Duet sometimes really slow? - I2C error or?

dc42

@deckingman said in Duet sometimes really slow? - I2C error or?:

@dc42 David,

For info, I ran "the sequence" again today after an over night shutdown, with @wilriker 's firmware 2.03RC3-M574C (2019-05-28b1) (based on your firmware but with end stop mapping enabled). No problems encountered with movement, no I2C errors reported, and no I2C resets.

Looking good so far........

Ian, thanks for the feedback. To be honest, I think you were just lucky that you had no I2C errors on this occasion. However, given the reports of behaviour when you and others tested the 2.03 releases, I am confident that the changes I made to the I2C driver have completely or at least partially solved the original issue. These changes (as well as all the other improvements in the 2.03RC) releases are already in the 3.0beta source code.

deckingman

@dc42 OK. I'll keep testing but only report back if something amiss happens. So no news will be good news.

Errrrr, just had a thought. Since last time when I could provoke errors consistently, I've changed my motor mounts. So when you said, I was just lucky I started to wonder....

These new mounts are aluminium and the old ones were plastic. So the "XYUVWA" motors will now be earthed through the mount and frame whereas before, they were insulated from the frame by the plastic mounts (which was one of the reasons fort changing them). Might it be possible that the reason why I didn't see I2C errors is not luck but more to do with the fact that I've earthed some of the steppers?

I'm not doubting that the firmware changes have fixed the issues but wondering if the root cause was stepper noise, which can be mitigated by earthing the steppers. Thoughts?

I guess it would be interesting to hear from other users if their stepper motors are earthed or not.

deckingman

@dc42

Ignore comments above above about my changing the motor mounts. Having run "the sequence" again today, subsequent M122 report shows:

I2C nak errors 0, send timeouts 1, receive timeouts 0, finishTimeouts 1, resets 1

My take on that is I2C errors occurred but the firmware caught and fixed any problem. So earthing the motors through the mounts didn't affect behaviour.

DocTrucker

@deckingman talking of earths have you got the earth spade near the ethernet connection grounded? Doubt this will effect your issue bit worth doing.

dc42

@deckingman said in Duet sometimes really slow? - I2C error or?:

I'm not doubting that the firmware changes have fixed the issues but wondering if the root cause was stepper noise, which can be mitigated by earthing the steppers. Thoughts?

Un-earthed stepper motors driving rubber belts can build up static charge (think Van de Graaff generator). If the stepper motors are well-insulated from the printer frame via plastic mounts, then the shortest path for the static charge to arc to ground may be to the stepper motor wires. That will in turn cause a ground transient on the Duet or the Duex (whichever one the stepper motor is connected to), and this will affect any I2C transaction that is in progress. So yes, using metal stepper motor mounts may have fixed the root cause.

But I see from your more recent post that you have had another I2C error.

deckingman

@doctrucker said in Duet sometimes really slow? - I2C error or?:

@deckingman talking of earths have you got the earth spade near the ethernet connection grounded? Doubt this will effect your issue bit worth doing.

What earth spade? Mine is a pre-production or at least the very first production Ethernet boards so it probably doesn't have one.

deckingman

@dc42 Yes I'm aware of the Van de Graff issues with belts - can cause all sorts of problems with model helicopters I believe because the tail rotor is often driven by a continuous belt and the resultant static plays havoc with the receiver. Also anti-static belts are often used in conveyor systems.

But I wonder if it's an issue with printers because generally the belts constantly change direction back and forth rather than always rotate in the same direction. It's an awfully long time since since schoolboy physics but if one reverses the direction of a Van de Graff generator, doesn't that reverse the polarity of the charge?