Uccnc not staying within acceleration parameters?

[dezsoe]

It was a closed loop stepper, but I was there to check the movements and the sound. It worked fine, but both controllers were ethernet type. I'll test it tomorrow with a UC100 and the machine with UC100 has classic steppers.

[rmtucker]

Further Tests today:-
Only Z axis has power and only Z axis enabled in Config.
Ran program at 800mm accel and no lost steps,just to double check that this was not a torque issue usually set at 250mm .It was running so fast it made the machine shake
So i think next step is to borrow a friends uc100 and try that.
There are two license files in the uccnc folder.
Uccnclicense.txt
Uccnclicense_9021E4.txt
Which one do i replace when i borrow his uc100.
Or do i just drop his license in there aswell.?

[rmtucker]

Further Tests today:-
Only Z axis has power and only Z axis enabled in Config.
Ran the attached shorter program 3 times at accel 800mm (only takes a couple of mins to run).
No lost steps.
Enabled the x and y in config as well as Z axis.
But still only power to The z drive.
Ran the program 3 times and it was showing lost steps.

[rmtucker]

It was a closed loop stepper, but I was there to check the movements and the sound. It worked fine, but both controllers were ethernet type. I'll test it tomorrow with a UC100 and the machine with UC100 has classic steppers.

Did you manage to test this on the other machine,even with the smaller program i posted?.
I have had to stop any more testing and try and decide what to do next!

[dezsoe]

Yes, here are the results, I've just finished. (Sorry, on the weekend I could not get to the machine. I'm married... )

So, the results. I used my oldest machine which is made of wood, and all the drivers are home made: this machine doesn't tolerate any bad signals.

First test with UC100 and your smaller g-code (when I saw the results I rehomed the machine to check again):

I was curious: is it a general problem or only UC100 related, so I went with the same config but replaced the UC100 with UC400ETH:

You can ignore these few steps, this machine is not precise. My conclusion is that the UC100 misses some steps.

I'll try to connect a UC100 to a newer machine (with closed loop steppers) to see if it fails. It will tell us important information to find the problem.

Thanks for your patience.

[rmtucker]

Thank you so much.
I thought i was going mad.
Try running the code with x and y disabled in the config and see what happens.
I find it never loses steps at all,so maybe that is relavent.

[dezsoe]

Getting really strange... Read the picture from the bottom (it's the status box):
- disabled X: Z error is the same as before (the 2 step difference is in the machine)
- rehomed, nothing interesting
- disabled Y: no Z error (ignoring the machines inaccuracy), X has a few, but I think it's also inaccuracy

There are 64 moves in Y, and about 72 to 76 steps error if Y is enabled. Hmm.

UC100test2.png (35.29 KiB) Viewed 9707 times

I couldn't connect UC100 to the closed loop steppers, I have to make some PCB (or cable) to interface them, but now I'm getting very curious, so I'll make it.

[rmtucker]

I have ran so many tests i can not remember them all!
But i am pretty certain that if both x and y disabled there are no lost steps.
I am also 70% certain that if i have the x and z enabled that it loses steps
This seems to contradict your findings.
I was married till i started with this problem

[rmtucker]

@Dezsoe
It seems your initial findings have trashed my idea of borrowing a friends UC100.
So i am not sure how to continue from here.
Please let me know if you have any other test findings too share.
I know you are working hard to investigate the problem,But i am not sure if i should go back to my old parallel/pc setup so at least i can use the machine till this is resolved.
I can not understand for the life of me how no one has highlighted this problem before now.
It makes me wonder if i should have gone with the UC400eth from the start but unfortunately Balazs dangled the free license with UC100 carrot in front of me a couple of months ago .

[dezsoe]

We did a lot more tests yesterday to find the problem. We are getting closer and closer, but give us some time to find it. I'll be back when I have any news, but I can't tell you when. It can be days or weeks: I can't prognose. If you have urgent job and you can switch back to do your job then do it.