RPI2 power issues?

[oneiroi]

Ok I've been facing some oddities and long hours debugging on this issue.

Namely when using a "high power" WiFi device, be it Realtek or Atheros the device will operate fine sometimes for hours, after which a vmem alloc failure will occur which causes the device to crash entirely.

The devices in question are comparable to an Alfa AWUS036H, using a Realtek "micro" wifi adapter does not cause any of these issues which is what leads me to think this is a power draw issue over the USB bus; is anyone else encountering these problems when using "high power" usb devices?

If not I may RMA this unit in the hopes I have received a faulty unit.

[oneiroi]

Ok I've "augmented" the power for the "high power" WiFi by using a "USB Y" cable and attaching the secondary power input to a battery pack and an in line power meter, it's draw is 5.01v @ 0.23A on this secondary input; thusfar the setup has been stable and running for some 2 hours.


Suggestions on how to further test this would also be welcome.

[drgeoff]

Try using a better PSU and/or cable feeding the RPi and see if the Y-cable can be dispensed with.

[oneiroi]

Confirming the USB Y cable has allowed for continuous operation

Code: Select all

root@raspberrypi:~# uptime
 09:12:04 up 11:15,  1 user,  load average: 0.57, 0.55, 0.58

[oneiroi]

Try using a better PSU and/or cable feeding the RPi and see if the Y-cable can be dispensed with.

Tried multiple cables, multiple PSU's, multiple systems in place of a PSU (laptop, workstation etc, etc); the only thing that's worked so far is the USB Y cable.

I don't really have the setup to carry out more extensive tests, I can only point of the following observations using an in line USB power meter.

- Pi USB: 4.99v @ -0.2A
- Secondary USB power: 5.01v @ 0.23A

This is stable though seems to imply a current flow back into the USB bus.

[tpylkko]

what psu/cable are you using?

[obcd]

Could it be RF interference from the stick into the Pi electronics?
I had some issues with a Huawei 3G stick.
When I put a cable between the stick and the Pi, the problems went away.
Maybe your Y cable is doing the same and making the distance from the stick to the Pi larger decreasing possible interference.

[oneiroi]

what psu/cable are you using?

USA A -> Micro cables from Nexus 5, 7, Oneplusone phone devices.

PSU: USB3, USB2 Desktop workstation, USB3, USB3 Laptop, Wall socket chargers supplied from afor mentioned phones / tablets (5c @ 2A).

Also tried a number of USB battery packs from portapow.

Note, these have all works absolutely fine with the original Model B pi and beagle bone black; only having issues with the Pi2

[oneiroi]

Could it be RF interference from the stick into the Pi electronics?
I had some issues with a Huawei 3G stick.
When I put a cable between the stick and the Pi, the problems went away.
Maybe your Y cable is doing the same and making the distance from the stick to the Pi larger decreasing possible interference.

Prior to the Y cable, was using a 1M cable; it did not seem to matter whether the Pi2 was right next to or away from the WiFi adapter unfortunately, the only thing I can confirm is having the power draw from a separate source for the WiFi adapters has solved the issue now over 13hours of continuous operation. I've have the wifi adapter right on top of and far away from the Pi2 without issue in this configuration.

Removing the secondary power feed however leads to the same crash I described earlier.

[gkreidl]

Assuming that your power supply + cable (!!!) really supply enough power without voltage loss, did you enable the usb high power mode
max_usb_current=1
in config.txt?

[oneiroi]

Assuming that your power supply + cable (!!!) really supply enough power without voltage loss, did you enable the usb high power mode
max_usb_current=1
in config.txt?

No I had not, in the interestes for as full a test as I can carry with what I have here please note the following.

I moved the inline power meter to sit between the PSU (laptop USB3) and PI;

NoWiFi: 4.97v @ 0.32A
WiFi (no Y split): 5.07v 0.41 -> 0.50A (fluctuates)
WiFi (Y split): 5.11v @ 0.2A


Added: max_usb_current=1

device descriptor read/64, error -110 noted on boot, had to unplug and replug the WiFi device

WiFi (No Y split): 5.07v @ 0.40A (not fluctuating, quiet stable draw)

Device running 5 minutes without crash so far

Edit: locked up again no error displayed this time w.r.t malloc fail, going to go back to Y split

[gkreidl]

enabling usb_max_current does only make sense, when you feed the Pi with a good (no crappy cables!) 5V2A power supply.

[tpylkko]

you should try a better cable. 20 awg or better.