is the duet 3 board able to kill an attached raspberry?

spllg · 19 Jan 2020, 14:49

configuration: duet 3 - 5v, controlled by dwc running on a raspberry pi 4 model b, powered via usb from the controlling raspberry (all jumpers: internal 5v, 5v->sbc and sbc->5v removed.

this configuration has been working fine for ~3 weeks but today at the end of a filament change, when i wanted to extrude some filament to get a clean color, the extrude spinner did not stop to spin and the raspberry did not even respond to a ping request. looking at the duet i recognized that none of the leds was on. i tired to power cycle the machine without success. i removed all connectors except power from the raspberry, tied to boot it but without any success - the raspberry does not even respond to a ping request and it's display stays dark.

i have been using rpis for several years now but never had killed a rpi.

question: is it possible that the rpi was killed by some sort of voltage from the duet 3? if yes, how to prevent it?

Danal · 19 Jan 2020, 20:06

Did the Duet power lights come back on when everything (except 12 or 24 V VIN) was disconnected?

If not, that is if both the Duet and the RPi are dead, this sounds like somehow 12 or 24V got on a 5V or 3.3V rail. This can and does happen via shorting in the moving cables on a hot end... look for that very carefully.

A Former User · 19 Jan 2020, 21:38

@spllg said in is the duet 3 board able to kill an attached raspberry?:

if yes, how to prevent it?

Look up digital isolators, something like TXS0108 might fit the bill. Or maybe the more traditional series resitors and TVS diodes could work, not sure if its a good combination with the speed the SPI bus is working at.

littlehobbyshop · 19 Jan 2020, 21:42

It couldn't be something as simple as the Pi SD card dying could it? Done that before.

Danal · 20 Jan 2020, 00:16

@littlehobbyshop said in is the duet 3 board able to kill an attached raspberry?:

It couldn't be something as simple as the Pi SD card dying could it? Done that before.

Would not affect power LEDs.

spllg · 20 Jan 2020, 06:32

@Danal do you you mean: shorting heater -> termistor? going to check tonight.

spllg · 20 Jan 2020, 06:34

@littlehobbyshop i do not think so, because mounting it from my laptop (after filesystem repair) does work fine.

Danal · 20 Jan 2020, 16:30

@spllg said in is the duet 3 board able to kill an attached raspberry?:

@Danal do you you mean: shorting heater -> termistor? going to check tonight.

Yes, that is one possible scenario. That would put 12 or 24 on the 5.

dc42 · 20 Jan 2020, 17:41

All generation 2 Duets survive shorting thermistor inputs to Vin, except some Duet WiFi v1.0 boards and the white Duet WiFi prototypes. See https://www.youtube.com/watch?v=uyWolKFzb-A.

Duet 3 boards have these and additional protection measures, which I hope to demonstrate in another video soon.

spllg · 20 Jan 2020, 21:56

@Danal visually checked and measured the resistance between 12v/24v heater cables and cables to all other components at the print head and could not find any shortage.

A Former User · 20 Jan 2020, 22:27

@spllg said in is the duet 3 board able to kill an attached raspberry?:

visually checked and measured the resistance between

can be tricky to locate a fault in a system that moves and heats up though. the meter also use much lower voltage to test, so if you have a leakage and not a straight up short it won't find it.

I should get some new terminals and housings one of these days, so I could try adding an isolator to my setup, but admittedly for the Pi to risk any damage odds are you're also risking damaging the Duet3, so imho it makes more sense to focus on protecting the Duet3 in the first place.

edit: just finished planning out the txs0108e, if anyone decides to try DO NOT use 5v to power the isolator on either the SBC or Duet3 side, both sides need 3.3v and you'll need to find 3.3v somewhere else on the duet (IO port, temp daughter board, swd port etc)
edit2: brain failure. txs does not isolate properly. will try some adum-something-something anther day

spllg · 21 Jan 2020, 16:35

just reconfigured (set jumpers 'internal 5v enable' and '5v->sbc') and powered the duet from my 12v psu.
the duet seems to be working - at least the 4 leds left of the reset button are all glowing and the red led right of the reset button is blinking. because the rpi appears to be dead i cannot proceed to do further tests.

jay_s_uk · 21 Jan 2020, 16:36

@spllg why not try it in standalone mode to confirm its fully functional?

spllg · 21 Jan 2020, 16:41

@jay_s_uk this would be for academic purpose only because i am already waiting for a duet 3 replacement (other issue).

spllg · 22 Jan 2020, 06:34

nevertheless it would be nice to know the reason for rpi's death.

was it faulty before?
did it commit suicide?
was it killed by an external event?
was it killed by the duet?

spllg · 22 Jan 2020, 06:38

@dc42 <All generation 2 Duets survive shorting thermistor inputs to Vin, except some Duet WiFi v1.0 boards and the white Duet WiFi prototypes. See https://www.youtube.com/watch?v=uyWolKFzb-A.

Duet 3 boards have these and additional protection measures, which I hope to demonstrate in another video soon.<

did you do the same tests using a duet 3 with an attached rpi? did the rpi survive?