase at every push of the "fwd jog"
> even though it just goes back and forth within the same pocket. Please get
> back if you want me to add any other signals and I will rerun the capture.
> One thing that crossed my mind; would a quick-and-dirty fix be to "add one"
> to the requested tool in carousel.comp, thus making it go one step to far
> before reversing? Since it only can go in one direction I think that might
> do the trick. But I havn't analyzed all aspects of that solution.
>
> Andrew: I will try and record a movie clip if that will help to identify
> the problem. But I think we've identified the problem. But a general
> solution for the carousel.comp might be the harder part to accomplish.
>
> Regards,
>
> Andreas
>
> Den lör 9 jan. 2021 kl 21:35 skrev andrew beck :
>
>> Just link a YouTube video here
>>
>> On Sat, Jan 9, 2021, 8:54 AM Andreas Linnarsson <
>> andreas.linnars...@gmail.com> wrote:
>>
>> > I could record a movie clip if it helps but I've commented the picture
>> > hoping that it would clarify what is happening during a tool change.
>> Can I
>> > send movie clips to the mailing list? It didn't accept zip-files the
>> last
>> > time I tried...
>> >
>> > Regards,
>> >
>> > Andreas
>> >
>> > Den fre 8 jan. 2021 kl 19:28 skrev andy pugh :
>> >
>> > > On Fri, 8 Jan 2021 at 18:03, Andreas Linnarsson
>> > >  wrote:
>> > > >
>> > > > I don't know if you've seen my attached picture before but it shows
>> the
>> > > > signals during a complete turn of the turret (slightly more than a
>> turn
>> > > to
>> > > > be honest).
>> > >
>> > > Yes, (but it isn't making much sense to me).
>> > > But I want to see what carousel is doing that doesn't match what I
>> > expect.
>> > >
>> > > --
>> > > atp
>> > > "A motorcycle is a bicycle with a pandemonium attachment and is
>> > > designed for the especial use of mechanical geniuses, daredevils and
>> > > lunatics."
>> > > — George Fitch, Atlanta Constitution Newspaper, 1912
>> > >
>> > >
>> > > ___
>> > > Emc-users mailing list
>> > > Emc-users@lists.sourceforge.net
>> > > https://lists.sourceforge.net/lists/listinfo/emc-users
>> > >
>> > ___
>> > Emc-users mailing list
>> > Emc-users@lists.sourceforge.net
>> > https://lists.sourceforge.net/lists/listinfo/emc-users
>> >
>>
>> ___
>> Emc-users mailing list
>> Emc-users@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/emc-users
>>
>

___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[andy pugh]

On Sat, 9 Jan 2021 at 21:32, Andreas Linnarsson
 wrote:

> target = current_position + jog_fwd - jog_rev + 1; // Added "+ 1"
>
> The only problem I face is that the current tool number increase by two for
> every fwd jog...

I worry that you are concentrating too much on jogging and not enough
on correct behaviour with a Tn command.

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1912


___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[Andreas Linnarsson]

You mean that the Tn might work correctly even though the fwd jog doesn't?
Will the carousel component behave differently if I request T4 from say T3
compared to forward jog from T3? I will try this out since my main goal
isn't to jog around the turret manually but make it work for the intended
use, of course.

Den lör 9 jan. 2021 kl 23:46 skrev andy pugh :

> On Sat, 9 Jan 2021 at 21:32, Andreas Linnarsson
>  wrote:
>
> > target = current_position + jog_fwd - jog_rev + 1; // Added "+ 1"
> >
> > The only problem I face is that the current tool number increase by two
> for
> > every fwd jog...
>
> I worry that you are concentrating too much on jogging and not enough
> on correct behaviour with a Tn command.
>
> --
> atp
> "A motorcycle is a bicycle with a pandemonium attachment and is
> designed for the especial use of mechanical geniuses, daredevils and
> lunatics."
> — George Fitch, Atlanta Constitution Newspaper, 1912
>
>
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>

___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[andy pugh]

On Sun, 10 Jan 2021 at 09:45, Andreas Linnarsson
 wrote:
>
>  Andy: Here's my first shot at using the HAL scope. As you can se the motor
> is stopped as soon as the pocket sensor that it backed away from triggers
> again. Thus the current-position increase at every push of the "fwd jog"

There is no obvious sign of the new line (curent_position -= 1) taking effect.

I think the trace needs to include the carousel.N.state parameter, so
we can follow the flow through the logic.

But, my first thought is that jogging uses state-5, whereas tool
change uses state-2 -> state-3 ->  state-4.
The state-5 logic is rather different (and does not include the timer
logic for the reverse lock move)

In fact I would say that the jog logic might be broken for your style
of tool changer.

You shouldn't actually need an M6 remap to run the carousel component,
it should connect to the standard HAL pins.
Part of the point of carousel was to remove the need of a remap in many cases.

It should just need iocontrol.0.tool-change connected to "enable" and
iocontro.0.tool-changed connected to "ready" in addition to the links
you already have.

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1912


___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[Andreas Linnarsson]

I reached milestone today when I got the tool change command to work - sort
of. The "iocontrol.0.tool.*" pins were not connected correctly. My
toolchange.ngc is now called correctly and it activates the carousel
component as expected. The first pocket trigger still generates problems
since the carousel continue to think it has gone when step further than it
really has. After homing it stops at pocket #1 and the
carousel.0.current-position also indicates #1. But it's after that things
gets a bit strange. Issuing T5 M6 will move the turret to the physical
pocket #4 whereas carousel.0.current-position will indicate #6.

Every tool change from there on will offset the commanded tool pocket
compared to the actual tool pocket by one, where the physical pocket is
lagging the commanded tool. If I try to shift just one tool forward it has
the same result as when I tried to jog one position earlier - it just goes
back and forth within the same pocket, although increasing
carousel.0.current-position by one . To make it change one physical
position forwards I thus have to command a tool that is two positions ahead.

Adding "+ 1" to the target on line 235 in carousel.comp takes care of the
problem from a physical aspect but carousel.0.pocket-number won't match
carousel.0.current-position afterwards as a result of that change. Due to
that carousel.0.ready will not go high and therefore the subroutine will
timeout.  I haven't found out how to deduct "one" from
carousel.0.pocket-number variable but if I do think things will match. Not
a global solution but it might be good enough for me.

*edit*
I got your last answer right before sending this so I will evaluate your
response and see what I can come up with. I didn't know that it wasn't
necessary to remap to make this work. I followed your advice to take a look
at the sim.axis.vismach.VMC_toolchange example in a thread at the forum.
But that part seems to work anyway.

Regards,

Andreas

Den sön 10 jan. 2021 kl 10:43 skrev Andreas Linnarsson <
andreas.linnars...@gmail.com>:

> Andy: Here's my first shot at using the HAL scope. As you can se the motor
> is stopped as soon as the pocket sensor that it backed away from triggers
> again. Thus the current-position increase at every push of the "fwd jog"
> even though it just goes back and forth within the same pocket. Please get
> back if you want me to add any other signals and I will rerun the capture.
> As you've already commented I'm having some problem with the call of the
> remapped M6 command but I'm a bit unsure about how that would change the
> behaviour of the carousel component. But I'm invesigating that as well now,
> of course.
>
> Andrew: I will try and record a movie clip if that will help to identify
> the problem but the behaviour I experience is logical, to me at least.
>
> Regards,
>
> Andreas
>
> Den lör 9 jan. 2021 kl 23:52 skrev Andreas Linnarsson <
> andreas.linnars...@gmail.com>:
>
>> You mean that the Tn might work correctly even though the fwd jog
>> doesn't? Will the carousel component behave differently if I request T4
>> from say T3 compared to forward jog from T3? I will try this out since my
>> main goal isn't to jog around the turret manually but make it work for the
>> intended use, of course.
>>
>> Den lör 9 jan. 2021 kl 23:46 skrev andy pugh :
>>
>>> On Sat, 9 Jan 2021 at 21:32, Andreas Linnarsson
>>>  wrote:
>>>
>>> > target = current_position + jog_fwd - jog_rev + 1; // Added "+ 1"
>>> >
>>> > The only problem I face is that the current tool number increase by
>>> two for
>>> > every fwd jog...
>>>
>>> I worry that you are concentrating too much on jogging and not enough
>>> on correct behaviour with a Tn command.
>>>
>>> --
>>> atp
>>> "A motorcycle is a bicycle with a pandemonium attachment and is
>>> designed for the especial use of mechanical geniuses, daredevils and
>>> lunatics."
>>> — George Fitch, Atlanta Constitution Newspaper, 1912
>>>
>>>
>>> ___
>>> Emc-users mailing list
>>> Emc-users@lists.sourceforge.net
>>> https://lists.sourceforge.net/lists/listinfo/emc-users
>>>
>>

___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[Andreas Linnarsson]

Forgot to attach the pictures in the last response. It shows the behaviour
without the "+ 1" on line 235.

Den sön 10 jan. 2021 kl 13:20 skrev Andreas Linnarsson <
andreas.linnars...@gmail.com>:

> I reached milestone today when I got the tool change command to work -
> sort of. The "iocontrol.0.tool.*" pins were not connected correctly. My
> toolchange.ngc is now called correctly and it activates the carousel
> component as expected. The first pocket trigger still generates problems
> since the carousel continue to think it has gone when step further than it
> really has. After homing it stops at pocket #1 and the
> carousel.0.current-position also indicates #1. But it's after that things
> gets a bit strange. Issuing T5 M6 will move the turret to the physical
> pocket #4 whereas carousel.0.current-position will indicate #6.
>
> Every tool change from there on will offset the commanded tool pocket
> compared to the actual tool pocket by one, where the physical pocket is
> lagging the commanded tool. If I try to shift just one tool forward it has
> the same result as when I tried to jog one position earlier - it just goes
> back and forth within the same pocket, although increasing
> carousel.0.current-position by one . To make it change one physical
> position forwards I thus have to command a tool that is two positions ahead.
>
> Adding "+ 1" to the target on line 235 in carousel.comp takes care of the
> problem from a physical aspect but carousel.0.pocket-number won't match
> carousel.0.current-position afterwards as a result of that change. Due to
> that carousel.0.ready will not go high and therefore the subroutine will
> timeout.  I haven't found out how to deduct "one" from
> carousel.0.pocket-number variable but if I do think things will match. Not
> a global solution but it might be good enough for me.
>
> *edit*
> I got your last answer right before sending this so I will evaluate your
> response and see what I can come up with. I didn't know that it wasn't
> necessary to remap to make this work. I followed your advice to take a look
> at the sim.axis.vismach.VMC_toolchange example in a thread at the forum.
> But that part seems to work anyway.
>
> Regards,
>
> Andreas
>
> Den sön 10 jan. 2021 kl 10:43 skrev Andreas Linnarsson <
> andreas.linnars...@gmail.com>:
>
>> Andy: Here's my first shot at using the HAL scope. As you can se the
>> motor is stopped as soon as the pocket sensor that it backed away from
>> triggers again. Thus the current-position increase at every push of the
>> "fwd jog" even though it just goes back and forth within the same pocket.
>> Please get back if you want me to add any other signals and I will rerun
>> the capture. As you've already commented I'm having some problem with the
>> call of the remapped M6 command but I'm a bit unsure about how that would
>> change the behaviour of the carousel component. But I'm invesigating that
>> as well now, of course.
>>
>> Andrew: I will try and record a movie clip if that will help to identify
>> the problem but the behaviour I experience is logical, to me at least.
>>
>> Regards,
>>
>> Andreas
>>
>> Den lör 9 jan. 2021 kl 23:52 skrev Andreas Linnarsson <
>> andreas.linnars...@gmail.com>:
>>
>>> You mean that the Tn might work correctly even though the fwd jog
>>> doesn't? Will the carousel component behave differently if I request T4
>>> from say T3 compared to forward jog from T3? I will try this out since my
>>> main goal isn't to jog around the turret manually but make it work for the
>>> intended use, of course.
>>>
>>> Den lör 9 jan. 2021 kl 23:46 skrev andy pugh :
>>>
 On Sat, 9 Jan 2021 at 21:32, Andreas Linnarsson
  wrote:

 > target = current_position + jog_fwd - jog_rev + 1; // Added "+ 1"
 >
 > The only problem I face is that the current tool number increase by
 two for
 > every fwd jog...

 I worry that you are concentrating too much on jogging and not enough
 on correct behaviour with a Tn command.

 --
 atp
 "A motorcycle is a bicycle with a pandemonium attachment and is
 designed for the especial use of mechanical geniuses, daredevils and
 lunatics."
 — George Fitch, Atlanta Constitution Newspaper, 1912


 ___
 Emc-users mailing list
 Emc-users@lists.sourceforge.net
 https://lists.sourceforge.net/lists/listinfo/emc-users

>>>
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[andy pugh]

I think that I might have a better fix.

Replace lines 198 and 214 with:

// Fix for unidirectional changers with a latch
if (inst_dir == 2 || motor_fwd) old_index = sense(1);

And remove the previous changes.

I think that will work for both jog and tool change.

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1912


___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[Andreas Linnarsson]

I've implemented your changes to carousel.comp but it still doesn't behave
as expected. It still moves so it didn't break anything. But the turret
behaves differently depending on if I first home the it using a jog-fwd
pulse or if I run Tn M6 directly letting it home as a part of that command.

Situation 1: The turret is homed using jog-fwd pulse. After homing;
current-position = 1,
pocket-number = 0
physical pocket = #1.

The first tool change (T4 M6) makes the turret end up in the correct pocket
(#4). After that
current-position = 5,
pocket-number = 5.
physical pocket = #4
.
From now on the turret will take one less physical step compared to what is
ordered in the same way as I described earlier.

Situation 2: The turret is homed as part of Tn M6 command. This makes the
turret move one physical step too far during the first tool change. After
T3 M6 change is issued we will thus have the following situation:
current-position = 4,
pocket-number = 4.
physical pocket = #4

As can be seen we end up with physical pocket #4 even though we wanted T3.
From now on it will continue to behave as described above, lagging one
physical step compared to what is demanded for every tool change.



Den sön 10 jan. 2021 kl 14:02 skrev andy pugh :

> I think that I might have a better fix.
>
> Replace lines 198 and 214 with:
>
> // Fix for unidirectional changers with a latch
> if (inst_dir == 2 || motor_fwd) old_index = sense(1);
>
> And remove the previous changes.
>
> I think that will work for both jog and tool change.
>
> --
> atp
> "A motorcycle is a bicycle with a pandemonium attachment and is
> designed for the especial use of mechanical geniuses, daredevils and
> lunatics."
> — George Fitch, Atlanta Constitution Newspaper, 1912
>
>
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>

___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[andy pugh]

On Sun, 10 Jan 2021 at 14:14, Andreas Linnarsson
 wrote:
>
> I've implemented your changes to carousel.comp but it still doesn't behave
> as expected.

Did you undo the other changes?

Are you using motor-vel or the fwd-rev outputs?

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1912


___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[Andreas Linnarsson]

Yes I undid all changes by downloading the carousel.comp again, just to
make sure I wouldn't mix up the files.

I'm only using the motor-fwd since the DC motor is controlled by a relay.

Den sön 10 jan. 2021 kl 15:22 skrev andy pugh :

> On Sun, 10 Jan 2021 at 14:14, Andreas Linnarsson
>  wrote:
> >
> > I've implemented your changes to carousel.comp but it still doesn't
> behave
> > as expected.
>
> Did you undo the other changes?
>
> Are you using motor-vel or the fwd-rev outputs?
>
> --
> atp
> "A motorcycle is a bicycle with a pandemonium attachment and is
> designed for the especial use of mechanical geniuses, daredevils and
> lunatics."
> — George Fitch, Atlanta Constitution Newspaper, 1912
>
>
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>

___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[andy pugh]

On Sun, 10 Jan 2021 at 14:53, Andreas Linnarsson
 wrote:
>
> Yes I undid all changes by downloading the carousel.comp again, just to
> make sure I wouldn't mix up the files.
>
> I'm only using the motor-fwd since the DC motor is controlled by a relay.

And motor-rev to pull it against the latch?

Can I see your HAL config? I think there must be something that I am
misunderstanding.

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1912


___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[Andreas Linnarsson]

No, I'ts not necessary to use the motor-rev with this tool changer. The
motor-fwd activates a switching relay that applies 24V to rotate the motor
forward. When motor-fwd is low the relay switches so that -12V is applied
instead, thereby reversing the motor until it locks in place. The -12V is
constantly on between tool changes in order to have a constant locking
torque applied. From that perspective it's quite easy to use.



Den sön 10 jan. 2021 kl 16:13 skrev andy pugh :

> On Sun, 10 Jan 2021 at 14:53, Andreas Linnarsson
>  wrote:
> >
> > Yes I undid all changes by downloading the carousel.comp again, just to
> > make sure I wouldn't mix up the files.
> >
> > I'm only using the motor-fwd since the DC motor is controlled by a relay.
>
> And motor-rev to pull it against the latch?
>
> Can I see your HAL config? I think there must be something that I am
> misunderstanding.
>
> --
> atp
> "A motorcycle is a bicycle with a pandemonium attachment and is
> designed for the especial use of mechanical geniuses, daredevils and
> lunatics."
> — George Fitch, Atlanta Constitution Newspaper, 1912
>
>
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>


toolchange_index.hal
Description: Binary data
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[andy pugh]

The HAL is rather strange. What drives the analogue and digital outputs?

All you should need is:

loadrt carousel pockets=6 encoding=index dir=1
addf carousel.0 servo-thread

#Tool turret encoder input
net car-index carousel.0.sense-0 <= [HMOT](CARD0).7i76.0.0.input-20-not
net car-pulse carousel.0.sense-1 <= [HMOT](CARD0).7i76.0.0.input-21-not

#Tool turret forward output
net car-fwd carousel.0.motor-fwd => [HMOT](CARD0).7i76.0.0.output-00

net tool-prep-loop iocontrol.0.tool-prepare iocontrol.0.tool-prepared
net tool-change-cmd iocontrol.0.tool-change carousel.0.enable
net tool-change-fb iocontrol.0.tool-changed carousel.0.ready
net tool-slot iocontrol.0.tool−prep−pocket carousel.0.pocket-number

loadusr sim_pin carousel.0.jog-fwd

Then when you issue an M6 Tn G43 the tool changer should go to the
numbered pocket.
(You can configure to change on Tnn like most commercial lathes do,
but that can come later, once it works)

Do you have the tools set up in the right pockets? Tnn looks in the
tool table to find which pocket Tnn is in.
Your example above has T3 setting pocket-number to 4, and T4 setting
pocket-number to 5. pocket-number is the _input_ to carousel which
seems to be coming via a G-code output.
If TnM6 is working at all then I think that your remap is still in
place, and is doing odd things.


-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1912


___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[Andreas Linnarsson]

Well, it all comes from the "sim/axis/vismach/VMC_toolchange" example in
LinuxCNC 2.8. I just implemented it as it was used there, but adapted it
for my lathe. So it uses the remap.py, remap.pyc, toolchange.ngc files,
plus the toolchange_index.hal.
But the toolchange.ngc is very small. Just move away the turret, set the
tool, wait for completion and then continue. But that part seems to work.

Regarding the tools I might have been unclear about the result. The lagging
of the physical pocket compared to the required isn't constantly one
position, the lag accumulates by one position for each change. Thereby it
will lag one, two, three, four positions, and so on after the same amount
of changes. The tool table itself is quite simple with T1 i pocket 1, T2 in
pocket 2 , all the way up to T6.

I will try your simplified HAL configuration to if it works but to be
honest I'm having trouble getting my head around how it's supposed to work
if the carousel doesn't ignore the first pocket trigger when changing
pocket. Adding one to the target it will behave physically correct but it
will not *carousel.**N**.ready* not be set high due to some other
discrepancy i suppose.


Den sön 10 jan. 2021 kl 17:34 skrev andy pugh :

> The HAL is rather strange. What drives the analogue and digital outputs?
>
> All you should need is:
>
> loadrt carousel pockets=6 encoding=index dir=1
> addf carousel.0 servo-thread
>
> #Tool turret encoder input
> net car-index carousel.0.sense-0 <= [HMOT](CARD0).7i76.0.0.input-20-not
> net car-pulse carousel.0.sense-1 <= [HMOT](CARD0).7i76.0.0.input-21-not
>
> #Tool turret forward output
> net car-fwd carousel.0.motor-fwd => [HMOT](CARD0).7i76.0.0.output-00
>
> net tool-prep-loop iocontrol.0.tool-prepare iocontrol.0.tool-prepared
> net tool-change-cmd iocontrol.0.tool-change carousel.0.enable
> net tool-change-fb iocontrol.0.tool-changed carousel.0.ready
> net tool-slot iocontrol.0.tool−prep−pocket carousel.0.pocket-number
>
> loadusr sim_pin carousel.0.jog-fwd
>
> Then when you issue an M6 Tn G43 the tool changer should go to the
> numbered pocket.
> (You can configure to change on Tnn like most commercial lathes do,
> but that can come later, once it works)
>
> Do you have the tools set up in the right pockets? Tnn looks in the
> tool table to find which pocket Tnn is in.
> Your example above has T3 setting pocket-number to 4, and T4 setting
> pocket-number to 5. pocket-number is the _input_ to carousel which
> seems to be coming via a G-code output.
> If TnM6 is working at all then I think that your remap is still in
> place, and is doing odd things.
>
>
> --
> atp
> "A motorcycle is a bicycle with a pandemonium attachment and is
> designed for the especial use of mechanical geniuses, daredevils and
> lunatics."
> — George Fitch, Atlanta Constitution Newspaper, 1912
>
>
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>
loadrt carousel pockets=6 encoding=index dir=1 
loadrt conv_float_s32
addf carousel.0 servo-thread
addf conv-float-s32.0 servo-thread


#Tool turret encoder input
net car-index carousel.0.sense-0 <= [HMOT](CARD0).7i76.0.0.input-20-not
net car-pulse carousel.0.sense-1 <= [HMOT](CARD0).7i76.0.0.input-21-not
#net car-pulse carousel.0.sense-1 carousel.0.strobe <= 
[HMOT](CARD0).7i76.0.0.input-21-not
#net car-strobe carousel.0.strobe <= [HMOT](CARD0).7i76.0.0.input-21

#Tool turret forward output
net car-fwd carousel.0.motor-fwd => [HMOT](CARD0).7i76.0.0.output-00


#M64 and M65 digital outputs
net car-enable motion.digital-out-00 => carousel.0.enable

#M68 Analog output
net car-pos-req motion.analog-out-00 conv-float-s32.0.in

#M66 digital inputs
net car-ready carousel.0.ready => motion.digital-in-00
net x-is-homed halui.joint.0.is-homed => motion.digital-in-01

#Carousel Soft Signals

net car-pos-s32 conv-float-s32.0.out carousel.0.pocket-number

net tool-prep-loop iocontrol.0.tool-prepare iocontrol.0.tool-prepared
net tool-change-loop iocontrol.0.tool-change iocontrol.0.tool-changed

loadusr sim_pin carousel.0.jog-fwd

o sub

G53 G0 X80 Z100; Go to safe position for tool change


M68 E0 Q# ;set the carousel to move to the right pocket

M64 P0 ; start carousel
G4 P2
M66 P0 L3 Q10 ; wait for carousel finished
o110 IF [#5399 LT 0]
(Abort, failed to align carousel)
o110 ENDIF
M65 P0; stop carousel



o endsub [1]


M2
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[andy pugh]

On Sun, 10 Jan 2021 at 19:03, Andreas Linnarsson
 wrote:
>
> Well, it all comes from the "sim/axis/vismach/VMC_toolchange" example in
> LinuxCNC 2.8. I just implemented it as it was used there,

But that is running a complex cycle of (simulated) pneumatics and sensors.

It shouldn't be necessary and doesn't seem to be helping. From what
you have shown so far it is sending the pocket number out via an
analogue output, and wrongly.

> I will try your simplified HAL configuration to if it works but to be
> honest I'm having trouble getting my head around how it's supposed to work
> if the carousel doesn't ignore the first pocket trigger when changing
> pocket. Adding one to the target it will behave physically correct but it
> will not *carousel.**N**.ready* not be set high due to some other
> discrepancy i suppose.

The change that I suggested was meant to make it ignore the
high-to-low transition when moving backwards. This in turn should mean
that the extra low-to-high is not seen.

However I think that there is a flaw in my logic. old_index will hold
the previous high value during the reversal, but will reset to low as
soon as motor-fwd is set.

Try moving the old_index update line:

case 'E': // index + position, both edges.
p = current_position;
if (homed){
if ( old_index != sense(1) ){
if (motor_fwd){
p += 1;
if (p > inst_pockets) p -= inst_pockets;
}
if (motor_rev) {
p -= 1;
if (p < 1) p += inst_pockets;
}
// Fix for unidirectional changers with a latch
if (inst_dir == 2 || motor_fwd) old_index = sense(1);
}
}
break;

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1912


___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[andy pugh]

On Sun, 10 Jan 2021 at 20:04, andy pugh  wrote:

> Try moving the old_index update line:

And if that doesn't work, post me your lathe and PC and I will sort it out.

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1912


___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Carousel component indexing/trigger problem

[Andreas Linnarsson]

OK, I've tested a few things now. First of all I went back to my most basic
machine configuration and implemented your version of the
toolchange_index.hal together with the unmodified version of carousel.comp.
Completely ditching the M6 remap solution used earlier. Thereafter I tested
the following:

1. Unmodified carousel.comp
Result: Homing works as it should, but in the same way as before the
physical pocket lags the demanded pocket by one unit for each tool change
and the error is accumulating for each successive tool change.

2. carousel.comp modified with "current_position -= 1;" on row 319 (within
case 3:)
Result. Same as for test 1.

3. carousel.comp modified with "if (inst_dir == 2 || motor_fwd) old_index =
sense(1);" instead of old_index = sense(1);on row 198 and 214.
Result. Same as for test 1.

4. carousel.comp modified with "if (inst_dir == 2 || motor_fwd) old_index =
sense(1);" moved to next row after 196 and 212, in accordance with your
mail from yesterday.
Result: Homing works fine but when issuing a Tn M6 command the turrent
won't stop rotating. After a couple of turns I get an error. I also tried
this using Edge encoding instead of Index for this modification and even
though it did stop by itself it only shifted half the amount of pockets
that was requested. I suppose it's because of the flank triggering for this
option.

I also tested the "target = current_position + jog_fwd - jog_rev + 1; //
Added "+ 1"" but interestingly enough this didn't have any effect as it had
earlier. It still behaved as in test nr 1.

Correct me if I'm wrong but it seems fair to say that this isn't a HAL
issue and that it needs to be addressed withinin the carousel component. Or
is there some functionality within the component that makes it ignore the
first pocket trigger when a reverse latching tool changer is used? What
lines of code are responsible for that functionality?

Regards,

Andreas


Den sön 10 jan. 2021 kl 21:07 skrev andy pugh :

> On Sun, 10 Jan 2021 at 20:04, andy pugh  wrote:
>
> > Try moving the old_index update line:
>
> And if that doesn't work, post me your lathe and PC and I will sort it out.
>
> --
> atp
> "A motorcycle is a bicycle with a pandemonium attachment and is
> designed for the especial use of mechanical geniuses, daredevils and
> lunatics."
> — George Fitch, Atlanta Constitution Newspaper, 1912
>
>
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>

___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users