psmedley
Posts: 10
Joined: Mon Aug 13, 2012 9:08 am

2x failed Pi 4's

Fri Nov 01, 2019 9:01 am

Hi All,
Looking for some help.

Ordered a Pi 4 from Pimoroni back in August. Set it up in a Flirc Pi4 case with an office power supply, and within 12 hours it had failed and wouldn't boot (orange and green LEDs lit constantly). Tried all the recovery options, nothing changed the LED status. Checked pin voltages and the 3.3v pin was < 2 volts. Pimoroni replaced under warranty and all was well.

Fast forward to this week, and the replacement Pi has failed in a very similar manner, asked weeks of running fine. Same issue with both orange and green LEDs lit constantly.

Both failures with the Pi 4 in the Flirc case, and with the official Pi 4 power supply.

1st failure, only USB peripheral was an unpowered USB hub + Aeotec zwave adapter
2nd failure: USB SSD, USB HDD + external powered USB case with 2x HDD in it.

Am I just _really_ unlucky or does anyone have any other tips?

Cheers,

Paul

jamesh
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 25442
Joined: Sat Jul 30, 2011 7:41 pm

Re: 2x failed Pi 4's

Fri Nov 01, 2019 9:31 am

I suspect just very unlucky. Completely different setups.

Wasn't that ZWave adapter the one with the odd USB fault on it? Hence the hub?
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Contrary to popular belief, humorous signatures are allowed. Here's an example...
“I own the world’s worst thesaurus. Not only is it awful, it’s awful."

psmedley
Posts: 10
Joined: Mon Aug 13, 2012 9:08 am

Re: 2x failed Pi 4's

Fri Nov 01, 2019 9:24 pm

Wasn't that ZWave adapter the one with the odd USB fault on it? Hence the hub?
Correct. Without the USB 2.0 hub it wasn't recognised by the Pi 4.

HypnoToad
Posts: 77
Joined: Thu Dec 11, 2014 6:00 pm

Re: 2x failed Pi 4's

Fri Nov 01, 2019 10:37 pm

psmedley wrote:
Fri Nov 01, 2019 9:01 am
2nd failure: USB SSD, USB HDD + external powered USB case with 2x HDD in it.
I'm running pretty much this exact same setup, without the external powered USB case with 2x HDD in it.

I've found you don't need to do the bolts on the FLIRC case up that tightly to get good heatsink contact.

I think that if you overtightened these bolts, it's possible to damage the case or even the Pi4 PCB.

A bit of a long shot maybe, but certainly a possibility.

renice123
Posts: 165
Joined: Tue Oct 15, 2019 3:56 pm

Re: 2x failed Pi 4's

Sat Nov 02, 2019 9:36 am

Please tell us about your power source. My friends and I are observing critical error messages of version 4 and almost always get messages that they used “official” power sources. It is so amazing that it interested me. I do not use an official power source, so I can not check it. However, the forums write about quality of an official 3 A power source (power noise, etc.).
Could you clarify the following.
1. Did you use only an official power source (5 V 3 A) in both cases?
2. You cannot check with an oscilloscope how “clean” the output signal from your power source is? Does the power supply hold a load (over 2 A) and how stable is the voltage under load?
3. Do you often have occasional blackouts in your home? Are there any other power-consuming devices in the house connected (washing machine, electric oven, etc.)
4. What is the average temperature in your home?
5. I do not know whether it is possible to disassemble the power supply without spoiling it, but I would really like to look at the circuitry of the power supply.
I would be grateful for the information. Here are the previous posts
https://www.raspberrypi.org/forums/view ... 8&t=251968

User avatar
davidcoton
Posts: 4666
Joined: Mon Sep 01, 2014 2:37 pm
Location: Cambridge, UK

Re: 2x failed Pi 4's

Sat Nov 02, 2019 10:31 am

renice123 wrote:
Sat Nov 02, 2019 9:36 am
I do not know whether it is possible to disassemble the power supply without spoiling it
If it's your PSU, you can of course dismantle it. You are unlikely to be able to reassemble it to a safe state. I know, I've tried to repair accidentally damaged PSUs. So assume that the investigation will cost you a new PSU.

NB Do NOT use PSUs or other mains equipment which is held together with tape or other improvised repairs, and do NOT add additional metal screws into plastic mains equipment housings.
Signature retired

renice123
Posts: 165
Joined: Tue Oct 15, 2019 3:56 pm

Re: 2x failed Pi 4's

Sat Nov 02, 2019 12:48 pm

If two Raspberries purchased at different times from different suppliers are out of order, then the first thing I would think would be improper operation or a poor-quality power supply. Therefore, in this case, I immediately checked the power supply (even if this power supply is recommended). Or at least measured voltages and currents in all modes (especially under load).
I live in a big city and use a lot of electronic devices. Every second power supply, even from well-known brands (Nikon, Sony, Sharp & Marantz & etc), seems to be a not very high-quality Chinese power supply and I prefer to replace them than pay a double price for a possible malfunction due to a bad power supply.
This became my habit and I immediately buy a high-quality (albeit expensive) power supply for a new electronic device. For Raspberry, I use “native” old power supplies from Blackberry 2 A (this is not charging!), They are convenient, reliable and work well. Now I am also thinking of buying a power supply unit 10–15 A from the Japanese company Murata, a small size and high-quality industrial power supply unit, which is well-known in amateur radio circles for Raspberry. Yes, it will cost more than a Chinese exercise (ha! For a few dollars!), But it will give me pleasure and I will be sure that in the future the new Raspberry will work well.
This is my opinion and my hobby, I am not trying to impose my views on anyone.

psmedley
Posts: 10
Joined: Mon Aug 13, 2012 9:08 am

Re: 2x failed Pi 4's

Sun Nov 03, 2019 7:43 am

HypnoToad wrote:
Fri Nov 01, 2019 10:37 pm
psmedley wrote:
Fri Nov 01, 2019 9:01 am
2nd failure: USB SSD, USB HDD + external powered USB case with 2x HDD in it.
I'm running pretty much this exact same setup, without the external powered USB case with 2x HDD in it.

I've found you don't need to do the bolts on the FLIRC case up that tightly to get good heatsink contact.

I think that if you overtightened these bolts, it's possible to damage the case or even the Pi4 PCB.

A bit of a long shot maybe, but certainly a possibility.
I had a similar thought, but wondered if the bosses we're quite level, such that when installed, it was putting some torsion on the board. I checked - and the flirc case seems pretty 'flat' if that makes sense - with all the bosses being in the same plane.

Pretty confident i didn't over-tighten, I was conscious of the fact there's a PCB in there!

psmedley
Posts: 10
Joined: Mon Aug 13, 2012 9:08 am

Re: 2x failed Pi 4's

Sun Nov 03, 2019 7:45 am

renice123 wrote:
Sat Nov 02, 2019 9:36 am
Please tell us about your power source. My friends and I are observing critical error messages of version 4 and almost always get messages that they used “official” power sources. It is so amazing that it interested me. I do not use an official power source, so I can not check it. However, the forums write about quality of an official 3 A power source (power noise, etc.).
Could you clarify the following.
1. Did you use only an official power source (5 V 3 A) in both cases?
2. You cannot check with an oscilloscope how “clean” the output signal from your power source is? Does the power supply hold a load (over 2 A) and how stable is the voltage under load?
3. Do you often have occasional blackouts in your home? Are there any other power-consuming devices in the house connected (washing machine, electric oven, etc.)
4. What is the average temperature in your home?
5. I do not know whether it is possible to disassemble the power supply without spoiling it, but I would really like to look at the circuitry of the power supply.
I would be grateful for the information. Here are the previous posts
https://www.raspberrypi.org/forums/view ... 8&t=251968
1. Yes - using the 'Official' pi supply in both cases - https://shop.pimoroni.com/products/rasp ... 1160737875
2. Sorry, no oscilloscope here
3. no blackouts here - we have a Tesla Powerwall 2 and large solar system installed
3. 20 - 23C
5. It's a sealed unit....

psmedley
Posts: 10
Joined: Mon Aug 13, 2012 9:08 am

Re: 2x failed Pi 4's

Sun Nov 03, 2019 7:48 am

renice123 wrote:
Sat Nov 02, 2019 12:48 pm
If two Raspberries purchased at different times from different suppliers are out of order, then the first thing I would think would be improper operation or a poor-quality power supply. Therefore, in this case, I immediately checked the power supply (even if this power supply is recommended). Or at least measured voltages and currents in all modes (especially under load).
Well given both failures were in the Flirc case with the same power supply, I'm likely to not use either going forwards....

I have a powered USB Hub on order, should be here on Tuesday, when I rearrange cabling for that, I'll switch over the power supply - just to be safe.

renice123
Posts: 165
Joined: Tue Oct 15, 2019 3:56 pm

Re: 2x failed Pi 4's

Sun Nov 03, 2019 8:37 am

thank! I hope the next raspberry will work well for you!

HypnoToad
Posts: 77
Joined: Thu Dec 11, 2014 6:00 pm

Re: 2x failed Pi 4's

Wed Nov 06, 2019 1:19 am

psmedley wrote:
Sun Nov 03, 2019 7:43 am
I had a similar thought, but wondered if the bosses we're quite level, such that when installed, it was putting some torsion on the board. I checked - and the flirc case seems pretty 'flat' if that makes sense - with all the bosses being in the same plane.

Pretty confident i didn't over-tighten, I was conscious of the fact there's a PCB in there!
Hmm, in that case I really have no idea why they failed.

All I can say is that my RPi4 has been fine in the FLIRC case, has currently an uptime of 12 days, and no issues for me so far, you may have been unlucky.

One more thing to check is the voltage from the PSU during use, just check it's between 5-6V.

wildfire
Posts: 581
Joined: Sat Sep 03, 2016 10:39 am
Location: Dundee, Scotland

Re: 2x failed Pi 4's

Wed Nov 06, 2019 8:28 am

HypnoToad wrote:
Wed Nov 06, 2019 1:19 am
One more thing to check is the voltage from the PSU during use, just check it's between 5-6V.
6v could kill the pi, check for 4v75 to 5v25
E8 85 A2 64 C9 64 81 94 64 81 95 64 89 84 89 96 A3
Still NF Shirls

renice123
Posts: 165
Joined: Tue Oct 15, 2019 3:56 pm

Re: 2x failed Pi 4's

Wed Nov 06, 2019 8:31 am

Unfortunately, just checking the switching power supplies does not work.
In our city, many are fond of radio electronics and in my youth I was also a radio amateur (but I am a programmer), so I know how often inconspicuous errors occur.
So, a switching power supply can work well without a load, but with a load, the voltage (hence the current strength according to Ohm's law) will “float” (change randomly). Switching power supplies may “make noise”, overheat, capacitors may “dry out” over time, relatively expensive low ESR capacitors, or even high temperature electronic components, must be used.
Sometimes the power supply network is “clogged”: if some kind of pulsed devices or devices that consume a significant amount of current (oven, air conditioning, washing machine, etc.) work in the house.
Therefore, it is best to check the operation of the power supply using an oscilloscope under a load of 2 to 3 A and at the time of on-off. Also check the quality of the mains voltage. It is possible that power surges occur in the power supply unit in some cases, and due to the circuitry, these power surges arrive at the Raspberry input - in expensive power supplies they use fast-acting suppressors and fuses (protecting against reverse polarity, overvoltage, high temperature, etc.), but cheap stabilization schemes can be primitive.

I must say that with regret and some anxiety, I see more messages about the "death" of Raspberries. So, recently in another branch it was reported that out of 6 devices, 3 computers stopped working. And in this case they used the “official” power supply. Here is this thread https://www.raspberrypi.org/forums/view ... 8&t=250183
FloFra writes: Now the next Pi4 died after 1 month (only HDMI and the network are connected - the original power supply). This one is even from another supplier ...
I really don't understand what might cause the problem.

wildfire
Posts: 581
Joined: Sat Sep 03, 2016 10:39 am
Location: Dundee, Scotland

Re: 2x failed Pi 4's

Wed Nov 06, 2019 8:38 am

renice123 wrote:
Wed Nov 06, 2019 8:31 am
I really don't understand what might cause the problem.
Neither do I, out of the dozen or so Pi's I've used over the years I've never had a failure yet (apart from a 1b which I stood on). All varying ages from several years to ~ 4 months, varying models from the zero through to 4B4's, all using official power supply and purchased from various sources.

I guess I'm just lucky.
E8 85 A2 64 C9 64 81 94 64 81 95 64 89 84 89 96 A3
Still NF Shirls

psmedley
Posts: 10
Joined: Mon Aug 13, 2012 9:08 am

Re: 2x failed Pi 4's

Wed Nov 06, 2019 8:40 am

Thanks all - to be safe, I've ordered a replacement power supply. For the safe of $A20 it's cheap insurance ;)

jamesh
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 25442
Joined: Sat Jul 30, 2011 7:41 pm

Re: 2x failed Pi 4's

Wed Nov 06, 2019 9:54 am

renice123 wrote:
Wed Nov 06, 2019 8:31 am
I must say that with regret and some anxiety, I see more messages about the "death" of Raspberries. So, recently in another branch it was reported that out of 6 devices, 3 computers stopped working. And in this case they used the “official” power supply. Here is this thread https://www.raspberrypi.org/forums/view ... 8&t=250183
FloFra writes: Now the next Pi4 died after 1 month (only HDMI and the network are connected - the original power supply). This one is even from another supplier ...
I really don't understand what might cause the problem.
I would not be too concerned about reports of failures. When you take in to account how many we are currently making per week (it in the hundreds of thousands!), the number of reports on here and elsewhere of 3 or 4 failures is statistically insignificant. In fact our failure rate is extremely low compared with industry averages. As for those people who are reporting multiple failures from a small batch, this is certainly down to something they are doing, rather than endemic failures of the Pi4 itself.
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Contrary to popular belief, humorous signatures are allowed. Here's an example...
“I own the world’s worst thesaurus. Not only is it awful, it’s awful."

FloFra
Posts: 8
Joined: Wed Sep 04, 2019 11:18 am

Re: 2x failed Pi 4's

Sat Nov 09, 2019 7:26 am

Hi,

I argee on the statistical insignificance of the failures, but they are there and something seems to be odd here

After some discussions and thinking of the externals in the cases where the pi‘s died I came to the conclusion that the issues has to be the PSU
All 3 units where connected to the same outlet (however with 2 different PSUs)
The only thing also connected is a flatscreen

I think that maybe there maybe some surges or something like the in the electrical net causing the PSU to misbehave and „fry“ the pi

However in my oppionion the OFFICIAL PSU should prevent such problems (I‘d rather have the PSU die than the raspberries)

I‘ve now put a Cubieboard at this location to see of that survives ;)

Does anyone know how I can check the dead PI in terms of what got killed? (Nothing via UART while power on)

Regards
Florian

jamesh
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 25442
Joined: Sat Jul 30, 2011 7:41 pm

Re: 2x failed Pi 4's

Sat Nov 09, 2019 8:34 am

FloFra wrote:
Sat Nov 09, 2019 7:26 am
Hi,

I argee on the statistical insignificance of the failures, but they are there and something seems to be odd here

After some discussions and thinking of the externals in the cases where the pi‘s died I came to the conclusion that the issues has to be the PSU
All 3 units where connected to the same outlet (however with 2 different PSUs)
The only thing also connected is a flatscreen

I think that maybe there maybe some surges or something like the in the electrical net causing the PSU to misbehave and „fry“ the pi

However in my oppionion the OFFICIAL PSU should prevent such problems (I‘d rather have the PSU die than the raspberries)

I‘ve now put a Cubieboard at this location to see of that survives ;)

Does anyone know how I can check the dead PI in terms of what got killed? (Nothing via UART while power on)

Regards
Florian
AGreed it is odd, and also agree it probably environmental, i.e. something being done to the Pi's rather than the Pi's themselves.
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Contrary to popular belief, humorous signatures are allowed. Here's an example...
“I own the world’s worst thesaurus. Not only is it awful, it’s awful."

psmedley
Posts: 10
Joined: Mon Aug 13, 2012 9:08 am

Re: 2x failed Pi 4's

Sat Nov 09, 2019 8:55 am

FloFra wrote:
Sat Nov 09, 2019 7:26 am
I argee on the statistical insignificance of the failures, but they are there and something seems to be odd here

After some discussions and thinking of the externals in the cases where the pi‘s died I came to the conclusion that the issues has to be the PSU
All 3 units where connected to the same outlet (however with 2 different PSUs)
The only thing also connected is a flatscreen
Being an engineer, and working in quality, I'd also love to understand root cause of my 2x failures. At this stage, I mostly suspect the power supply. Whilst the Flirc case was also common to both failures, I can't see how it could have caused the failures. I've now replaced the power adapter, and am using an official Pi4 case rather than the Flirc.

Cheers,

Paul

emma1997
Posts: 591
Joined: Sun Nov 08, 2015 7:00 pm
Location: New England (not that old one)

Re: 2x failed Pi 4's

Sat Nov 09, 2019 3:57 pm

If PSU is suspect that can be checked with a scope, even $10 Ebay one. Poorly designed switchers are susceptible to a sometimes fatal startup-overvoltage design flaw.

One of my recent buck prototypes showed a 21v pulse on the 5v rail every time powered up. Yikes! Unlikely to damage anything due to duration/energy but still... redo that layout! I have not seen this with 'store bought' modules, even cheepie Chinese. A few may have a minor bump at startup due to feedback loop but nothing serious in the ones I checked.

Like I said easy to check if you are in sleuth mode.

User avatar
mahjongg
Forum Moderator
Forum Moderator
Posts: 12665
Joined: Sun Mar 11, 2012 12:19 am
Location: South Holland, The Netherlands

Re: 2x failed Pi 4's

Sat Nov 09, 2019 8:41 pm

emma1997 wrote:
Sat Nov 09, 2019 3:57 pm
One of my recent buck prototypes showed a 21v pulse on the 5v rail every time powered up.
If you have equipment that can show the pulse, the pulse is long enough to destroy equipment.

It doesn't need a second or so, it will happen in a billionth of a second.

emma1997
Posts: 591
Joined: Sun Nov 08, 2015 7:00 pm
Location: New England (not that old one)

Re: 2x failed Pi 4's

Sat Nov 09, 2019 10:48 pm

Scope was a 60ghz Lecroy so able to catch quite small pulses. Typically picosecond or less.

I agree on the timeframe. In a previous life working in wafer fab I learned that things can happen very fast. Depending on type of abuse it's possible to damage a junction quickly. Not so much in cases of heating which can take seconds, minutes, or even days, but typically in the realm of microseconds for avalanche.

Also depends on IC layout and how often too. They don't always drop dead right away, just get worse over time. In any case common glass fuse ain't likely to protect anything down there.

However a lesser known secret is just plugging/unplugging cables can generate HUGE reflected pulses many orders of magnitude greater than what I saw on that screen. Hundreds of volts. What I'm talking about is not just voltage but total energy content. Whether it's ESD breakdown or thermal EOS (term not coined yet when I was there) things can happen very fast. Fortunately nearly all modern chips have clamping diodes, intentional or parasitic, which do take lots of abuse of both types.

Basically it's about coulombs, how many and for how long. Unlikely the the circuit I was working on would cause any trouble. But it was a sign of bad ground/power plane layout so I fixed it with some trace relocation and better pour.

I'm always surprised how well these cheap Ebay products perform. Probably because most are just clones of mfg reference design. Also surprising how bad some can be regardless of provenance.

Return to “Troubleshooting”