U3 "A4403" started to smoke



  • Hi

    My rather new duet2 wifi rev 1.04b started to smoke when I was probing my bed. U3 burned a hole in itself just as others have shown in pictures.
    I have used a 12V M12 inductive probe for 2-3 weeks, but yesterday I changed to a smaller 5V M8 inductive probe. (The 12V for the old probe was from an external step down converter.)
    For the 5V probe, I used the 5V from the duet expansion connector. I would say that it did around 40 probe points, but then suddenly U3 started to smoke. My rail was 25.8V. No other active components were connected to the card more than the probe and the main PSU.

    Anyone knows why? Have anyone checked the 5V AC power spectrum, with a 5V inductive probe connected and modulating the distance. My first thought when I found others with smoked U3, was regarding the phase margin of the closed loop of U3 when attaching a non-resistive load to 5V. But it is just a very wild guess.

    I ordered the card directly from duet3d.com 16/04/2019.


  • administrators

    I'm sorry that U3 has failed on your board. We've had a few unexplained failures of that chip recently, after 2 years with no failures at all. Please go to https://www.duet3d.com/warranty and follow the link to the warranty form to arrange a replacement.



  • Hi

    Thanks, I filled in the Warranty Form.
    I forgot to mention that the PanelDue 5i was connected to the board. I am not convinced that it survived.
    I can test it with a current limited bench supply @ 5V, (if you tell me what to limit the current supply ), or I could just send it in as well so it can be verified or replaced. The cards were bought together.
    The micro-USB connector X3 on PanelDue 5i, should it be able to power the board stand alone?


  • administrators

    If you connect the PanelDue to a PC or a 5V power supply using a USB cable, then it should display the main screen even though it is not connected to a Duet.



  • Connecting a USB cable to X3 results in a rapid temperature increase in (at least) DA1.
    Connecting the bench supply to X1:

    50 mA , 1.8V
    100mA , 2.0V
    200mA 2.2V
    300mA 2.4V
    400mA 2.6V
    500mA 2.7V
    600mA 2.8V - Aborted

    U2 seems to get hot, no signs of light from the panel.



  • Sales helped me to solve the issue very quickly. 🙂

    Is there any information on how to avoid the problem happening again from a user perspective? 🤓

    I will probably add a small fuse on the 5V and thereafter a shunting ~ 5.4V zener to the ground on the cable between duet2 and the panel to avoid future overvoltage of the panel. Are the Absolute Maximum Ratings 6.5V for the 5V input of the panel?


  • administrators

    @mikael said in U3 "A4403" started to smoke:

    Sales helped me to solve the issue very quickly. 🙂

    Is there any information on how to avoid the problem happening again from a user perspective? 🤓

    I will probably add a small fuse on the 5V and thereafter a shunting ~ 5.4V zener to the ground on the cable between duet2 and the panel to avoid future overvoltage of the panel. Are the Absolute Maximum Ratings 6.5V for the 5V input of the panel?

    Unfortunately we don't know what causes these occasional U3 failures. It might be caused by a bad batch of chips, or to a design change we made recently. I will conduct further stability tests on the 5V regulator later today or tomorrow.

    A 5.6V power Zener diode across the 5V rail might provide additional protection, if it can handle the over-current before either U3 or the fuse blows.

    The 3.3V regulator on PanelDue is rated at 6V absolute maximum, but I know from experience that in practice they tolerate a lot more before they fail. The other critical component on PanelDue is the backlight inverter on the display panel.



  • I tested my probe at 5.15V and it still worked as before.
    The DC current consumption was ~ 700uA when the integrated LED was off, and ~5mA when the LED was on due to close proximity to a screwdriver.


 

Looks like your connection to Duet3D was lost, please wait while we try to reconnect.