Jump to content

Orange Pi PC randomly stop responding


giacomo892

Recommended Posts

Hi all.

I'm running armbian in all my orangepi boards since I've them and the distro is great and we'll developed. (thanks!)

 

I run them with telephone chargers (htc and Samsung) 2/1.5A directly connected to the board via a 30cm cable. I don't know what's the status of data pins (d+ d-) but I think they are left open by the cable. They are used for sdr use running dump1090 all the time. In the aim of reducing the heat I've disabled two cores (only recently) out of four and installed a copper heatsink on the cpu.

 

The problem is that from time to time I found my boards unresponsive from ethernet and serial port, so completely stuck. Only a reboot works. Do you have any idea of what can cause this behaviour? Thanks

Link to comment
Share on other sites

I am guessing it is power adapter problem. You need test cpu on full load and measure voltage drop on easy accessible usb or gpio pins to be sure your power source is suitable.

But if you want lowerdown power consumption try set frequency to 480 or 640Mhz. It can help you save about 40% of energy compare to 4 cores running on 1296Mhz and on 100 load.

 

Sent from my Redmi Note 2 using Tapatalk

Link to comment
Share on other sites

I am guessing it is power adapter problem. You need test cpu on full load and measure voltage drop on easy accessible usb or gpio pins to be sure your power source is suitable.

But if you want lowerdown power consumption try set frequency to 480 or 640Mhz. It can help you save about 40% of energy compare to 4 cores running on 1296Mhz and on 100 load.

 

Sent from my Redmi Note 2 using Tapatalk

Thanks. I'll measure the voltage as soon as I can. 

Link to comment
Share on other sites

 

My boards running perfectly with armbian system and I can tell you only 1 or 2 of 10 power adapters/charger for phones are suitable for singleboard computers asking for 5V. Problems is there are lot of fakes on market and adapters with very poor/simple circuit design not able to deliver current and voltage specified by manufacturer. This is not problem for phone charging, because for most people is not difference between charging time 2 or 4 hours, when you charging phone overnight but for computers it is. But if you like corrupted data...
 

Check link on the right, there are some test of power adapters... http://www.lygte-info.dk/info/usbPowerSupplyTest%20UK.html
 

One of my Orange Pi PC has been powered by power adapter with rated output 5V 3A. I bought it together with Orange Pi mini 2 board, but it works with Orange Pi PC and Orange Pi PC One as well.
 

You really need proper power source. Do not underestimate it.
 

And for second. If you have stability issues ending in reboots run cpu stability tests immediately:

http://linux-sunxi.org/Hardware_Reliability_Tests

Link to comment
Share on other sites

My boards running perfectly with armbian system and I can tell you only 1 or 2 of 10 power adapters/charger for phones are suitable for singleboard computers asking for 5V. Problems is there are lot of fakes on market and adapters with very poor/simple circuit design not able to deliver current and voltage specified by manufacturer. This is not problem for phone charging, because for most people is not difference between charging time 2 or 4 hours, when you charging phone overnight but for computers it is. But if you like corrupted data...

 

Check link on the right, there are some test of power adapters... http://www.lygte-info.dk/info/usbPowerSupplyTest%20UK.html

 

One of my Orange Pi PC has been powered by power adapter with rated output 5V 3A. I bought it together with Orange Pi mini 2 board, but it works with Orange Pi PC and Orange Pi PC One as well.

 

You really need proper power source. Do not underestimate it.

 

And for second. If you have stability issues ending in reboots run cpu stability tests immediately:

http://linux-sunxi.org/Hardware_Reliability_Tests

 

My boards running perfectly with armbian system and I can tell you only 1 or 2 of 10 power adapters/charger for phones are suitable for singleboard computers asking for 5V. Problems is there are lot of fakes on market and adapters with very poor/simple circuit design not able to deliver current and voltage specified by manufacturer. This is not problem for phone charging, because for most people is not difference between charging time 2 or 4 hours, when you charging phone overnight but for computers it is. But if you like corrupted data...

 

Check link on the right, there are some test of power adapters... http://www.lygte-info.dk/info/usbPowerSupplyTest%20UK.html

 

One of my Orange Pi PC has been powered by power adapter with rated output 5V 3A. I bought it together with Orange Pi mini 2 board, but it works with Orange Pi PC and Orange Pi PC One as well.

 

You really need proper power source. Do not underestimate it.

 

And for second. If you have stability issues ending in reboots run cpu stability tests immediately:

http://linux-sunxi.org/Hardware_Reliability_Tests

Thanks. As soon as I can i'll try with a new power supply and trying the stability tests!

Link to comment
Share on other sites

I'm measuring 4.76V (using a cheap multimeter) on the GPIO 5V pin using an original Samsung 5v 2A tablet charger. Isn't that enough? 

 

Nope. If you measure 4.76V when your device is idle, then simply do a 'stress -c4' and watch your board dying (and voltage dropping down to maybe 4.0V or below).

 

It's really simple: If you try to operate your board in an unreliable way (using crappy phone 'chargers') it will become unreliable. Everything as expected.

 

BTW: we include both stress and sysbench in our default package list for a reason: Since it's pretty easy to detect insufficient power supply using both of these tools. :)

Link to comment
Share on other sites

Well, in french we use to say "Tomber en marche" for such problems. It is a computer scientist joke. It means that a software bug is no longer occuring and we don't know why.

 

Connectors and PSU are more or less crappy and wires too thin. Multimeters cannot show peak amperage demands. Boards (software and hardware) and accessories generate unpredictables peak loads so fiddleling with cables is often sufficient to crash the board some time later or on reboot.

 

Is there a software or hardware solution to trace (and record !) power failures before the system crash ?

Link to comment
Share on other sites

The board is not idle when I measure 4.76V. It has attached an RTLSDR key and decoding ADSB packets with main process lying at 55% of CPU :)

 

I'm going to try the stress and measure the voltage again.

 

EDIT:

 

Done the measurements:

 

With the same load as stated before + stress -c 4  the voltage is lying at 4.71-4.72V

 

I'm leaving the stress on for some time 

 

EDIT2:

 

The board is able to run 4h with stress -c 4 without any issue.

Link to comment
Share on other sites

Is there a software or hardware solution to trace (and record !) power failures before the system crash ?

 

Unfortunately not with H3 boards. With A10/A20 it's different. Those SoCs are accompanied by a PMIC (AXP209) which contains several ADC converters that can be used on some/most boards to monitor DC-IN voltage (at least with A20, I never had an A10 board).

 

This here is from an Olimex A20-Lime2 that I misuse as UPS equipped PSU for a Banana Pi M2+ (I use the Lime2's left USB port to connect to BPi M2+'s Micro USB port -- cable is short, has low resistance, so voltage drops aren't an issue and I get a direct network connection using the same cable for free):

 

Bildschirmfoto%202016-07-16%20um%2016.04

 

When Lime2 has to also power BPi M2+ then input voltage drops down to 5.0/5.01V, if not then it's 5.1V. So when being able to query PMIC for input voltage it's quite easy to spot power related problems. I still have my favourite 'crappy USB cable' lying around to demonstrate how moronic Micro USB as power connector is: http://forum.lemaker.org/forum.php?mod=viewthread&tid=8312&extra=page%3D1

Link to comment
Share on other sites

"Unfortunately not with H3 boards."

 

If I anderstand well, H3 boards have no voltage monitoring hardware. So I should find an external hardware solution.

 

How do you retrieve PMU stats ? Perhaps could you power an A20 board threw an H3 device to check power stability ?

 

Thanks for your works anyway : I got the first usable distro for BPI-m1 as desktop with armbian (ubuntu) and my bpi m2+ will be ready for production in a few days as firewall/gateway/ap (customized gentoo with armbian kernel/uboot)

 

N.B. I use ap62xx for wifi and external dongle for bt since bluez is (one more time) buggy - in order to use external adapter, disabling bt in script.bin is necessary.

Link to comment
Share on other sites

If I anderstand well, H3 boards have no voltage monitoring hardware. So I should find an external hardware solution.

 

How do you retrieve PMU stats ?

 

AXP209 used with A10, A13 and A20 provides access through I2C (same with any other PMU from X-Powers used together with other Allwinner tablet SoCs like A31/A31s, A83T, A64 and so on -- but there the drivers to readout information pretty much suck).

 

Some details: http://forum.armbian.com/index.php/topic/1631-tutorial-marriage-between-a20-and-h3-ups-mode-sunxi-pio-utility/?p=12714

 

So what can be read out depends on the board and powering method in question. And vendors providing H3 boards could of course add voltage monitoring hardware... nobody would buy later since the selling points of these boards are IMO: Fast enough (quad core at 1.2/1.3GHz), many real USB ports, HEVC video decoding HW accelerated and... being dirt-cheap.

Link to comment
Share on other sites

The board is not idle when I measure 4.76V. It has attached an RTLSDR key and decoding ADSB packets with main process lying at 55% of CPU :)

 

I'm going to try the stress and measure the voltage again.

 

EDIT:

 

Done the measurements:

 

With the same load as stated before + stress -c 4  the voltage is lying at 4.71-4.72V

 

I'm leaving the stress on for some time 

 

EDIT2:

 

The board is able to run 4h with stress -c 4 without any issue.

If you want plug some usb device to your board you are out of range 4,75-5,25V what usb specification says about voltage. I doesn't mean 4.75V is OK. Your board should rather stay between 5,0-5,1V to guarantee stability.

Link to comment
Share on other sites

Guest
This topic is now closed to further replies.
×
×
  • Create New...

Important Information

Terms of Use - Privacy Policy - Guidelines