Position loss UC-300ETH

If you think you've found a bug post it here.

Re: Position loss UC-300ETH

Postby cncdrive » Sat Sep 09, 2017 7:24 am

Hi Derek,

I would like to see the graph to cross out the possibility of a communication problem.
We made the communication latency view to see how much time do the communication cycles take, so the user can see if there is a problem with the communication.
How it works is the PC side software sends a packet and receives the response and it measures the time between the two.
For problemless operation it is a requirement to receive the answers within 20msec.
If the response times are longer means somekind of issue with the communication.

With USB controllers it can be that the USB hub is overloaded, e.g. some other USB devices connected to the same hub using too much bandwidth and so the UCxxx controller do not have enough bandwidth to do it's own communication fast enough.
With ethernet controllers it can be a problem with the network card or with it's drivers or it can be a problem with the router if a LAN connection was made.

I want to cross out the possibility of a communication problem, because if it is a comm. latency problem then the issue is outside the UCCNC and then you will have to look at the network components instead of debugging the UCCNC. So, it would be good to verify this first.
cncdrive
Site Admin
 
Posts: 4695
Joined: Tue Aug 12, 2014 11:17 pm

Re: Position loss UC-300ETH

Postby Derek » Sat Sep 09, 2017 11:14 am

Here is a pic showing idle, a big spike when I push run, and then running the code. This is with the ETH version.

Image
Derek
 
Posts: 341
Joined: Mon Sep 05, 2016 9:57 am

Re: Position loss UC-300ETH

Postby cncdrive » Sat Sep 09, 2017 4:37 pm

Hi Derek,

That spike is still OK, because it is not reaching 20msec, actually it is far from reaching it, so it is OK.
I will ask my collegue next week start to run some tests with a logic analyser using your profile.

BTW, is this issue only happens on the Y axis or on all of your axes?
I mean it does not matter if you run:

G00 Y-10
G00 Y0

or

G00 X-10
G00 X0

Both having the same issue?
cncdrive
Site Admin
 
Posts: 4695
Joined: Tue Aug 12, 2014 11:17 pm

Re: Position loss UC-300ETH

Postby Derek » Mon Sep 11, 2017 1:26 pm

I have loss on x as well but no data.

House made it through the storm fine. Under generator power. Won't know about the shop for another day because it's still very windy.
Derek
 
Posts: 341
Joined: Mon Sep 05, 2016 9:57 am

Re: Position loss UC-300ETH

Postby cncdrive » Mon Sep 11, 2017 1:33 pm

Hi Derek,

Today was a not too easy day for us. :(
My collegue started to test with your profile and the first thing he did was he connected the wrong powersupply connector to the UC300ETH giving it 12Volts instead of 5Volts which the UC300ETH however survived, but the step/dir counter did not. The laptop also got the high voltage through the step/dir counter's USB connection, but fortunately it did not damage.
Next we realised that we can't reproduce the step/dir counter electronics today, only tomorrow...
The other bad thing is the logic analyser did not want to start up, it developed somekind of problem which did not seemed to be an easy fix, so we could also not use it now...

So, finally we used an industrial frequency counter instrument as we had no better tools available and counted the step/dir signals.
And it seems to be that there is a problem, the step/dir counts does not match at the end when using your profile file.
What we could figure out so far is that it seems to only happen on 25, 50 and 100kHz and with pretty high steps per value which you have in your profile and we normally do not test-run with this high steps per values is how it is possible that we did not see this issue before. We do not see the issue clear enough yet to know why is it happening, will need to restore the measuring tools first.
We also tried to switch to 200kHz and 400kHz and with those the issue does not happen, the position is always correct.
So, this is what we could figure out today, will further investigate the issue tomorrow to see why is this issue exactly happening.

We also found another problem while debugging with your profile and it is with the B axis, that it reads the comp.acceleration value from the wrong register and so the B axis makes strange movements with your profile, because with your settings that other register contains a far value from what should be there and the comp acceleration sets the absolute limit for the acceleration, so the wrong value causes a problem even if backlash is not enabled.

For now I ask you to please try to switch the kernel frequency ot 200kHz and it should work ok with that, except the B axis, we will have to fix that.
What you could do is to temporarily, until we fix the problem, configure the B axis to the C axis, that does not have this issue.
cncdrive
Site Admin
 
Posts: 4695
Joined: Tue Aug 12, 2014 11:17 pm

Re: Position loss UC-300ETH

Postby cncdrive » Mon Sep 11, 2017 1:34 pm

Hi Derek,

Great, good to hear that you are safe.
cncdrive
Site Admin
 
Posts: 4695
Joined: Tue Aug 12, 2014 11:17 pm

Re: Position loss UC-300ETH

Postby Derek » Mon Sep 11, 2017 1:53 pm

Hi Balazs.

The b axis is for the power drawbar. That weird motion started a few versions back.

It may be weeks before I have power at the shop. Providing there is still a shop.

Derek
Derek
 
Posts: 341
Joined: Mon Sep 05, 2016 9:57 am

Re: Position loss UC-300ETH

Postby CT63 » Mon Sep 11, 2017 4:05 pm

Not to hijack the thread but could this also be happening in the 400ETH as well? I've noticed intermittent step loss on my Y axis but have attributed it to random noise issues. I'm running at 25khz because I only need 13khz max at the steppers. Using a Gecko 540 that will take 200khz.
CT63
 
Posts: 115
Joined: Sat Sep 03, 2016 11:19 pm
Location: Connecticut, USA

Re: Position loss UC-300ETH

Postby cncdrive » Mon Sep 11, 2017 9:18 pm

We don't know the details yet, because as mentioned in my previous post caused by the problems with the measuring tools we had today we could not investigate the issue through yet, will probably need a few days to do it and then we will know what the problem exactly is. What we see now is that when the steps per parameter is high, around 10000 value then is when this issue happens and with 25, 50 and 100kHz kernel frequencies and we only tested this today with a UC300ETH-5LPT. We test every releases for possible step loss, just not with this high steps per values, so we did not explore this issue before.
cncdrive
Site Admin
 
Posts: 4695
Joined: Tue Aug 12, 2014 11:17 pm

Re: Position loss UC-300ETH

Postby CT63 » Tue Sep 12, 2017 1:51 am

cncdrive wrote:What we see now is that when the steps per parameter is high, around 10000 value then is when this issue happens and with 25, 50 and 100kHz kernel frequencies and we only tested this today with a UC300ETH-5LPT. We test every releases for possible step loss, just not with this high steps per values, so we did not explore this issue before.


The 10,000 steps per unit is an interesting find. I'm using 10,176 steps per unit @ 25kHz. My problem is slightly different in losing Y axis steps intermittently. I first noticed it running at 100kHz then dropped to 25kHz. It's still happening but much less frequently.

Before changing over to 400ETH & UCCNC I was using a Ethernet SmoothStepper/Mach 3 and never had a noticeable step loss problem. Had a ton of CV and other problems with Mach 3.... :evil:
CT63
 
Posts: 115
Joined: Sat Sep 03, 2016 11:19 pm
Location: Connecticut, USA

PreviousNext

Return to Report a bug

Who is online

Users browsing this forum: No registered users and 4 guests