Sep 18, 2019, 16:26 by [email protected]:

> Thank you Justin and Bas for quick response.
>
> I know about Mesa cards, but it takes too long to have it in my country. 
> Legal procedures etc. or I need to spend a lot for quick cargo and custom 
> taxes.. Besides I have two analog and one digital servo motor and drive thus 
> I need at least two Mesa daughter card and one anything card like the one for 
> pi or pcie etc. Which is very confusing for me.  So I am trying to find a 
> solution for that. I am good at python programming and I know that using NML 
> with machinekit is possible to control motors. I already have a pi3b+ and a 
> pc :) so I just wanted to know if it's a good idea to spend some time on it 
> or not.
>
> I have no idea about pic programming nor electronics so I wanted to do 
> something with what I already have.
>
> About step frequency I have tested linuxcnc with parallel port and I have got 
> 1500 mm/min speed rate if just two axis runs at the same time. I do relief 
> machining a lot and when z axis runs with X and y, speed drastically reduce 
> to 600 mm/min. This is because of the latency issues. Not to have latency 
> issues I thought it would be good idea if I seperate the GUI and the step 
> generator.
>
> Since forums is here for brainstorming I just got the idea and shared with 
> you to learn if it's a good idea or not.
>
> Thank you again for your interest.
>
> Berkdan
>
Hello,
I can feel your pain. The distribution part of Mesa business is very bad. It is 
a little bit better that it was but still they could learn a lot from Chinese 
sellers.

However, Raspberry Pi with default GPIO configuration for step-gen is not good 
solution. I have heard some interesting news about the GPIO toggling on 
Raspberry Pi: https://evlproject.org/pipermail/evl/2019-May/000019.html 
<https://evlproject.org/pipermail/evl/2019-May/000019.html> but fo far I didn't 
play with it.

About distributed system, to have proper distinction between RT tasks specific 
hardware and non-RT tasks specific hardware is the way forward, I think. It is 
clutch from LinuxCNC days to have everything on one PC, from times when 
electronics, PCs and whatnots were very expensive. Today, there is no need to 
not have one physical system running the RT kernel and tasks and other taking 
care of the planning and another taking care of UI. Video frame buffer will 
always cause unnecessary latencies on RT systems.

However, Machinetalk in the current form is not going to take us there. 
Unfortunately.

If you want Raspberry Pi with something like Mesa FPGA firmware, try the 
BeagleBone Black.

And, if you are up to challenge, you can try to implement simple core on some 
$10 FPGA developer board from Alibaba. The HostMot2 is nice, but it is 
universal and so overcomplicated. Look for example at the Pluto code or 
internet tutorials. Or the LAN9252 + ARM or XMC4800 for Mesa card DYI 
alternatives.

Cern.

-- 
website: http://www.machinekit.io blog: http://blog.machinekit.io github: 
https://github.com/machinekit
--- 
You received this message because you are subscribed to the Google Groups 
"Machinekit" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/machinekit/Lp4RDSK--3-1%40tuta.io.

Reply via email to