Might have finally killed a DuetWifi

[DjDemonD]

So my first board which was one of the first 500 production boards, I think is dead. I haven't pulled it out of the machine to really dig into it, but during an upgrade to some hotend parts last night, I think I've shorted something. I left it last night with the duet board powered up (via ext USB) but the main power off. However this morning the paneldue was showing a black screen and the hotend was warm (presumably heating from the 5v supply).

Are these symptoms of any known failure mode? Might guide me in my investigation for which I have limited time.

[dc42]

I'm sorry to hear of your misfortune. The only way I know of killing both a Duet and an attached PanelDue is to short either +3.3V or +5V to +12V or +24V. This can happen if a head-mounted Z probe shorts to the hot end metalwork, and you get a short within the heater cartridge.

Is the PanelDue still blank if you disconnect it from the Duet and provide power throught the USB socket? Perhaps it's just pining for the fjords.

[DjDemonD]

Thanks David. I will if I get chance, do a little investigating. I am quite curious as having read/replied to so many posts here I've never heard anyone reporting a hotend heating, with no VCC power attached, although I suppose a failed mosfet (coupled with shorted heater cartridge) might connect +5v to GND and therefore warm-up. I couldn't read the temperature as the board was non-responsive, but it was only hand-hot so I'd guess around 50 deg C.

I have a lesser used machine that can donate its controller for now, if this one is not repairable.

[DjDemonD]

I accept this isn't very interesting but if anyone knows what these signs and symptoms mean I'd love to know about it.

So with usb power attached I am seeing 3V+ light (red) and a red light to the left of the USB connector (viewed from outside). No other lights.

The resistance across E0 heater output isn't stable its cycling from very low 10ohm all the way to 0.9MOhm.
I get 20 KOhm from ground to right leg of the E0 mosfet, but on the E1 mosfet its 0 ohm. So clearly the E0 mosfet is allowing some leakage of current.

VSSA fuse is not blown.

The paneldue is odd. The screen is fine, I moved it over to another machine and tested it. The paneldue board does nothing, when connected to the other machine, it does nothing when connected to a PC, even after pressing erase and reset, no "bossa program port" showing in device manager, so I presume I've killed it somehow.

[BPisLife]

@djdemond your 5V regulator could have bit the big one. Does the board have an option to use an external 5V instead of an internal one like v1.03 has? If so, try that. If not, then measure the output from the 5V regulator and see what the voltage is. Ohming a FET is not going to tell you all that much unless you know what state it is supposed to be in mainly because a "0 ohm" across the drain to source means it might just be "closed" or "on".

Try measuring resistance from gate to source with the board not powered on. That resistance should be high otherwise it would show a short between the gate and drain ie a blown FET. That is only one failure mode. I would start with the 5V regulator and work your way down the line.

[dc42]

If you did get a short that caused the +5V supply voltage to reach a much higher voltage, then the voltage regulator on the PanelDue has probably failed. The fact that the screen is still working suggests that in doing so it may not have fed excessive voltage to the remainder of the circuit, if it is the 4.3 or 5 inch screen.

[DjDemonD]

I will check the 5v Reg, but it won't work on ext 5v. That being said this machine had both the int 5v jumper and ext 5v jmper connected at the same time (I forgot to remove the int 5v jumper when I switched to using a relay module to control 240v to the machine and powered the board from a USB supply permanently) . Also the memory card (original 4gb one supplied) is dead got too hot to touch when I transfered it into another duetwifi board, so I quickly removed it, also got too hot to touch in my card reader on my PC. So another piece of the puzzle?

I am wondering if any progress has been made on a backup facility to (ideally) automatically backup the sd card to pc when connected via DWC? I only say this as my last manual backup of the sd card was 5 months ago and is almost certainly well out of date.

[DjDemonD]

So not having a great day. Put everything back together with another Duet board in it, and this time magic smoke came out of this chip:

(shown is the chip from another board not the damaged one).

What does this chip do?

The 2nd board I've just murdered was a chinese clone so not too upset, I think I am going to have to rebuild this machine part by part, to track down the fault.

[T3P3Tony]

@djdemond U11 is a 74HCT02 it drives the Heater mosfets by level shifting the PWM signals from 3.3V to 5V and also providing protection from the heaters coming on permanently if 3.3V was shorted to ground.

It looks like you have VIN being fed back into either 3.3V or 5V as both those go to U11:

[DjDemonD]

Thanks Tony I'm thinking hotend heater cartridge short but I'll investigate. Might see if someone can repair the board if there aren't any other issues.

[dc42]

Shorts between the hot end heater cartridge and it's casing appear to be fairly common. That's why we protect both sides of the thermistor inputs, because shorts between thermistors and the hot end metalwork are also common - maybe less so since E3D switched to cartridge thermistors.

To damage a Duet like that, you need a short between something powered from VIN and either +5V or +3.3V. One such mechanism is a short between the +3.3V supply to the Z probe and the hot end metalwork, coupled with a short inside the heater cartridge. That's why the fitting instructions for the IR sensor insist that 2 or more layers of Kapton tape are put on the back of the board, to insulate it from the heater block if the hot end rotates (as E3D hot ends do).

HTH David

[DjDemonD]

Thanks, I've been scratching my head, as having started to test components individually, the hot end heater cartridge is not shorted to the block, or to negative of the PSU, or to the PT100.

The PSU is outputting 24.1v.

I will continue to investigate, there is a possibility something entirely different is going on, a full strip down and rebuild is almost certainly required.

The thing is I'm not sure if I can be bothered, I've concluded reluctantly that this machine (cutting-edge Kossel XL from 2016) is no more accurate a 3D printer than a CR10 300 which costs around 1/3 of the cost, is basically ready-assembled and just works. Since I might be planning to do some travelling I might just dismantle the XL and sell it off in pieces, and continue to use my CR10 for day-to-day printing. Will give forum members first refusal on the parts.

[dc42]

If there is a short between the hot end heater cartridge and its casing, the short may only be present when the cartridge is hot.

There are of course other ways to get a short between +3.3V or +5V and VIN, such as mounting the Duet too close to a conductive plate behind it.

In principle, a faulty 5V regulator chip could cause a short between VIN and +5V. However, on the returned boards showing this type of symptom that we tested, I found that the 5V regulator was working correctly in all cases.

[DjDemonD]

So I finally found the time to do a full diagnostic of the machine and found the issue. A short between the hotend fan + and E0 heater cartridge, which occurred in the network cable (stranded not solid core) that I have been using to connect my smart effector to the duet.

This explains why even on USB power only (PSU off) with the original board (v1.00) the heater cartridge was warm in the morning, it had been heating up under 5V power from the fan + to the heater cartridge all night. Probably damaging the 5v reg in the process but since its a buck converter IC, not a simple reg I'm not sure how to test it.

The second board (v1.02 clone) instead blew the U11 74HCT02 chip immediately when Vin power was connected.

So its possible the second board might be repairable if anyone can swap the 74HCT02 chip for me cheaply?

Still doesn't quite explain how the paneldue died or the 4gb sd card from the original board. Edit - the second board also killed an 8gb sd card.

But at least I now feel confident to put a third board in the machine once I have replaced all the wiring from Duet to Smart Effector.

It might be just the quality of network cable I chose or how I have implemented/routed it, but I would consider going for something better quality than network cable for this purpose, they are designed to be static, not moving around on the end of a delta effector at high acceleration for hundreds of hours. Cheap cable has cost me 2 duetwifi and one (maybe two) paneldue! I will look into multicore highly flexible quality cable.

[T3P3Tony]

@djdemond thanks for the detailed investigation and cautionary note about the cables.

The PanelDue has 5V from the Duet so that would have received 24V as David said above, how much damage that has done to the rest of the PanelDue/screen is unknown. As the SD card has failed I guess that the 3.3V regulator also failed, putting 24V onto the 3.3V supply which would have blown all the 3.3V ICs if applied for long enough. You may find the processor, stepper drivers and other assorted ICs are also damaged. I guess the 3.3V LED comes on becasue it was protected from overcurrent by the current limiting resistor.

On the second board you will need to do more investigation to see if it is just that IC that is blown. with 5V only applied to the second board does the 3.3V LED light up and can you connect to the Duet?

[DjDemonD]

I am going to try as an alternative to network cable 13 core towing-electrics cable. It has the following wire sizes within it

7 x 0.65mm2 1 x 1.5mm2 and 4 x 2.5mm2

Which should provide 7x 2 amps, 1x 10amps and 4x 20 amps of current handling, with I am sure much tougher insulation. £5/meter.

The only drawback is I suspect it will be rather heavy compared to 2x cat 5 network cables.

[Dougal1957]

@djdemond said in Might have finally killed a DuetWifi:

I am going to try as an alternative to network cable 13 core towing-electrics cable. It has the following wire sizes within it

7 x 0.65mm2 1 x 1.5mm2 and 4 x 2.5mm2

Which should provide 7x 2 amps, 1x 10amps and 4x 20 amps of current handling, with I am sure much tougher insulation. £5/meter.

The only drawback is I suspect it will be rather heavy compared to 2x cat 5 network cables.

Simon

Why not look to using silicon covered wire and make your own bundle and then just encase it all in braid 13 core trailer wire is heavy Thick and very Stiff.

Got all mine from Hobbyking

Silicon Wire

I would suggest 20awg for hot end power (at 24 Volts or 18 at 12) and 26 for all signal cables

Braid

HTH Doug

[DjDemonD]

Hi Doug,

That sounds like a good solution. I will have a look at both options. My effector has a fair amount of inertia these days as I run water cooling so I am running much lower XY acceleration (1500) than I used to eliminate artefacts. But the silicone cabling look nice.

At this point I am inclined to over-engineer it, this incident has been quite an expensive one, for the sake of a bit of wire.

I also wonder whether to remove the possibility of this happening again whether it might be worthwhile to have a fuse box between the hot end and controller, where every connection is suitably fused. Sounds like overkill but an automotive fusebox and fuses are a lot cheaper than blown controllers and displays.

[Dougal1957]

Yep know what you mean and I am soon to fir a water cooled hotend to mine and will re-wire my effector at the same time have been meaning to do it for some time, (doing so I will be able to eliminate 4 cores of the wiring as well as won't need any Fan wires at all).

[deckingman]

I once looked into using these https://www.igus.co.uk/chainflex/unharnessed-cable?tab=1

Bit expensive though...............