Advanced Search

the command return to me :

Tinine wrote:

Yeah I understand this but I want to use my own controllers and my own PID, etc. 

That is all good as explained (i tried) above.

Tinine wrote:

I don't want to rely on an OS for the motion control. Is this even an option with LinuxCNC?

This makes no sense as LinuxCNC is a Motion Controller that runs inside an OS, Linux in this case.
I think i might be on the right track as to what you are looking for, but first some stuff to get out of the way
-you have drives with PID loop in them, they are not motion controllers
-motion controller in this context implies opening gcode, parsing it, planing the trajectory for the axis/joints, sending understandable signals to drives/outside world.
-
You can use separate parts of LinuxCNC on their own, mostly, so you can probably use the gcode parser and trajectory planner, but you would also need the sending signals to drives part, so......
any closer?

Yeah I understand this but I want to use my own controllers and my own PID, etc. 

I don't want to rely on an OS for the motion control. Is this even an option with LinuxCNC?

So looks like the DKMS driver (I think the default driver with latency issues "8169")

So not sure whats going on.

What does your kernel command line look like?

cat /proc/cmdline



 

I played with PID myself some, and moving away from P250 I0 D0 just made things worse for me, the deadband was what I needed, but I was also hunting for a 4rth digit of following accuracy. Glad things are working better!

What kind of set up do you have? Do you have Stepper Motors with Feedback, without feedback or Servo Motors?
I am running Steppers without Feedback. I wonder how the PID works in that case. If you don't have feedback, how can the PID work. My theory, the Position is commanded and the Steppgenerator reports the "theoretical" position back. Not knowing it the motor made the step or not.

When I played around with it, I lowered the values quite a bit and got a following error. Seems like to confirm my theory above. I did not notice a change from the default PID P 300 to like 30, but the following error came with P 5 or less.

I was hoping that the way my motors stall would be an overrun of the P Max value, but increasing that did not help either. I have to do more testing. It still might not be a problem of RIO, since my other motors are running just fine.

Thanks again.

Should not affect work/use, LinuxCNC is very good at reporting issues, so if LinuxCNC is not complaining, move on, all is good.

Thanks a lot, I'll check it out. Do you think running with this error message is risky?

That would be analog control, like old servo drives or velocity control drives, so yes you can control them with LinuxCNC and Mesa boards, see below, and would also require encoder feedback to LinuxCNC.
Mesa combos that can do +-10V :
7i97T all in one
7i92TM with 7i77
7i92TM with 7i85 and 7i83 and 7i84 for IO if needed
7i96S with 7i85 and 7i83.

I played with PID myself some, and moving away from P250 I0 D0 just made things worse for me, the deadband was what I needed, but I was also hunting for a 4rth digit of following accuracy. Glad things are working better!

That reminds me, is there a way to configure minimum delay time between a DIR change and a movement command? some steppers are more picky about that. I also need to flip the action of the EN pin, my steppers are enable-low and it's currently outputting high when enabled.

Dell Optiplex 760 (very old Core2Duo) and 980 (older i5) have parallel ports in Tower cases, both work good with LinuxCNC for software stepping and 980 is good for Mesa.
HP Z210, Z220, Z230 should all have a parallel port header and usually come with the adapter/port in, also very good for LinuxCNC and built like tanks.
Some Fujitsu i5 and i7 are also very good, but no idea what models they vere, and i have a Fujitsu laptop that is very good for Mesa, unfortunately gave it to a client and got it back with a broken screen.
As for laptops, i have bought several older Lenovo laptops and strangely most are ok for Mesa use, example X220 and T420S were unusable, X240 seems OK from some short testing! A T470 i have in use for over 6 months on a production machine, works fine with Mesa 7i92.
There are several topics here on the forum with plenty of used PC/laptops tested with LinuxCNC for Mesa and parallel port, but as with everything, yout milage may vary, as i had two of the same Acer laptops where one was very good and the other was terrible for latency, the only difference was CPU i think.

The motor for rotating axis?
It had a long Nema 23 with 1:15 reduction, but had an accident while being transported from the old shop to the new shop and broke the reduction, so i changed it to chain reduction of 1:7 and Nema 34 motor. Also had to reduce acceleration a bit as it was cutting 193mm diameter tube with 4mm wall and 3 meter length.
Use very low inductance motor for this as they can spin faster, you might need it for smaller diameters.

I have created closed-loop motion controllers using the RPi Pico, each one reading 4 quadrature encoders (2.8M quad-counts/sec)
I am using a separate Pico for each axis though. +/- 10V motor command, 12-bit DAC.
I guess I could simply take pulse & direction from LinuxCNC but I would like to not have to worry about LC's latency and jitter issues.

Craig

What do you mean by Motion Controller?
Using 2 of those would be reduntant and unnecessary.
Do you mean closed loop in the drive or in LinuxCNC? Both can be used, and LinuxCNC has PID controll and settings for everything required. Also can use PID closed in Drives and in LinuxCNC, and much more.

Find a used PC with a parallel port, install LinuxCNC from the downloads page, play with it, it can run with no machine in Sim mode, it can controll stepper drives directly from parallel port just mind the wiring, attach some LEds to parallel port so you can see what the pins are doing when you press buttons, etc, etc.