Re: [Emc-users] mesa 5i25, pncconf working but not linuxcnc (bruno) [SOLVED]
thanks for the explanation. I tried 0 backlash and the reversals caused no error. Then I used the values you suggested for ferror and min_ferror, that still works with low backlash. I will experiment with higher scale (microstepping) and ferror in order to use the correct backlash required by the machine. And work on lowering the backlash... Thanks everybody for the help in figuring this out ! On 2/10/14 7:51 AM, emc-users-requ...@lists.sourceforge.net wrote: > Date: Sun, 9 Feb 2014 23:00:31 + > From: andy pugh > Subject: Re: [Emc-users] mesa 5i25, pncconf working but not linuxcnc > (bruno) > To: "Enhanced Machine Controller (EMC)" > > Message-ID: > > Content-Type: text/plain; charset=ISO-8859-1 > > On 9 February 2014 20:52, bruno wrote: > >> >- if I stop the movement and continue in the same direction : no >> >problem, it will start moving (and stop) smoothly >> >- if I stop the movement and go jog in the opposite direction, I >> >immediately get a following error. > This points very strongly to the problem being the backlash > compensation. (which won't be active in pncconf). > > In fact you have a backlash compensation of 0.05 and an f-error of > 0.005. So on every reversal you are_automatically_ out of position by > 10x your f-error limit. > > Your step scale of 314 steps per mm means that you have a resolution > of 0.003mm, so your f-error is 1 microstep. That's pretty tight, > especially for the high-speed f-error. > Try 0.015 for min_ferror and 0.1 for f-error and things should run > rather better. > > -- atp If you can't fix it, you don't own it. > http://www.ifixit.com/Manifesto -- Managing the Performance of Cloud-Based Applications Take advantage of what the Cloud has to offer - Avoid Common Pitfalls. Read the Whitepaper. http://pubads.g.doubleclick.net/gampad/clk?id=121051231&iu=/4140/ostg.clktrk ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] mesa 5i25, pncconf working but not linuxcnc (bruno)
On 9 February 2014 20:52, bruno wrote: > - if I stop the movement and continue in the same direction : no > problem, it will start moving (and stop) smoothly > - if I stop the movement and go jog in the opposite direction, I > immediately get a following error. This points very strongly to the problem being the backlash compensation. (which won't be active in pncconf). In fact you have a backlash compensation of 0.05 and an f-error of 0.005. So on every reversal you are _automatically_ out of position by 10x your f-error limit. Your step scale of 314 steps per mm means that you have a resolution of 0.003mm, so your f-error is 1 microstep. That's pretty tight, especially for the high-speed f-error. Try 0.015 for min_ferror and 0.1 for f-error and things should run rather better. -- atp If you can't fix it, you don't own it. http://www.ifixit.com/Manifesto -- Managing the Performance of Cloud-Based Applications Take advantage of what the Cloud has to offer - Avoid Common Pitfalls. Read the Whitepaper. http://pubads.g.doubleclick.net/gampad/clk?id=121051231&iu=/4140/ostg.clktrk ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] mesa 5i25, pncconf working but not linuxcnc (bruno)
On Sun, 9 Feb 2014, bruno wrote:
> Date: Sun, 09 Feb 2014 21:52:52 +0100
> From: bruno
> Reply-To: "Enhanced Machine Controller (EMC)"
>
> To: emc-users@lists.sourceforge.net
> Subject: Re: [Emc-users] mesa 5i25, pncconf working but not linuxcnc (bruno)
>
> I tried also with 25% more on stepgen_maxvel and maxaccel. Not better.
stepgen_maxaccel needs to be about 25% more than the maxaccel for each axis
or you _will_ have following errors
>
> I checked the enable, it works fine. Pncconf was generating a different
> named net. Once fixed at least the axis moves. But I get following
> errors and occasional grinding if I increase the MIN_FERROR
>
> I rechecked within pncconf just to make sure I was not dreaming, it
> works great in there. I can let it go back and forth over 100mm at
> 2800mm/min (1/8 microstepping, 0.2 inch per turn leadscrew => scale of
> 314...something...) for as long as I want. Very nice movement, no
> grinding, no missed steps, it stops and reverses smoothly without problems.
>
> then I finish pncconf, let it write the configuration, startup linuxcnc
> and get a following error if I get past like 1mm/min...
>
> So I would say pncconf testing is definitely using something different
> from linuxcnc. How can I check what pncconf is using ?
>
halcmd show param
> Another experiment I just did that gives some interesting results:
> I went back to my stepper driver specs and found I was using values for
> dirhold, dirsetup, steplen and stepspace that were way too small. After
> adjusting them, I can jog an axis up to 2800mm/m, no problem ! however:
>
> - if I stop the movement and continue in the same direction : no
> problem, it will start moving (and stop) smoothly
> - if I stop the movement and go jog in the opposite direction, I
> immediately get a following error.
>
> Bruno
>
> On 2/9/14 8:43 PM, emc-users-requ...@lists.sourceforge.net wrote:
>> Message: 3
>> Date: Sun, 9 Feb 2014 17:08:01 +
>> From: Chris Morley
>> Subject: Re: [Emc-users] Emc-users Digest, Vol 94, Issue 17
>> To: EMC
>> Message-ID:
>> Content-Type: text/plain; charset="iso-8859-1"
>>
>>>
>>> seems like it would be best to specify 0 for both of those as the per
>>> axis limits would still be in place ?
>>> I am going to play with that and see... thanks for the reply
>>>
>> Don't set them to 0. It doesn't work reliably.
>> Pncconf sets them to 25% more.
>>
>> Does your drives require an enable ? I would check that signal with
>> halmeter or a real meter.
>>
>> While pncconf uses the same numbers as it tunes ( as long as you clicked
>> accept)
>> It doesn't use the same HAL code in the test as in linuxcnc so problems can
>> creep in and bugs are always possible.
>>
>> Chris M
>>
>>
>
>
> --
> Managing the Performance of Cloud-Based Applications
> Take advantage of what the Cloud has to offer - Avoid Common Pitfalls.
> Read the Whitepaper.
> http://pubads.g.doubleclick.net/gampad/clk?id=121051231&iu=/4140/ostg.clktrk
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>
Peter Wallace
Mesa Electronics
(\__/)
(='.'=) This is Bunny. Copy and paste bunny into your
(")_(") signature to help him gain world domination.
--
Managing the Performance of Cloud-Based Applications
Take advantage of what the Cloud has to offer - Avoid Common Pitfalls.
Read the Whitepaper.
http://pubads.g.doubleclick.net/gampad/clk?id=121051231&iu=/4140/ostg.clktrk
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] mesa 5i25, pncconf working but not linuxcnc (bruno)
I tried also with 25% more on stepgen_maxvel and maxaccel. Not better. I checked the enable, it works fine. Pncconf was generating a different named net. Once fixed at least the axis moves. But I get following errors and occasional grinding if I increase the MIN_FERROR I rechecked within pncconf just to make sure I was not dreaming, it works great in there. I can let it go back and forth over 100mm at 2800mm/min (1/8 microstepping, 0.2 inch per turn leadscrew => scale of 314...something...) for as long as I want. Very nice movement, no grinding, no missed steps, it stops and reverses smoothly without problems. then I finish pncconf, let it write the configuration, startup linuxcnc and get a following error if I get past like 1mm/min... So I would say pncconf testing is definitely using something different from linuxcnc. How can I check what pncconf is using ? Another experiment I just did that gives some interesting results: I went back to my stepper driver specs and found I was using values for dirhold, dirsetup, steplen and stepspace that were way too small. After adjusting them, I can jog an axis up to 2800mm/m, no problem ! however: - if I stop the movement and continue in the same direction : no problem, it will start moving (and stop) smoothly - if I stop the movement and go jog in the opposite direction, I immediately get a following error. Bruno On 2/9/14 8:43 PM, emc-users-requ...@lists.sourceforge.net wrote: > Message: 3 > Date: Sun, 9 Feb 2014 17:08:01 + > From: Chris Morley > Subject: Re: [Emc-users] Emc-users Digest, Vol 94, Issue 17 > To: EMC > Message-ID: > Content-Type: text/plain; charset="iso-8859-1" > > > > >seems like it would be best to specify 0 for both of those as the per > >axis limits would still be in place ? > >I am going to play with that and see... thanks for the reply > > > Don't set them to 0. It doesn't work reliably. > Pncconf sets them to 25% more. > > Does your drives require an enable ? I would check that signal with > halmeter or a real meter. > > While pncconf uses the same numbers as it tunes ( as long as you clicked > accept) > It doesn't use the same HAL code in the test as in linuxcnc so problems can > creep in and bugs are always possible. > > Chris M > > -- Managing the Performance of Cloud-Based Applications Take advantage of what the Cloud has to offer - Avoid Common Pitfalls. Read the Whitepaper. http://pubads.g.doubleclick.net/gampad/clk?id=121051231&iu=/4140/ostg.clktrk ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
