6HC+2x 3HC+SBC. 3.2 Crashes at high print speed.

Phaedrux

I've flagged this for review. Will just need to wait for time zones before anyone can take a look.

jrockland

@Phaedrux
Im having that fault every 10 min now while trying normal print speed..:
HardFault undefInstr, Platform spinning, available RAM 145856, slot 2
since I updated the boards.. Ill just shut it down for now.

no worry I know we are not in the same time zones. thank for the help

Phaedrux

If you disconnect that tool board entirely does the problem go with it?

jrockland

@Phaedrux
seem like it.
as per my config.g file do you think I can move X-Y to the main board and the 4 Z to the extension boards ? (older version wouldn't accept my Z sensor on extension boards, and also the sensor will not be on the same board as all the Z (only 3 ports, 4 motors)..

But that would solve the communication issue as Z barely receive anything compare to x and y..

dc42

@jrockland, there have been fixes to CAN communication at high movement rates in 3.3beta. But I don't think any of them accounts for the reset you are seeing. I will analyse the reset data.

dc42

I've looked at the reset data. It returned from a function call, but then complained that the instruction following the call wasn't valid - whereas it certainly is in the firmware. So either the flash memory or the cache memory has returned the wrong data. The most likely cause I can think of for this happening is a power brownout.

jrockland

@dc42 so we are looking at the boards power controller failing at high rate ? as for the power supply they are stable +/- 0.15v during the whole process, and under 30% load.

what can we do to stabilize the power distribution on the board ?

jrockland

@dc42 my friend is bringing over is marlin boards, as we still want to see stability vs speed of the machine. we will be using the same power supply for comparison.

Also I now need to downgrade the boards as since I updated they are not useable anymore..

jrockland

@dc42 might actually be power distribution to the sbc.. I know they are unstable and I wouldn't trust a raspberry pi to be anywhere near a functional tool.
I will look into it.

jrockland

@jrockland
negative, same problem with the sbc self powered.

jrockland

@dc42 power distribution is stable: 120.4 at source, 24.1 at the power supply tested while printing and when it failed again. going back to version 3.1 as I dont remember having that issue then.

Phaedrux

@jrockland said in 6HC+2x 3HC+SBC. 3.2 Crashes at high print speed.:

--Why 2 power supply : at first I tough I was running out of amps when everything was kicking together.. actually helped when I added the second one.

Is it possible the first power supply was having issues browning out?

jrockland

@Phaedrux I tested both while running and they are stable and way underload.
I added a second power supply because of that exact issue where I taught I was running out of power.

We also tried with my friend marlin's boards (using the same power supply) and we have no issue. I can push it twice as fast without any power loss.

I will be flashing those board with klipper and see if it is firmware related, if it is the board I guess I will have to learn how to program marlin and go with their board.

jrockland

@Phaedrux when we pulled out the marlin board we realised the main difference was the grounding; the exp boards dont have any external grounding connection.

I added grounding using io0 gnd on each ext board and it fixed the issue of the main board shutting down.

I still have issue with the ext board loosing connection for a sec under high load (not high power, just when they receive lots of commands: If I do a giant square it work good but a giant circle make them fail).
But I think moving Z motors to the ext board and x/y back to the main should solve that.

dc42

@jrockland said in 6HC+2x 3HC+SBC. 3.2 Crashes at high print speed.:

@Phaedrux when we pulled out the marlin board we realised the main difference was the grounding; the exp boards dont have any external grounding connection.

I added grounding using io0 gnd on each ext board and it fixed the issue of the main board shutting down.

I still have issue with the ext board loosing connection for a sec under high load (not high power, just when they receive lots of commands: If I do a giant square it work good but a giant circle make them fail).
But I think moving Z motors to the ext board and x/y back to the main should solve that.

It's not normally necessary to add extra grounding to expansion boards if they are powered from the same PSU as the main board. However, we have found one situation in which CAN connection can be disrupted. In summary:

If you use a high powered bed heater powered directly from the Duer 3 MB6HC main board, then either:

Use very short and thick wires between then Duet 3 MB6HC and the PSU; or
Connect the PSU to the Duet 3 MB6HC, then daisy-chain the expansion board power wires from the MB6HC power connections, instead of running them direct from the PSU.

jrockland

@dc42 O im still far from the heated bed. Still have to get those 3hc to work properly.

jrockland

@dc42 I did manage to get the 4 z axis to work with an inductive sensor on 2 different 3HC boards, wouldn't work before the last update. Good job with that !

I used one of the fan output (out6) as power for the sensor (running 12v.. still have to figure the hz but it work anyway) instead of the 12v output on the 6HC, and used io1 instead of io7 from the 6HC.
Configured it as a normal endstop as it wouldnt read it as a z-probe (no more mesh bed leveling.. but I wasnt a fan anyway because of the constant Z movement causing issue because of the size of my machine..)

I was not sure about sending 12v to that io port but it was already like that on the 6hc as described in your setup manual.. so I guess it is fine. My inductive sensor probably just send about 3v out anyway.

dc42

@jrockland said in 6HC+2x 3HC+SBC. 3.2 Crashes at high print speed.:

Configured it as a normal endstop as it wouldnt read it as a z-probe

It should have worked if configured as a Z probe, type 8.

jrockland

@dc42 I will give a try, but as it work now I wont touch it (the board isn't the part im testing and I have to finish testing my latest gantry/extruders..)
I also ordered 2x (actually 5 but 2 for that printer) 1500w 24v power supply, 1 for the boards and 1 for the beds when they will be added. that should clear any chance that power was an issue.

jrockland

@dc42 I managed to get it up to 250mm/s before having a memory crash error ( I should had written it down..) but moving 4 stepper + extruding + sending code to the 2x3hc boards at that speed will create issue for sure... and I was also doing a shape, not a square so the file was heavy.. poor board he is getting worked hard. nothing blew up tho !