Page 1 of 1

My 1st ever dead Pi.

Posted: Fri Nov 23, 2018 1:42 am
by DavidS
I finally had a Raspberry Pi die, about half an hour ago. It is the only BCM2837 based RPi I have, my Raspberry Pi 3B. I have not had it very long either, I got it about two months after the RPi 3B was released, so it is only about 2 and a half years old, making it my newest RPi.

All of my RPI B and B+ models are still going, lost my BCM2836 based RPi 2B during a move.

I do like the the newer BCM2837 RPi's seem to run cooler by all reports. So when money allows I may get a 3B+. Though I will definitely be getting a few more B+ boards as they have more than proven themselves many times over.

Re: My 1st ever dead Pi.

Posted: Fri Nov 23, 2018 3:35 am
by DavidS
This definitely highlights how good the Raspberry Pi is, only having one out of 15 die in using a system for this many years is pretty good for sure. I do not think I could say that about any other system I have ever had (while some last a long time, if you put 15 of them together the chances of having more than one dead is fairly likely in the first 4 years).

Re: My 1st ever dead Pi.

Posted: Fri Nov 23, 2018 7:59 am
by Heater
I guess it could not handle all that million digit Fibonacci calculation :)

Re: My 1st ever dead Pi.

Posted: Fri Nov 23, 2018 9:53 am
by gordon77
Any cause of failure identified?

Re: My 1st ever dead Pi.

Posted: Fri Nov 23, 2018 1:12 pm
by DavidS
gordon77 wrote:
Fri Nov 23, 2018 9:53 am
Any cause of failure identified?
Nope. All I know is that a few days ago it started to show undervolt warning randomly, so I checked power supply and test points on the board, everything was good so I set the over_volatage=2 and it still showed the low voltage lightning bolt at random. The low voltage warning never lasted more than about half a second and was only maybe once or twice an hour. Then it quit, completely, no Led, no warm up of the SoC, no detectable activity at all, to be sure did try multiple SD's with multiple different OS's and different firmware releases, and even with nothing hooked in at all other than power and SD.

Even after failing the test points show as if it is working, though it is a brick.

Re: My 1st ever dead Pi.

Posted: Fri Nov 23, 2018 6:27 pm
by Burngate
DavidS wrote:
Fri Nov 23, 2018 1:12 pm
... so I set the over_volatage=2 and it still showed the low voltage lightning bolt at random.
Not sure I understand ... the overvoltage merely changes what's feeding the core, it doesn't change anything to do with the 5v rail.
Or am I just wrong?

Re: My 1st ever dead Pi.

Posted: Fri Nov 23, 2018 6:50 pm
by DavidS
Burngate wrote:
Fri Nov 23, 2018 6:27 pm
DavidS wrote:
Fri Nov 23, 2018 1:12 pm
... so I set the over_volatage=2 and it still showed the low voltage lightning bolt at random.
Not sure I understand ... the overvoltage merely changes what's feeding the core, it doesn't change anything to do with the 5v rail.
Or am I just wrong?
My understanding is that the low voltage warning comes from the core voltage going to low. Or has something changed?

The 5v rail was not the issue, as that was slightly above 5v (and well within safe limits). So the issue had to be the voltage at the core, or am I missing something.

Re: My 1st ever dead Pi.

Posted: Fri Nov 23, 2018 7:06 pm
by davidcoton
DavidS wrote:
Fri Nov 23, 2018 6:50 pm
My understanding is that the low voltage warning comes from the core voltage going to low. Or has something changed?
The low voltage warning is and always was associated with the 5V supply rail only. It indicates that the supply is below (IIRC) 4.65V after the polyfuse. Of course a false warning could be associated with some impending failure mode.

Re: My 1st ever dead Pi.

Posted: Fri Nov 23, 2018 9:20 pm
by DavidS
davidcoton wrote:
Fri Nov 23, 2018 7:06 pm
DavidS wrote:
Fri Nov 23, 2018 6:50 pm
My understanding is that the low voltage warning comes from the core voltage going to low. Or has something changed?
The low voltage warning is and always was associated with the 5V supply rail only. It indicates that the supply is below (IIRC) 4.65V after the polyfuse. Of course a false warning could be associated with some impending failure mode.
Ok so I have always had it wrong. And what does the low voltage warning mean when the 5V test points on the board (after the poly fuse) are between 5.0 and 5.12 volts?

I am updating my notes, as I had noted it wrong in 2012, having noted that it was the core voltage.

Re: My 1st ever dead Pi.

Posted: Fri Nov 23, 2018 9:42 pm
by drgeoff
DavidS wrote:
Fri Nov 23, 2018 9:20 pm
And what does the low voltage warning mean when the 5V test points on the board (after the poly fuse) are between 5.0 and 5.12 volts?.
it means the warning circuitry is up the swanee.

Re: My 1st ever dead Pi.

Posted: Fri Nov 23, 2018 10:08 pm
by PeterO
drgeoff wrote:
Fri Nov 23, 2018 9:42 pm
DavidS wrote:
Fri Nov 23, 2018 9:20 pm
And what does the low voltage warning mean when the 5V test points on the board (after the poly fuse) are between 5.0 and 5.12 volts?.
it means the warning circuitry is up the swanee.
Or that the low voltage is so short that the measuring device doesn't notice it !

PeterO

Re: My 1st ever dead Pi.

Posted: Sat Nov 24, 2018 2:43 am
by DavidS
Call me confused. Just to see if I could figure anything out I hooked the dead RPi 3B up to a monitor, keyboard, mouse, osciloscope, and active sampling voltometer. Then I inserted a backup copy of my RISC OS boot SD card, and a current copy of my RISC OS boot SSD HDD.

The end result is that I am currently typing this from RISC OS running on that Raspberry Pi, I have unhooked the scope and voltometer, and decided to just see how long this lasts.

Call me confused, what could cause this type of behaviour. I have yet to see a low voltage lightning bolt on the screen since it came up.

So I was wrong, so far. I still do not have a dead RPi.

Though I have added a Raspberry Pi 3B+ or 5 to my shopping list as soon as I have enough money to get one or more. With this RPi still running (after acting like it died), and the 3B+ running cooler, I am very much interested in the 3B+ (even though I will likely clock it the same as I have this one for over a year, arm_freq=700, arm_freq_min=500, core_freq=250, only going up to full speed for testing things).

Re: My 1st ever dead Pi.

Posted: Sat Nov 24, 2018 3:41 am
by Tzarls
You must have tripped your Pi's polyfuse, maybe when testing voltages. Now it has reset, that's why it is working again. I had something like this happen once to one of my Raspis, one that was doing almost nothing, just displaying photos over and over. One day it was dead. After a coupe of days it was back and it is still working after several months. I suspect some sneakyi fingers must have done something they shouldn't have...

Re: My 1st ever dead Pi.

Posted: Sat Nov 24, 2018 8:30 am
by rpdom
DavidS wrote:
Sat Nov 24, 2018 2:43 am
Though I have added a Raspberry Pi 3B+ or 5 to my shopping list as soon as I have enough money to get one or more. With this RPi still running (after acting like it died), and the 3B+ running cooler, I am very much interested in the 3B+ (even though I will likely clock it the same as I have this one for over a year, arm_freq=700, arm_freq_min=500, core_freq=250, only going up to full speed for testing things).
Don't forget that the 3B+ has better thermal management and throttling than the 3B has. It has three clock speeds instead of two. It runs at 1400MHz when loaded, but not hot at all, then drops to 1200 (the 3B's top speed) if it reaches the low trip temperature (65 degrees C, I think), then drops to the current minimum (700) if it still gets hotter. The whole board gets warmer, especially the metal of the various ports as they are being used as an extra heatsink.

Re: My 1st ever dead Pi.

Posted: Sat Nov 24, 2018 2:30 pm
by DavidS
rpdom wrote:
Sat Nov 24, 2018 8:30 am
DavidS wrote:
Sat Nov 24, 2018 2:43 am
Though I have added a Raspberry Pi 3B+ or 5 to my shopping list as soon as I have enough money to get one or more. With this RPi still running (after acting like it died), and the 3B+ running cooler, I am very much interested in the 3B+ (even though I will likely clock it the same as I have this one for over a year, arm_freq=700, arm_freq_min=500, core_freq=250, only going up to full speed for testing things).
Don't forget that the 3B+ has better thermal management and throttling than the 3B has. It has three clock speeds instead of two. It runs at 1400MHz when loaded, but not hot at all, then drops to 1200 (the 3B's top speed) if it reaches the low trip temperature (65 degrees C, I think), then drops to the current minimum (700) if it still gets hotter. The whole board gets warmer, especially the metal of the various ports as they are being used as an extra heatsink.
I had missed that it has 3 clock speeds for throttling. Still going to likely set the speed down. It is amazing how much faster an Cortex A53 (Pi 3B/B+) is at 700MHz than a ARM1176JZF-S is at 1100MHz. Even only using a single core in each case (RISC OS). And now that RISC OS is beginning to support SMP, the future looks bright.