[SOLVED] [...ish?] weird i2c errors, also cpu voltage warnin

Moderators: adafruit_support_bill, adafruit

Please be positive and constructive with your questions and comments.
Locked
tintar
 
Posts: 24
Joined: Tue Jun 24, 2014 2:00 pm

[SOLVED] [...ish?] weird i2c errors, also cpu voltage warnin

Post by tintar »

UPDATE: maybe solved?
the combination of removing the stray HDMI cable, and keeping the governor on "performance", seems to have removed all of the messages about i2c and about tps65217 and voltage.

hm, now that I'm stuck on the idea of hdmi, looking back through logs there are some odd messages about virtual cape slot 6.




so, I don't use any i2c on my bbb. its only connections/peripherals are:

- 5v@2a power brick
- ethernet
- powered 4 port usb hub, with
--- 64gb flash
--- bluetooth 0a5c:21e8
--- gsm modem 12d1:1506
--- R820T radio receiver 0bda:2838

right now, the only thing permanently running on it is rtl-sdr/1090 in a screen session
http://www.satsignal.eu/raspberry-pi/dump1090.html
which averages a load of ~1 at 1ghz, but load is more 6-8 at 300mhz. (I know traditional load concept is somewhat meaningless on android/ARM, but no clue if the same is true for debian/ARM?)

so... ok: apologies, I have tried searching the adafruit forums but no luck.
I did find this:
https://groups.google.com/forum/#!topic ... DOVeiaGCQk

and have all the exact same symptoms/warnings as the people there. it sounds like simply nuisance warnings. but I'd like to be sure there's nothing to worry about.

with the default ondemand governor, I get the following set of lines once every 1 second, maybe even faster than that:

[ 1771.612760] tps65217 0-0024: Read from reg 0xf failed
[ 1772.620214] omap_i2c 44e0b000.i2c: controller timed out
[ 1772.630027] tps65217 0-0024: Read from reg 0x11 failed
[ 1772.635943] cpufreq_cpu0: failed to scale voltage down: -110
[ 1774.004886] omap_i2c 44e0b000.i2c: controller timed out
[ 1775.012790] omap_i2c 44e0b000.i2c: controller timed out
[ 1776.022573] omap_i2c 44e0b000.i2c: controller timed out
[ 1776.032712] dummy 0-0034: Error -110 reading from cec:0xfe

(apologies if that is too much log text spam, please let me know and I'll use BANNED)

cpufreq shows the cpu spending about 60% of time at 300mhz, 40% at 1000mhz. 600mhz & 800mhz, both almost literally zero.

if I change the governor to performance, the tps65217 and voltage warnings stop. but the i2c / dummy-error continue.

at least 1 person in the google thread said that constantly running at 1ghz eliminated the i2c warnings. for me, all it does is slow them from appearing every 1 second, down to every 10 seconds.

I also tried the suggested "while true; do i2cdetect -y -r 0; done > /dev/null" from the thread, which they said (mostly)eliminated the messages. in my case it caused them to increase in amount and also in speed, ha ha!

I know very little about i2c, but it sounds like the messages may be due to no resistors on i2c0 and it's not possible to add them. i.e. not defective or dying hardware. if that's the case, I won't worry, and maybe stop the messages from spamming the logs.

I do worry a little about heat and electricity usage. running ondemand, the am335x-bandgap-isa-0000 is mostly 50dC. in performance usually ~61dC.
Last edited by tintar on Sat Jul 05, 2014 3:46 pm, edited 1 time in total.

firobug
 
Posts: 2
Joined: Sat Jul 05, 2014 9:43 am

Re: weird i2c errors, also cpu voltage warnings

Post by firobug »

im a total newbie but i just thought maybe have you looked over your code....did you simply paste a full copy of some library into your project and have it doing some test or operation to some item that is not in your build? it sounds to me like the device is trying to send signal to some sensor like maybe a 360o servo that it is anticipating feedback from one of those little slot reading thingeys....sorry terminology fails me at the moment....if this is the cause it could be as simple as removing that bit of code and your are good on that part....to tell you honestly i came here trying to find out what the safe threshold on the device was supposed to be in reguards to current as i know it says +5 volts and recomends a 9v battery....how much juice can i hook up to the uno is what im trying to find out.....i would like to power something for a long durration and have an 8aa battery clip from radioshack that has a 9v plug with nihm id put peaks at about 12v with running at about 9.6 and some modulation/studdering....im thinking of making a cnc/home 3d printer out of this thing but im not trying to fry this thing because i know i dont know what im doing....i just saw your post and from the way i read it it sounds to me like its trying to fetch some info that it is not recieving a response for and a library issue is my best guess if true and i helped you there please let me know because im totally unaware myself if this would actually cause issues(useless/unused code)so i would like to know for my own self and if you or anyone else can help me answer my question because i cant seem to find any myself....and while what im seeming to see in the texts that while factors may need concidered as to what all i have in my build how much current the total resistance totals im still to new to this thats why i got a beginners kit but there is totally not enough info put out on this kind of stuff to answer all my questions and i dont know where to begin....it has been hell for me to get this far into the site this far...i mean i used grease monkey to enable the text entry fields to try getting to customer support.....now i created an account and i cant see how to post my own private question to this site....i dont know where to start looking for support but hope i was able to help you and hope maybe i can get help myself

tintar
 
Posts: 24
Joined: Tue Jun 24, 2014 2:00 pm

Re: weird i2c errors, also cpu voltage warnings

Post by tintar »

thanks for the ideas. but --

apologies, I neglected to mention that I get the errors regardless of whether the 1090 software is running. and regardless of if I have any of the usb devices present or not.

rtl-sdr libraries do have some i2c support in them, but everything I'm doing is 100% usb and tcp. but I don't disagree, it's possible.

but: already tried unplugging every single usb item, even the hub. killed the 1090 software. killed transmission-daemon. basically the only 2 things not from the factory debian image.

still the same messages. then I had an odd thought - originally I had assumed that this line

Code: Select all

dummy 0-0034: Error -110 reading from cec:0xfe
all of cec:0xfe was a hex address.
but then I wondered, what if cec was referring to HDMI CEC? hdmi actually does have i2c running over it.
and then, the fact I had the HDMI cable connected to BBB but the other end connected to nothing (not enough ports on my TV and I don't need a screen anyway, ssh is sufficient)
this cable is around a lot of other electronics - could it be simply picking up a lot of noise and confusing the BBB?

I could be completely wrong about the cec == CEC part, and simply created a story in my head. but as soon as I unplugged the HDMI cable, the messages stopped! maybe only coincidence.

even after reconnecting usb hub and all usb devices, restarting 1090 software, restarting transmission - ZERO MESSAGES. hurray!

Locked
Please be positive and constructive with your questions and comments.

Return to “Beagle Bone & Adafruit Beagle Bone products”