Dead Duet Wifi

  • administrators

  • administrators


    BTW. I think that erase button is the dumbest idea. It should use jumpers similar to clearing CMOS on a computer so people never accidentally hit that button. Putting that button next to the reset button is just asking for it to pressed by accident.

    Indeed. I did the design when we still did not know how well the whole upload&flash over network would work and wanted to be able to get to it from the edge of the board in an installed position in case we needed to manually erase before each firmware reflash. #hindsight

  • @dc42:

    Have you read ?

    Yes, I have gone through all of those steps. I will try the last step one more time this evening, but I'm fairly certain I did hold down the erase button for a few seconds followed by a reset at some point while I was troubleshooting yesterday.

  • administrators

    Are the LEDs correctly lit? Is the microcontroller chip getting hot?

  • The LED's are correctly lit - the 3.3v and 5v are lit in addition to the two LEDs directly on either side of the USB connector. The only thing I noticed getting warm to the touch was the wifi module, but I'll be able to verify that again in a couple of hours.

  • Just to confirm, the micro is not getting warm. The wifi module does, however, get quite warm (I don't recall whether or not this was the case before). I also re-checked the Erase then Reset procedure, and that is a no-go. At this point I'm just going to get a new board on order tomorrow as I need to get the machine back up and running. Still, I'd feel a lot better about putting in the replacement if I knew exactly what happened to this one.

  • administrators

    It's normal for the WiFi module to get warm.

    I'm wondering whether the USB cable is faulty. Have you tried another one?

  • Yes I had the same thought last night so I tried another one, same result.

  • administrators

    I think it's time for you to ask your supplier for a replacement under warranty.

  • I assume you guys will be doing autopsies on the dead Duet's to see what killed them ? Not a simple task that I know, but perhaps a pattern will emerge.


  • Yes, they're all being returned to T3P3. Worth noting that the current failure rate is 3% - and some of those may have been caused by issues outside the Duet itself such as wiring or user error.

  • @elmoret:

    Yes, they're all being returned to T3P3. Worth noting that the current failure rate is 3% - and some of those may have been caused by issues outside the Duet itself such as wiring or user error.

    Probably would have been nice to see some TVSS, MOV's, etc. in certain areas to try and keep problems down. I know the micro has some very limited protection against static and such, but it's not much. I think the biggest problem will be people with bad wiring on motors tossing hundreds or thousands of volt spikes around and perhaps inducing voltages into endstop wires, thermistors, etc. I killed my micro , I know how I did and I replaced the chip…no biggie, I've also seen some really stupid videos on You-Tube on how to crimp wires, if people follow some of these "tutorials" I can see failure rates increasing.

    Edit: Not a reflection on the OP.


  • The following is a subset of reported failure causes, selected randomly from the full set in order to not release sales volumes.

    Duet Wifi Incomplete soldering
    Duet Wifi Failed PS_ON functionality
    Duet Wifi Flaky bootup
    Duet Wifi Dead stepper driver
    Duet Wifi FAN0 is dead
    Duet Wifi Intermittently bad SPI comms - cannot update firmware

    Some of these (incomplete soldering, SPI comms, flaky boot up) are manufacturing problems.

    Some of them (dead stepper driver, fan 0 is dead) are user error, for example not understanding the thermostatic functionality of the fan channel - but that was not discovered until after the warranty exchange. Now we know to ask about that when troubleshooting 🙂

    Failed PS_ON functionality - not sure what caused that.

    Looking through the full set of failure data, I don't think any, with the possible exception of the dead stepper drivers, could be the result of a voltage spike. Protection is always a balance of cost vs predicted failure rate. If adding protection would increase board cost by $3 but only 0.5% of users would have experienced that damage, it doesn't make sense to implement on a ~$200 product.

    With all that said, it would be nice to see Vin reverse polarity and overcurrect protection on the next revision, and I've requested as such to T3P3. Would also be nice to see input protection (inline resistor + diodes), space permitting.

  • Yeah, it's always a balancing act. I think that 0.5% could be mitigated through some education. Forums like this certainly help.


  • administrators

    A little more information about protection on the Duet WiFi.

    We increased the protection on the Duet WiFi compared to the Duet 085. So far we have no indication that voltage spikes have caused any Duet WiFi failures, with two exceptions:

    • A couple of people connected brushed DC motors to fan outputs (e.g. Berd Air) and didn't connect a flyback diode across the motor (in fairness, at that time we didn't tell then that they needed to). This caused the fan mosfet to fail because of the spike due to the back EMF from the inductive load. Had we realised that some people would connect brushed DC motors, or seen this problem occur with the Duet 085, we would have included flyback diodes on the board for the fan outputs. We'll be adding them at the next PCB revision.

    • If stepper motors are disconnected under load, this produces very large voltage spikes that can blow the stepper drivers. I'll be looking at whether we can protect against this in future version of the Duet. I know that there are some very expensive external drivers that incorporate such protection. RAMBO boards attempt to provide this type of protection, but I've still seen complaints about blown drivers on RAMBO boards so I guess the protection they added isn't 100% effective. I have a feeling that providing 100% protection would require some inductors to be put in the circuit, which would be expensive and require a larger board to accommodate them.

    We have ZERO evidence that ESD has caused any Duet WiFi failures. So I'm far from convinced that TVS diodes would contribute anything to reliability. It's very easy to add protection devices without any evidence that they solve a problem. OTOH we know from experience with the Duet 06 and 085 that shorts between heaters and thermistors (causing 12V or 24V to be fed into the thermistor inputs) are very common, which is why we designed the Duet WiFi to survive those - at the expense of the VSSA fuse in some cases.

    It's also possible for the user to connect VIN with reverse polarity. Some boards attempt to provide protection against this, using either a fuse and crowbar diode, or (in the case of Smoothieboard) a P-channel mosfet. I'm not convinced that the crowbar diode will give full protection - the voltage drop across it before the fuse blows may still allow too much current to flow through the stepper drivers to damage them. The circuit used on Smoothieboard should protect better, but needs more space and increases cost. However, I have never heard of anyone applying reverse VIN power to a Duet, so it doesn't appear to be a significant risk and we didn't include that type of protection. I guess it's possible that a few people have reversed the VIN polarity but been too embarrassed to admit it!

  • I don't think reverse protection is really needed. Kind of one of those don't drink and assemble rules.

  • @dc42:

    I think it's time for you to ask your supplier for a replacement under warranty.

    When I inquired with the supplier, they weren't sure about the warranty or replacement policy and had sent a note to T3P3. If you can point me to an official warranty policy that'd be appreciated.

    I've already purchased a replacement board and had it overnighted, as I had not heard back yet on my inquiry with the supplier and needed to get the machine back up and running. Since we weren't able to identify the cause of the failure, I still assume it could've been something I did, in which case I have no problem with paying for a replacement. Either way, I'd be happy to ship my dead board back for analysis if it might reveal the cause or help improve future products.

  • administrators

    Please email with more details (I.e. who the supplier is, the failure stuff is well documented in this thread). We will certainly want the original board back for investigation however normally this will be arranged as part of the warranty replacement process.

  • I (Filastruder) was the supplier, Roland was CCed on an email last night. The only oddball items is that the board was delivered back in September as part of the pre-order.

  • administrators

    Hi Tim, sorry I missed that. Thanks for clarifying quickly!

    cchaffee, ignore my previous comment, Tim is all over this.

Log in to reply