4w ago

Duet3D fire hazard. Software issue? (manufacturer doesn't care)

I run half a dozen Duet but one of them started to give me a lot of headaches and only this one (unique hardware + configuration so no other to compare to).

issue: tool head 3 heater is always enabled. The software detects a thermal runaway but the heater stays enabled. The only option to stop the machine from catching fire is to remove the main power.

Hardware:

Duet 2 wifi
Due5X

troubleshooting done (hardware):

measuring the mosfets => good
checking the driver for obvious issues with a multimeter => no issue found
moving toolheads hotends around. Issue persists on #3 spot (4th toolhead) so it is unrelated to the tool head. disconnecting it entirely also causes the heater to stay on
removing the SD-card => issue is gone.
sometimes it works like it should. Most of the time it doesn't.
on this particular unit there were some other minor anomalies regarding the WIFI earlier (wouldn't connect to the network) but those fixed themselves after a few rounds of resetting the wifi module, adding it to the wifi network again, waiting a few days for it fail once more.

**TL;DR Is my assumption correct that this indicated an issue with the Firmware from Duet which isn't just on this device but also on the newer Duet 3?

Did anybody has run into a similar issue and was able to resolve it?

In the wider picture: Is every duet product unsafe?** This is an older Duet 2 but it runs the same/similar firmware as the current generation Duet 3. So one of the big questions is if all of the Duet products are unsafe. Would be bad as this has been my controller of choice meaning a lot of work to replace all of them.

Btw. If you buy a duet: don't expect ANY support from the manufacturer.

Here is the main config file:

config-overwrite:

5 comments

Is the FET gate floating? Maybe try adding a pull down/up resistor as a sanity check. I have no Duet experience. Based on what you are saying, I would troubleshoot that hardware first. Cut the trace and see if the gate signal is going off as expected... things like that.
- here is the schematic: https://github.com/Duet3D/Duet-2-Hardware/blob/master/Duet2/Duet2v1.06/Duet2_Schematic_v1.06.pdf
  https://github.com/Duet3D/Duet-2-Hardware/blob/master/Duet_Expansion/Duex5v0.9a/Duex_Schematic_v0.9a.pdf
  There is a 74HCT02 which can do up to 20mA so no pulldown resistor is required.
  Not looking forward to cut traces on it for troubleshooting. It is an original board (no clone) so it should be exactly the parts in this schematic assuming the linked revision matches the board revision.
  I would troubleshoot that hardware first
  Was my first instinct as well as I didn't do any firmware updates prior.
  The continuity test doesn't show a short and the test Duet recommended in one of the documentation was to remove the SD-card. If the issue stays it is a hardware issue. If it is "gone" it isn't and me removing the SD-card causes this behavior to stop further indicating a software issue. Reading the schematic right now for the first time and I am schoked that there is no ESD-protection at all. Nothing. Just a straight connection from the mosfet to the output... what the hell where they thinking ....
  Would need to have an in-depth look at the schematic but likely I can't probe the MOSFET gate voltages in the circuit.
  What I could do is probe the signal/heater enable signal from the MCU.
  
  Removing the SD card shows whether there is a burned or shorted connection somewhere, but does not show if a gate is staying high after it is triggered. HCT logic is translating CMOS threshold voltages on the input to TTL levels at the output. Still while the window of where bit logic high and low are specified, the actual logic zero voltage is relatively high. You also have to remember that all chip manufacturers have an error rate distribution curve and send bad parts or parts that are right on the edge of acceptable. It is quite possible that you happened to get a FET with a low gate threshold voltage and a logic gate with abnormally high logic zero threshold. I would just tack on a through hole resistor between gate and source similar to how Fan 2 MOSFET control is configured and see if that solves the issue.
  I'm not effectively able to help with the software, so this may still be a red herring, but I do not think removing the SD card effectively addresses the question of whether the gate is driven low correctly.