Pi3: usb 1-1-port2: over-current change

[jbeale]

After running MotionEyeOS without trouble on my RPi model 3 for a month or so, I notice the red power light blinking and the webpage is still live but "can't connect to camera". The USB Camera still works fine on another computer. With no USB devices connected, the red power light no longer blinks, but logging into the pi via SSH and typing 'dmesg' yields a whole lot of this sort of thing:

[ 763.865601] usb 1-1-port2: over-current change
[ 764.121635] usb 1-1-port2: over-current change
[ 764.377672] usb 1-1-port2: over-current change
[ 764.633703] usb 1-1-port2: over-current change
[ 764.889743] usb 1-1-port2: over-current change
[ 765.145842] usb 1-1-port2: over-current change

Note that NOTHING is plugged into any of the 4 USB ports, just the ethernet cable and the power. I've tried swapping in a known-good 6-inch long USB cable and known-good 5V2A power supply but same result. So after a month of working fine, this happens? Any ideas what this means?

I found this: viewtopic.php?f=28&t=110164
which casts suspicion on the AP2553 chip on the RPi board. I would really love for one of the RPi hardware designers to comment on this finding!

[jbeale]

With nothing connected to any USB port, I plugged in a USB power monitor and it was dead; there is no +5V output on the USB ports, so it is a hardware failure. Pi is running and responds normally via SSH. From the GPIO pins I measure +5.14 V on the internal power rail, and using a special cable and a voltmeter I measure 0.0V coming out of all four USB ports. I wonder how many other people are seeing this?

U13 is a 6-pin package marked "BK5r" (I think... very small letters) and may be the AP2553 http://www.diodes.com/_files/datasheets/AP255x.pdf
measured U13 pin voltages:
1: 5.14 V (USB V Input)
2: 0 V (GND)
3: constantly pulsing, 0.4V avg (ENABLE)
4: constantly pulsing, 0.15V avg (FAULT)
5: 0 V (ILIM)
6: moving around near zero, 1.5 mV to 4.8 mV (USB V OUT)

[jbeale]

Looked at the signals on U13. It keeps trying to turn on and then quickly shuts off after 8 msec, repeating at about 3 Hz. This is with nothing at all in any USB socket, in fact nothing connected except the SD card and +5V power in.

Photo of board showing U13 (USB +5V current-limit power switch)
https://goo.gl/photos/7Jm35PGRBkKuoiga8

Scope photos of U13 pin 3 and pin 4 on my board (misbehaving)
https://goo.gl/photos/W8iPZkDfCdUyCztr9
https://goo.gl/photos/gAgPKJhiegXfX9NJ9

[MaxK1]

Have you looked inside the USB sockets to see if any of the "fingers" have peeled back or if something has worked its way in to cause a short?

[jbeale]

I checked the USB sockets, they looked clean to me. I should also mention this is intermittent, at one point the system went back to working normally, but quickly failed again. I am thinking one of the ceramic capacitors on the USB output power rail (maybe C97 ?) may have an intermittent short. I tried to measure it in-circuit with the power off, and got around 700 ohms which seems low but I'm not sure what it is supposed to be. The way to prove it is I suppose to unsolder the cap.

[jbeale]

The RPi3 was working again briefly after removing the ceramic cap C97 (which means I won't be able to hot-plug USB devices) but has now failed again. In case of interest, I measured C97 to be 38 uF, not sure what it was supposed to be. Photo of board, now without C97: https://goo.gl/photos/envY8zDwa9tS2KE7A

Update: removed pin 3 from U13. Now there are no more error messages in dmesg, but the USB ports are still without power.

Finally, I replaced C97 as it seemed to be OK, and simply shorted U13 pin 1 (power input) to U13 pin 6 (USB power output). Now everything is working again.

Needless to say, I don't think the end-user should be expected to do this sort of thing

[jdb]

There are three capacitors for Vbus decoupling - C97/98/99.

MLCC capacitors fail soft short-circuit (dielectric puncture). Is it possible/likely that the USB or 5V input has been subjected to voltage overstress (backpowered hub) maybe?

[jbeale]

Thanks for the reply. In operation the USB ports were all just loads (USB camera, mouse + keyboard on passive hub, USB memory stick). During initial setup I may have had a powered hub connected, but that would still have provided only +5 V so I don't know how bypass caps would have been overstressed.

EDIT: The powered USB hub I used does have the common 5.5 x 2.1mm DC power plug input and I do have +9 and +12V power supplies with the same connector, and my desk is a mess of wires. While I don't remember such an incident, it is conceivable at some point I connected a +9 or +12 V supply to the input of the powered USB hub by accident, but I would have expected more dramatic results if so.

[esa1966]

Hi,
I get the same issue on a Pi 2 Model B.

There is no more +5V power on the USB socket. I read the other posts which speak about AP2553 :

• viewtopic.php?f=29&t=88789
  https://93.93.130.214/forums/viewtopic. ... 4&p=760997

The signal on pin EN is a rising edge which decreases after +/- 10 ms. The cycle is about 250 ms.
Nothing connected on USB port. See picture below :

En PIN signal

Rpi-V2-USB-noPower-1b.jpg (38.91 KiB) Viewed 16510 times

I though something was consuming current. So, I have placed a simple 150R resistor between +5V (PIN 2 GPIO) and PP27 (VUSB)
I measured the voltage drop which is about 1 V. 1V/150R gives less than 10 mA. I do this test without SDcard (so not running and can explain this low current as there are less power consumed by IC ).
But, it shows that there is no over current and VUSB appear from AP2553 by cycling and gives only 4V.

VUSB with 150R connected to 5V

Rpi-V2-USB-noPower-2.JPG (40.81 KiB) Viewed 16510 times

This is the resistor placement :

R From +5V

Rpi-V2-USB-noPower-3.JPG (34.67 KiB) Viewed 16510 times

Finally, I place a resistor of 1R between +5V and VUSB. When there is nothing connected, VUSB is +/- 5V.
When I connect a Wifi Dongle the power fall at 4.91V.
When I start the Rpi (SDcard inserted), VUSB fall up to 4.7 V for a short time (I suppose other IC are powered ON).

The Wifi dongle is detected corretly on 3 USB port but not on the upper USB socket near the RJ45 connector. I don't have an explanation.

Question : I don't understand which component is wrong : AP2553, the Rlimit, a capacitor, the sense current,...

[davidcoton]

I don't understand which component is wrong : AP2553, the Rlimit, a capacitor, the sense current,...

The only thing I know that can affect one USB socket out of four is physical damage to the socket itself, or the tracks or passive components that serve only that socket.

[gregeric]

That 10ms decay is ~ the RC time constant of the 220n x 100k coupling the FAULT_N output & ENABLE input of the 2553. I'd start with the easiest components to remove/refit: C97,98,99. Does the cyclical fault/enable disappear as you remove these caps? If not, and the sockets look OK, then I'd suggest it's the AP2553.

[esa1966]

If AP2553 is a protection , then it should be throng enough and protected against short circuit.
It appears that this protection circuit isn't enough robust and instead of acting as a protection acts as a fuse

[ana3bdelr7man]

[quote="jbeale"]The RPi3 was working again briefly after removing the ceramic cap C97 (which means I won't be able to hot-plug USB devices) but has now failed again. In case of interest, I measured C97 to be 38 uF, not sure what it was supposed to be. Photo of board, now without C97: https://goo.gl/photos/envY8zDwa9tS2KE7A

Update: removed pin 3 from U13. Now there are no more error messages in dmesg, but the USB ports are still without power.

Finally, I replaced C97 as it seemed to be OK, and simply shorted U13 pin 1 (power input) to U13 pin 6 (USB power output). Now everything is working again.


please can you explain what you do with U13 by pictures .... please

[handfish]

Any chance you (jbeale) can give any further information on the mods you made to U13? Did you simply cut pin 3 -- does that need to happen?) and link 1 to 6? Any hints on the locations of each of the pins: I'm having trouble orienting.

My RPi3 has had the same problems. I've gotten no power when checking PP27 against PP3.

Thanks for any help in advance.

[handfish]

The RPi3 was working again briefly after removing the ceramic cap C97 (which means I won't be able to hot-plug USB devices) but has now failed again. In case of interest, I measured C97 to be 38 uF, not sure what it was supposed to be. Photo of board, now without C97: https://goo.gl/photos/envY8zDwa9tS2KE7A

Update: removed pin 3 from U13. Now there are no more error messages in dmesg, but the USB ports are still without power.

Finally, I replaced C97 as it seemed to be OK, and simply shorted U13 pin 1 (power input) to U13 pin 6 (USB power output). Now everything is working again.

Needless to say, I don't think the end-user should be expected to do this sort of thing

Just gone with your fix of shorting U13 pin 1 to pin 6. It seems to have fixed the the USB ports in that an SD card is now registering. I'm still getting the "usb 1-1-port2: over-current change" message. To get rid of that I have to remove pin 3?

Thanks for the help!