Duet 2 Wifi U3 failure
-
Yes the board was powered by 24V VIN when I connected to the PC.
-
Please contact Printed Solid and initiate a warranty exchange. Include a link to this thread as authorization.
It's possible the USB ground loop contributed, but it's also possible it would have happened anyway due to infant mortality of the chip itself.
In future though, do take care to avoid/mitigate USB ground loops.
-
@Phaedrux Thank you. I've already filled out the reseller warranty form through Printed Solid. I will update the link in the form to point to your post instead of the general thread.
-
I don't know if any of the engineers read these posts, but the curiosity was getting to me. I plugged the board in again to a micro usb breakout with a current limited power supply with nothing else plugged in to see what components were heating exactly. D2 seems to be passing a lot of current. Even after using hot air to desolder U3, D2 was still heating up enough to give me a little diode shaped burn on my finger after only a few seconds.
U3 seems to be the victim of some other failure mode rather than with the chip itself. If there's any particular points any of the developers would like me to probe, let me know.
-
They do frequent the forum, a down right surprising amount.
D2 shouldn't be conducting anything at all with U3 out of the picture, and if memory serves it has a significant reverse voltage rating so the most likely reason for it to fail would be too much current. I'd check the Vbus of the computer you connected and look at the USB connector and traces on the Duet for signs of excessive current.
Of all the failed U3 parts back when they had a suspected bad batch of the A4403, afaik, only one case of D2 and U3 both failing, also USB related as opposed to the majority of the U3 failures.
-
@bearer You're right, the reverse voltage on that diode is 40V and maximum current is 3A.
Tomorrow I've got a few things to check.
Vbus of the computer I connected to, though it hasn't cause me problems with any other devices in 5 years.
Current running through the USB port. It should be limited to 500mA.
I will see if I can't also get the voltage across D2, R85, and R86.
And I will check for continuity between pin 8 of the U3 pads and any other nets that it's not supposed to.Thanks for your input. I'm getting the impression the Duet community is pretty robust.
sources
https://raw.githubusercontent.com/Duet3D/Duet-2-Hardware/master/Duet2/Duet2v1.04/Duet2_1.04c_Schematic.pdf
https://www.mouser.com/datasheet/2/315/DB2W40300L_E-1141713.pdf -
VBUS on my PC is 4.6V (seems low but it's consistent across all my ports?) so that's probably fine. I only have to worry about ground loops now regarding that port.
I desoldered D2 and it is indeed a jumper now, not a diode.
With D2 out of the way, I'm no longer reading half an ohm of resistance between pad 12 of U3 and GND, and the duet 2 wifi seems to have kicked its smoking habit. R85 and R86 are both just 0.15Ohm so I'm guessing that heat was from 4.6V across 0.075Ohm through those two resistors. Yikes. It's possible that my finger tip was burned by R85 and R86, not D2.
-
@veloxsouth said in Duet 2 Wifi U3 failure:
VBUS on my PC is 4.6V (seems low but it's consistent across all my ports?) so that's probably fine.
its quite low, but I would think d13 would block the voltage. but if d13 didn't do its job the 5v regulator might be trying to increasing its output to get the voltage back up to 5v and possibly end up on a feedback loop?
anyways, weather d2 or r85/86 burned you would depend on which had the higher resistance as it would dissipate the most of the heat when vbus was more or less shorted to ground.
-
I like your theory, @bearer . I wanted to see what condition D13 was in but I can't seem to find it anywhere on my board. I will have to fire up kicad and search for it on the board layout to be sure of where it is.
-
I just got word from printed solid that it's time to return the fried board so I'm going to wrap this up. Thanks you both for the time and attention. At least we learned that ground loops need to be taken seriously, and that U3 and D2 have certainly failed.
