Re: [Emc-users] Big Tree Tech Re: Mesa Card Stepgens?

[Nicklas SB Karlsson]

> > ...
> 
> If it can revert back to the way LinuxCNC works ( which requires realtime), 
> there is no real advantage to queueing at all. Queueing is used to make 
> up for the lack of realtime. For example Mach3 can do decent spindle 
> synchronized motion, but only if this feature is implemented in the interface 
> hardware. This would be a big step backwards for LinuxCNC.

And then it will come back to the problem how to send enough queued data to 
actuators.

Queue may make sense if for example sending position commands to follow over an 
Ethernet network as position is small
compared to the header. In that case sending many positions in one message 
reduce overhead.


Nicklas Karlsson



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

Re: [Emc-users] Big Tree Tech Re: Mesa Card Stepgens?

[Todd Zuercher via Emc-users]

What happens if your "NOW" time stamped command, is delayed in transmitting to 
the hardware's que and it isn't read until after it's time stamp has happened?  
(Things like this happen without realtime.)  There could be situations where 
too late could almost be worse than not at all.

Todd Zuercher
P. Graham Dunn Inc.
630 Henry Street
Dalton, Ohio 44618
Phone:  (330)828-2105ext. 2031

-Original Message-
From: Chris Albertson 
Sent: Wednesday, April 17, 2024 8:51 PM
To: Enhanced Machine Controller (EMC) 
Subject: Re: [Emc-users] Big Tree Tech Re: Mesa Card Stepgens?

[EXTERNAL EMAIL] Be sure links are safe.

No it does not fail, we can always send new commands in real time to be 
executed “right away” (at the current time) and make the queued execution 
system look and act as if there were no queue.   A command taged with the 
current time will jump to the front of the queue.

The system also allows for very delayed execution of commands, like “in five 
minutes”.   This works well because you can queue a macro to do something like 
poll the voltage of a battery and then place itself back in the queue for five 
minutes later.  But as said with a current time or a special tag that says 
“now” it reverts back to the way LCNC works.

Remember that a time-tagged queue is not executed in order.  It is not  FIFO, 
so a command that is placed in the queue, even with 100 commands waiting does 
not need to wait for the others.


> On Apr 17, 2024, at 5:05 PM, Ray Henry  wrote:
>
> On Wed, 2024-04-17 at 09:13 -0700, Peter Wallace wrote:
>>
>>> LCNC really should be doing this.  If the Measa cards would maintain
>>> a queue and a synchronized clock we would not care at all about
>>> latency.
>>> Klipper
>>> proves the idea works.
>>
>> This is basically what Mach 3/4 do
>>
>> This fails as soon as you need the control to repond to feedback in
>> real time (Spindle synchronized motion, plasma THC etc etc) Its
>> possible of course to build this feedback into the interface
>> hardware, but there is a big advantage to having this feedback in
>> LinuxCNC where is works with any interface hardware, is extensible
>> and open, and where the feedback control math/logic has unlimited
>> resources.
>>
>>
>>
>> Peter Wallace
>
>
> The message on my 25+ year old coffee cup from the NIST Intelligent
> Systems Division reminds me at least three times a week to "SENSE ->
> MODEL -> ACT." in that order.  Thanks.
> Peter.
>
> Ray
>
>
>
>
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://list/
> s.sourceforge.net%2Flists%2Flistinfo%2Femc-users&data=05%7C02%7Ctoddz%
> 40pgrahamdunn.com%7Ccd85dee376d64f54939b08dc5f41bfc7%7C5758544c573f47c
> ebee96c3e0806fb43%7C0%7C0%7C638489983214968539%7CUnknown%7CTWFpbGZsb3d
> 8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C
> 0%7C%7C%7C&sdata=bdsmSu1FEmeuDj6DNsClXkmGOwjVOMp4n5fcAew1Mjs%3D&reserv
> ed=0



___
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] Big Tree Tech Re: Mesa Card Stepgens?

[Peter Wallace]

On Thu, 18 Apr 2024, Nicklas SB Karlsson wrote:


Date: Thu, 18 Apr 2024 13:33:20 +0200
From: Nicklas SB Karlsson 
Reply-To: "Enhanced Machine Controller (EMC)"

To: emc-users@lists.sourceforge.net
Subject: Re: [Emc-users] Big Tree Tech Re: Mesa Card Stepgens?


...


If it can revert back to the way LinuxCNC works ( which requires realtime),
there is no real advantage to queueing at all. Queueing is used to make
up for the lack of realtime. For example Mach3 can do decent spindle
synchronized motion, but only if this feature is implemented in the interface
hardware. This would be a big step backwards for LinuxCNC.


And then it will come back to the problem how to send enough queued data to 
actuators.

Queue may make sense if for example sending position commands to follow over an 
Ethernet network as position is small
compared to the header. In that case sending many positions in one message 
reduce overhead.


Nicklas Karlsson



The equivalent of sending many positions is already done to some extent with 
interfaces that interpolate (or are velocity or velocity+acceleration 
controlled) This includes most stepgens, Ethercat servo drives etc


Queueing does make sense for some higher data rate applications like 
motion synchronized video data for engraving.






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



Peter Wallace
Mesa Electronics


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

[Emc-users] VCP Panel Options?

[Todd Zuercher via Emc-users]

It has been a few years since I last built a VCP panel for a machine, and well 
time marches on and things change.  What would be the considered the most main 
stream, best supported VCP panel system?

The machine I'm overhauling had been set up with a PyVCP panel along the Axis 
UI.  Later machines I did used GladeVCPs.  Since then there have been newer QT 
based things for building VCPs (I know next to nothing about them.)  My old 
PyVCP was rather complex, it had check buttons to enable/disable each of the 8 
router spindles and displays to indicate each spindles run status feed back 
from the modbus control (run, stop, accel, decel, fault,...)  as well as a 
number of convenience features (cycle time display, jogging of work coord. 
Vacuum clamping...)  Most of these I could easily reproduce in a GladeVCP, 
except I'm not sure how to do the Spindle status displays.  In the PyVCP I used 
the image_u32 widget to display an appropriate image/lable for each possible 
status feedback from all of the VFDs.

There had been a lot of fuss with tools and depency depreciation for Glade 
about the last time I tried to use it  But QtVCP at that time was still not 
quite ready for simple use (needed more programming skills than I have.)

Is there or what might be the simplest way to achieve the same or equivalent 
display features I had before with Glade or QT VCP?  There are other changes I 
need to make to the VCP necessitated by changes made to the machine, so I need 
to re do it.  The old PyVCP is a bit of a bear to work with, expecially when 
things get complex.  So just weighing my options before I jump in and try to 
learn or relearn what I need to get this done.

Todd Zuercher
P. Graham Dunn Inc.
630 Henry Street
Dalton, Ohio 44618
Phone:  (330)828-2105ext. 2031


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