3v short from servo use on 1.04c, now board dead?



  • I just pulled my duet 2 wifi from my machine as part of an upgrade. With everything unplugged except VIN and the duet off I plugged in a servo to the GND, +5V and heater3 pins on the expansion header. I then fired up the duet to test the servo and got it working. I jogged the servo back and forth a few times and then restarted the board from DWC to add the servo commands to config.g.
    I then ran m280 to move the servo again, which moved it and then suddenly the 3v led shut off. Tried restarting/resetting the board but no luck. Turned off board and unplugged servo, still same issue no wifi module lights either. USB power only lit the 3.3v led and reset led. The board is mounted with nylon screws and spacers onto a piece of wood. I checked for shorts etc and blew both sides with compressed air. Anyone know what is wrong? Is my board fried? Bought 3-4 months ago from filastruder.
    IMG_1426.JPG


  • administrators

    What size is the servo? Servos push power back into the 5V rail when they decelerate, and that can result in the voltage on the 5V rail exceeding the ratings of the mosfet buffers and the 3.3V regulator. The larger the servo, the more power it pushes back into the 5V rail.

    Is there any sign of damage to chips U2 and U3?

    Which LEDs light when you apply USB power? You said the 3.3V LED lit, which is the green one, but that seems unlikely if it doesn't light on VIN power.

    With USB power applied, do any chips get hot? Also check the WiFi module and the SD card socket.



  • The servo is a ds3225 25kg. I was aware of this issue with the 5V rail but thought there was the preventative capacitor on the 1.04 boards. No sign of damage to U2 or U3. On usb power the 3.3v LED is red as well as another red LED next to the reset button but those are the only ones lit.
    Nothing getting hot on VIN or USB.
    Also I should mention, when powering on VIN, I can see the 3.3v LED flash green for a tenth of a second and then go dim as in the photo above.



  • So how do you think I should proceed? Is this more repairable than say a 5V-VIN burnout? Is warranty coverage an option since I was following the protocol from the wiki? Let me know.

    Even with a replacement board, I need to use servos with this project and am scared of just burning another duet. It seems like the capacitor did not do it's job, maybe a larger one is in order? Or another form of over-voltage protection. I also need to run two servos in parallel, is that going to be an issue?


  • administrators

    @tristanryerparke said in 3v short from servo use on 1.04c, now board dead?:

    ds3225 25kg

    That's a larger servo than I've ever tested with. the largest I tested with put an 8V pulse into +5V. So I am fairly sure that is the cause of the problem.

    I am surprised that nothing is getting hot. If the 3.3V LED glows briefly then goes dim, I would expect at least the 3.3V regulator to be getting hot, probably also either the SD card or the WiFi module.

    Which country are you in?

    To drive that servo I suggest one of the following options:

    1. A separate 5V or 6V power supply (and don't drive anything else from that PSU), preferably with capacitor of 220uF or more connected between the output and ground.
    2. A Schottky diode from Duet +5V feeding a capacitor of 220uF or more, then use the output from that to power the servo. The diode will prevent the pulse from the servo getting back into the Duet. This is the arrangement we have adopted on the Duet 3 boards.

    It might be that just adding a capacitor across the 5V output would be sufficient, but without measuring the amount of power fed back from the servo, i can't be sure. Two servos will feed back twice the amount of power, increasing the size of the voltage surge.



  • @dc42 Thanks, I'm in the USA northeast.

    My servo said 4v-7.8v on the side, but I guess the braking can exceed that?

    With this separate 5v supply idea would you recommend opto-coupling the PWM signal from duet and keeping the two circuits separate? I just want to find the safest way to run these servos consistently. I also have an SX1509 that I was planning to drive servos with, would that be an easy method to protect the duet by powering that separately?

    I will plug it back in today and check for heat on those chips again.


  • administrators

    If it's a 5V PSU without a mains ground connection, no need to opto isolate the servo control signal.



  • Ok, I'll order a some 5V PSUs, needed one for the RPI+duet2 setup anyway.
    What do you think about warranty replacement/repair?


  • administrators

    Yes I approve a warranty replacement.


Log in to reply