Pi 3 shutting down-ish

it happens every time on my Raspberry Pi 3B (with the same sd card), sometimes after more than an hour, sometimes after 10-15 minutes

That is mostly what I find, it can happen commonly quite soon after start up.

Edit Here was a Pi ROM build in which gamma was not enabled in BCMVideo 0.50. It has been removed as a request has been made for further testing with the “official” beta ROM. It had been good here for a couple of days on the RPi3B+.

OS5.24 is near, that will sort out the problem one way or the other.

Hi David

So far so good, with the Pi rom build you mentioned, my RPi3 is running fine for hours now.

Thank you!

François

Update: 3 hours later, still no issue! The experience on RPi 3B is so better that RPi 1 :)

If some of the other downloaders could post back with success/failure reports then that could be beneficial.

I had some problems with my Pi3B+ some time ago. At the time, I thought it was just overheating, so I put it away and used a Pi2.

Having read this post in detail, I now think I can report having had the same problem with a Pi3B+

Here is a Pi ROM build in which gamma is not enabled in BCMVideo 0.50.

This afternoon I downloaded this ROM and the Pi3B+ has been stable for a few hours, although still running noticeably hotter than a Pi 2.

It would be useful if anyone who thinks they’ve seen the screen blanked issue to

1. Go back to the 12th April beta ROM with gamma enabled; there’s not going to be any progress if nobody’s running the accused ROM
2. Update to the 9th April Pi firmware which I’ve grabbed from github, this seems to be better but I’m only a sample of 1
3. Test it
4. Report back what happens (along with roughly what you were doing at the time, how long it took, which model of Pi)

Hello Sprow

I can tell you that with Beta Rom of 8-Apr-2018 and the latest firmware I also got from github, i had the issue.
Every session ends up with this issue, even when doing nothing special.
Raspberry Pi 3B
On the Raspberry Pi 1B, the issue never occurs.

Regards

I can tell you that with Beta Rom of 8-Apr-2018

Please follow all of steps 1-3 in my request. Don’t jump straight to step 4.

Hi Sprow

Here we go, I am exactly in the configuration you requested, even I am sure I was already in this configuration before I use the Rom mentioned by David Pitt.
So, we will see if sth wrong happens within the next hour.

Reagards

Hi Sprow

My RPi 3B is running for a couple of hours now with the latest DEV build and the firmware you provided…
All is fine up to now, which already seems an improvement…

I will leave my RPi running the whole morning to confirm the fix.

Regards

Franços

My experience of these shutdowns is that though they can occur soon after start up it is also possible to go for days without one happening at all. Any notions such as the fault always appears within 12 hours is not so here.

I built a gamma disabled ROM to confirm that that cleared the fault. On the evidence so far that does seem to be true, no one has reported a shutdown that is (yet?).

A question has been raised with the Raspberry Pi developers which has so far not yielded a result.

Firmware has been updated to the latest throughout.

P.S. The current beta ROM and latest firmware are on the case here now. The one piece of evidence I would like to try to grab is the chip temperature at shutdown (via vnc on the iMac). I have the impression that the chip is hotter than usual immediately after a shutdown.

As a request for further testing has been made using the official beta Pi ROM and the latest firmware, the gamma-disabled ROM has been withdrawn.

The one piece of evidence I would like to try to grab is the chip temperature at shutdown (via vnc on the iMac).

I glued a thermocouple onto the metal can of the 3B+ processor, but the temperature looked much the same before and after the video cut out (approx. 35C above ambient, ie. 55-60C). I also heated the chip with a hot air gun to 100C and held it there for 5 minutes with no ill effect, so if there is a thermal issue then it’s not accelerated within that timeframe.

My experience of these shutdowns is that though they can occur soon after start up it is also possible to go for days without one happening at all.

The testing is very linear in nature. Prior to the combination I’m hoping people will test, the apps I loaded each time caused the video to cut out in 4h-12h kind of window. With the combination it’s been > 3 days, but the more people can run the test in parallel the better the odds of understanding are.

but the more people can run the test in parallel the better the odds of understanding are.

Bingo, the RPi3B+ has just obliged with a shutdown after about 6 hours. The Pi continued to work over vnc until attempting to open a StrongED TaskWindow from the TaskManager iconbar menu. That stiffed the Pi completely, and this is something I have seen before.

There was nothing untoward about the measured temperature, 55℃. I have had the feeling that after a shutdown the processor feels a lot hotter than usual, as digitally measured with an uncalibrated finger. Removing the Pi case lid seems to reduce, or stop, the shutdowns and this one happened with the lid on. Hmm!!

CTRL-Break restarted the Pi.

Latest ROM and firmware.

*fx0
RISC OS 5.23 (12 Apr 2018)
*
bootcode.bin  09Feb18
fixup.dat     09Apr18
start.elf     09Apr18

Meanwhile, my Pi’s steadfastly refuse to crash. I’ve had a Pi 3B+ running Sprow’s tests for 54 hours with the April 9th firmware, and before that I’ve had a Pi 3 running for 20 hours running older firmware, along with various other Pis running for various other amounts of time (generally many hours/days until I get bored). But I have definitely seen the crash at least two or three times in the past.

Hi

My RPi 3B has been running for 12 hours without problem.

Cheers

François

Looking OK so far running with Sprow’s suggestions on my Pi 3B. Ran about 5 hours Sunday evening and 14 hours so far yesterday/today.

Unless it falls over in the next few minutes, I’ll leave it running overnight.

Mine’s lasted 24:48 so far, although as noted that doesn’t guarantee that it’s fixed. I’ll certainly keep it running (at least until 5.24 is released) and will post back if it fails.

I’m quite happy to run the system for weeks on end, so we’ll see what happens!

Mine did not survive overnight :(

I’m using the 9/4 firmware and 12/4 ROM as linked in Sprow’s post. It’s a Pi 3 (not a 3+), I have hdmi_blanking enabled in config/txt and I was doing absolutely nothing when it failed (I only had NetSurf, Edit and maybe one or two other things running). I woke up this morning, moved the mouse, and the screen didn’t come back on.

Has managed to run 38 hours OK. Now shutting down; not sure how much of a run I’ll be able to manage later today.

No luck

The issue occured again this morning, less than 20 minutes after switching on my RPi3B. I was doing nothing, but taking my breakfast.

:/

Regards

Which version do you have? The problem shouldn’t occur with the 18-Apr-2018 version of BCMVideo (which is still numbered 0.50).

No luck

Some luck here, but that is all it is, just luck I think.

The RPi3B+ did shutdown four days ago but there have been no further occurrences since with the latest ROM and firmware.

*fx0
RISC OS 5.25 (17 Apr 2018)
*
bootcode.bin  09Feb18
fixup.dat     16Apr18
start.elf     16Apr18

The issue occured again this morning, less than 20 minutes after switching on my RPi3B. I was doing nothing, but taking my breakfast.

I had several days of good luck, then decided to roll the dice and power cycle, whereupon the Pi3B+ video dropped out again after a couple of hours. So there’s certainly some random element…

Some luck here, but that is all it is, just luck I think.

Interestingly I note there’s another new set of firmware where the SDRAM frequency has been reduced. Perhaps that’s important for our problem too?

In the interests of science, we should take care to change only 1 thing at once. So, using the same RISC OS ROM and firmware as I pointed to before but adding the line

sdram_freq=450

Interestingly I note there’s another new set of firmware where the SDRAM frequency has been reduced. Perhaps that’s important for our problem too?

That applies to the RPi3B+, we have the problem on the RPi3 as well. The default sdram_freq on the RPi3B is already 450

My hopes were here :-
https://github.com/raspberrypi/firmware/issues/971

My run of luck ran out a couple of times yesterday. First after about an hour whilst I was was trying to create a new directory via the filer. Second time after about two hours during web browsing.

After seeing Sprow’s post from earlier today I added sdram_freq=450 to my CONFIG.TXT file. It lasted about 35 minutes before falling over.

OS 5.24 on the Pi is gamma disabled and so should not shutdown inconveniently.

OS 5.25 is gamma enabled.