Duet 3 fried

NeoDue

@elmoret that very much looks like the board is screwed onto some bent aluminium plates to me... and the vout_0+ connector also has a rather unhealthy colour. Judging from what can be seen, I fear there is a nice short-circuit caused by those metal plates down there...

new_to_me901

@oliof the documentation doesn't state what the capacitors and chips are connected to. I've looked. Thank you, though.

Edit: nevermind, I'm dumb. This is different than what I've seen. I will search this now. Thank you for your help.

new_to_me901

@NeoDue @elmoret the metal plates are attached to the wooden base, the board is attached to the wooden base. The board and the metal plates don't have contact. There are also plastic spacers between the wood and the board. I can take photos and post them later today if that's helpful? You can see the edges of the spacers in that photo at the bottom of the board.

As for the v_out0, yes this looks like damage but that photo was from two weeks ago. I'm not sure if it had this discoloration when it was installed, but it's attached to a cooler and has been running for months without obvious issues. Could it have slowly been damaged? Is there anywhere else which you'd recommend to check for damage?

This system has been running for about year with changes/upgrades here and there. But no major changes to this hardware recently. I had unplugged a heater connected to the out_8 to change the housing, then plugged it back in. When I first switched the system on, the chip sparked and flamed within a couple seconds. The vout didn't have apparent damage at that time. There wasn't an arc or anything.

Thank you for your help

NeoDue

@new_to_me901 might be good to show some more pictures - both from all the sides in assembled state, and from the underside of the Duet and the empty place where it ought to go with the Duet unscrewed. On the second picture of the first post, it very much looks as if the wood is below the metal on the back of the board. And since the metal seems to be anodised aluminium, this would also explain the delayed failure: the aluminium oxyde layer created by anodisation creates a nonconducting "cover" on the metal which prevents short-circuits - but this is far from being reliable since the thin layer can easily be destroyed by even slight scratches. Even unplugging or replugging a connector might cause this.

By the way, wood is also conductive if it is damp enough (which again might change over time by storing conditions, e.g. if you left your printer in a no-so-dry cellar over winter), therefore putting a piece of wood into direct contact with the contacts of a PCB might not be the best idea either, especially if you have higher supply voltages.

And if I take a look at my Duet, the plastic washers supplied are not thick enough to provide enough distance between the board and a base. As with any other PCB, the user needs to provide some additional space on the underside - which is also relevant for cooling. That is something one can quite easily achieve by printing a simple mounting base or housing for the Duet before installind it into the printer. There are numerous ready ones to choose from, or you design your own if there are special constraints.

A minimal, yet perfectly working mounting base can be seen here: https://forum.duet3d.com/topic/14216/show-me-your-duet-3-enclosures/3
And if you want it more difficult (significant height constraints, only predefined mount points available, active cooling of the Duet underside and such) this is what I cooked out for my current printer: https://www.myminifactory.com/object/3d-print-snapmaker-j1-halter-für-holder-for-duet-2-6hc-329588

new_to_me901

@NeoDue You are right, it was actually mounted to the metal. I am unsure if it is aluminum but will be able to check with the person who built this setup tomorrow. I have found multiple damaged chips (in addition to the ones found before) and am assuming this board is now unusable. I will be able to obtain a new one. Can you please help me to ensure it is safe going forward?

I have included pictures of how it was mounted. I am wondering if putting the board above the aluminum would have made a safe difference, even if it's still mounted to the aluminum? I also have a spare small mounting bracket printed that I found extra from someone else's setup. It does not have the fan, like yours does. I've included potential mounting set ups for this, are any of them appropriate? I think I should be able to implement your setup fairly easily. But what materials are appropriate for that to sit on top of? Is the wood okay, and are there any I should avoid completely?

I am concerned for the other Duet3's being used in my work space too. I have also included the photos of the set up for those. They do not have fans directly cooling the boards. They also seem to be mounted to metal (with plastic in between). Is this a safe set up for long term use?

We are using 24V power supply (also pictured).

I so appreciate your help. I am utilizing the system but am unfamiliar with how it actually works. I'm very much out my depth here and you've been a great help in identifying the problem. I sort of short circuited too, and was in a panic when I first posted. You have helped tremendously. Thank you.
Duet3 mounting pictures

NeoDue

@new_to_me901 Okay, the spacers you used are high enough, the board does not rest on the metal as I feared. But you (and the other guy who made the plastic frame as well btw) ignored this part of the manual:

Do not use metal washers under the mounting screws, or screws with large heads (e.g. button head). Nylon washers are recommended and supplied with the board. With a metal washer or large screw head there is a risk of creating a short.

I do not know which parts the screws you used might have contacted, but you should definitely add the supplied washers between screw head and PCB.

A second issue may arise due to the hand-bent metal elements that are far from being even or even parallel to each other. If you tighten the screws on those, the PCB will get bent since each screw tries to force it to be parallel with the corresponding contact surface - which is something you should avoid if you want to keep it working for longer. Therefore yes, the alternative holder is definitely the better choice. Just use any non-conductive (!) plastic you have available for printing - standard PLA, PETG etc. will be fine. Just keep away from any plastic that is marketed as "conductive", "for shielding", "ESD" or such. If in doubt, check with a multimeter to be sure. With your Duet seemingly being in open space, you probably won't need active cooling (depends on your stepper current though).

However, the pictures do not clearly show a reason for the death of the Duet unless I overlooked something. Apart from updating the mounting frame, I therefore suggest you do check and clean up your wiring.

new_to_me901

@NeoDue Okay, so it could possibly be the screws, or possibly the lack of plastic spacer between the screws. I'll try to implement these changes along with the plastic mount. But otherwise the answer isn't obvious..

Circling back to the vout_0, I found older pictures of it which do not show damage to this wiring. Is it possible that this equipment (a cooler) failed and caused the short? Is there a recommended way to test each piece of equipment individually for fault? Without frying another board? We have multimeters available for use.

Thanks again for your help!

NeoDue

@new_to_me901 the Duets are very well protected, therefore I personally doubt a simple failure of one element that is powered to V_IN as your cooler is would have created this. If the failure is in the wiring, I would search for some short that causes a higher potential to be inserted where it does not belong. For example, any ungrounded metal parts that carry more than one electric component might be relevant since they could act as a "wire" that connects these two components.

But without knowing the exact schematic of whatever you are doing with the Duet, it is impossible to say this for sure - especially since you seem to have some pretty nonstandard use case, with a cooler at a connection where a heater belongs and a heater at a place where a fan belongs... an ice-cream printer maybe?

For testing, do the usual:

does each component work and draw a suitable amount of current if it is connected separately to a suitable power source?
Is the sum of all 3.3/5/12V currents within the range of what the Duet is capable to deliver?
can you measure any shorts between any two of the component's wires or any wire and the metal housing/frame of the component if there is any?

new_to_me901

@NeoDue okay I'm going to try these things! Thanks again

new_to_me901

@NeoDue it looks like I have a short where the thermocouple daughterboard connects. When I connect the multimeter (on ohm setting) to the +3.3V pin on the back of the duet 3 board and any other ground, the multimeter shows resistance. ~2.8-3.8. However when I do the same to the daughterboard itself there is no resistance.

I now see it would have been relevant to state the daughterboard has always given me trouble. Particularly when I had two thermocouples connected to it, it would power on with the Fault light showing on one of the two. I would simply disconnect and reconnect, power everything back on, and that would fix it. Sometimes I'd have to repeat many times. I tried multiple of these daughterboards and multiple duets, and still it happened. So I resigned myself to just check for the fault at each power up of the system. That seems like it was a mistake. Does this sound like it would have been the cause of the short?

I will continue to do more testing in the meantime! The continuity was the fastest/easiest to start with. Wires are next.

NeoDue

@new_to_me901 Hm, I fear I have no personal experience with the thermocouple daughterboard.

@oliof do you happen to know these addon boards or have someone in mind who could tell if that resistance is something that could be within expected ranges?

jay_s_uk

@new_to_me901 the temp daughterboard info is here if you need it https://github.com/Duet3D/TempDaughterboards

new_to_me901

@NeoDue Well as far as I know there shouldn't be any resistance at all between these two pins. It should read 0L, as it does for all other pins when connected to the ground. I also have a new duet3 board that hasn't been used yet, and it does not have resistance when touching the same pins: 3.3V and ground.

When I search the forum it looks like I'm not the only one who has had problems with the thermocouple daughterboards. Maybe I can get some good information from those threads.

dc42

@new_to_me901 when thermocouples give bad readings or the daughterboard error lights turn on, the most common reasons are:

Short circuit between one of the thermocouple wires and another wire (e.g. a heater wire)
Thermocouple element is not isolated from the hot end. The thermocouple elements supplied by E3D do have isolation between the element and the metal case. Many other types of thermocouple do not. If there is a short between the hot end metalwork and the heater (for example, because the heater insulation has broken down), or the hot end metalwork is connected to mains ground but the Duet ground is not, then this could feed a high voltage or transient into the daughter board, which could result in the chip failing and passing that voltage or transient t the +3.3V rail. That is likely to blow the 3.3V regulator and other components.

Can you disconnect the thermocouple from the daughterboard and use a multimeter to check whether there is any continuity between the thermocouple wires and the hot end metalwork?

new_to_me901

@dc42 Thank you for your replies. The thermocouples are insulated from the hot end.

I have checked continuity for these and have not found any continuity where it should not be. I've tried every combination possible, thermocouple to heater, thermocouple to insulation, thermocouple to heater wires, thermocouple to heater wire insulation, and then the same combinations through the heater power wiring.

I'm now working on powering the heater through an older heat controller so as to not fry another Duet if it is the culprit. Just want to see if it still works as intended. If it still works, I'm sort of at a loss. I've also seen that length should be taken in account for thermocouples. The thermocouple wiring is about 19.75 inches. Is this an acceptable length?

FYI we are using the v1.1 daughterboard with a k type thermocouple.

dc42

@new_to_me901 thermocouple wire length can be extended but you should use proper K-type twisted pair thermocouple wire to extend it. If you use plain copper wire then then you will move the cold junction to where the wires join, so the reference temperature measured on the daughter board by the chip won't be the same.

Please ensure that both the hot end metalwork and Duet ground are connected together, in case the problem was caused by ESD.

When and where did you buy the 6HC?

new_to_me901

@dc42 We did not actually extend the thermocouple wiring, so it seems like that isn't the culprit.

I will check for what you advise when I get back to the workspace later today and let you know.

We started using these boards more or less than 2 years ago, from filastruder a distributor in the US.

dc42

@new_to_me901 we may be able to suppy a refurbished board to you. Please email sales@duet3d.com with your details including a link to this thread.

NeoDue

@dc42 Out of curiosity: is it possible from your experience to fry a chip that much as it is visible from the last picture in post 1 with ESD currents alone? I have seen numerous failures from ESD from invisble but dead components to exploded ones - but never a chip housing softened and melted like that. I always thought ESD discharges are too fast to melt plastics like that, hence the question.

dc42

@NeoDue said in Duet 3 fried:

@dc42 Out of curiosity: is it possible from your experience to fry a chip that much as it is visible from the last picture in post 1 with ESD currents alone? I have seen numerous failures from ESD from invisble but dead components to exploded ones - but never a chip housing softened and melted like that. I always thought ESD discharges are too fast to melt plastics like that, hence the question.

I don't believe that the energy in a "normal" ESD would be enough to fry a chip in that way directly; but it may trigger a failure mode in the chip that causes it to draw a lot of power, or that subjects another component to excessive voltage. In particular, small 3.3V linear regulators usually have an absolute maximum input voltage rating of 6.5V and any higher causes a breakdown that causes them to draw a large current.

We know that ESD can directly fry small (0402 size) resistors.