Pi 3 is Destroying SD Cards

[MarkDaniels]

I have 2 Pi 3 boards. One has been configured and run for several days trouble free. In each case the SD card was updated in a Pi 2 and transferred to the new Pi 3. The hardware setup for each Pi is virtually identical. However, on initial power up of one of them, the LEDs lit briefly then went out. The SD card was removed and transferred to the working Pi, which failed to boot from it and caused a PSU problem (power loss). Next, the SD card from the working Pi was put into the non-working Pi, which still refused to boot - LEDs lit up. The SD card was returned to the working Pi, which then also refused to boot, so was transferred to a Pi 2 with the same result. Inspecting the cards with GParted on an Ubuntu PC showed the first card to have a capacity of 32 MB (yes, megabytes!) instead of 32 GB. The card was getting hot. The second card was not recognised at all by GParted.

I have tried an unmodified Pi 2 SD card in the originally working Pi 3 and get a persistent colour splash and failure to complete the boot process exactly as expected, so that Pi appears to still be working. However, this Pi will no longer boot from the original power supply and has to be connected directly to a 5V 2A PSU to work. The original PSU still works and will run the Pi 2. So, clearly something has gone wrong with both Pi 3s.

EDIT: the original PSU is rated at 5 V 3 A.

I am using stock Raspbian and various hardware setups which are known to work and have been tried and tested over a very long period. There are no software configuration errors of any relevance to this problem, other than as stated in the previous paragraph. I am reluctant to test the dead Pi further with new SD cards, unless this is a known problem with a solution. Any ideas? This has me stumped.

[jlukanc]

I think you need a 5v 2.4A outlet for increased power draw for usb ports.

[MarkDaniels]

jlukanc posted:

"I think you need a 5v 2.4A outlet for increased power draw for usb ports."

Thank you for your response, but this is clearly not the issue as the Pi 3 was working from the original PSU prior to this problem. If you read through my full description of the problem you will see that the PSU is of adequate capacity.

[MarkHaysHarris777]

The most likely event here is a hardware failure in the original PSU. More likely than not it is no longer capable of providing full clean output for the PI 3, but is still able to provide enough current for the PI2. (Even though you shouldn´t do it in my opinion) you can run the PI2 on very little current (less than one amp) to boot and very lite activity. The PI3 does not appear to be so forgiving; its power hungry. The fins of my PI3 SoC heat-sink get very warm just idling (I haven´t measured the current yet, later today).

Hopefully you don´t have a power problem that is frying the SBC nor the SD; but at this point (pure speculation) I would be willing to suspect your 3A PSU as a culprit.

Good luck.

[stderr]

I have 2 Pi 3 boards. So, clearly something has gone wrong with both Pi 3s.

It is very unlikely that 2 pi R screwed. Especially given your explanation of how the second pi3 started behaving badly. Almost anything is more likely including dodgy sdcards and especially problems with power. You are saying that the power source is fine apparently due only to what the printing on the wall wart claims. Never trust that, always use a completely different known good supply.

. I am reluctant to test the dead Pi further with new SD cards,

2gig cards can be used with Jessie Lite and I have a 128meg card I'm using with Tiny Core. Any testing should be done with the least valuable cards you have.

But I think you should try to figure out what is going on with the sdcards that now don't work right. Obviously you don't want to damage anything else of value, do you have any sdcard reading device that isn't really important, like an old slow netbook you don't care that much about?

I can't image this trashing the entire thing but I recall plugging in flash drives into ports and the flash drive gets hot and the port stops working right, perhaps only for a period of time. If you think about it, the amount of current that goes into the flash drive is at least partly defined by the drive itself and not the source. If the flash drive negotiates a high current that it can't deal with and then draws that current for no reason, where is it going to go but into getting hot?

Another thing is that I was trying to reflash some flash drives and I put them into a port on a netbook and they kept claiming the drives were, I think it was 242 meg. The drives were 4 gig or 8 gig. Nothing worked right until I rebooted the netbook, which was running x86 Ubuntu 14.04 LTS. So it is possible to get incorrect size readings that clearly are coming from the host hardware/software in some sense.

In your case, the sdcards show up as 32meg no matter where you put them. Assuming that they aren't getting hot just putting them in any reader, I would use a second sdcard reader, or boot to Tiny Core Linux and use the built in one, and with dd write zeros to the drive. Then I'd remove the drive and put it somewhere and see if its size was correct again.

[MarkDaniels]

Thank you for your in-depth response. Please see my notes inserted, below, which I hope will not cause offense. If you read to the bottom you will find an explanation regarding part of the problem.

It is very unlikely that 2 pi R screwed. Especially given your explanation of how the second pi3 started behaving badly. Almost anything is more likely including dodgy sdcards and especially problems with power. You are saying that the power source is fine apparently due only to what the printing on the wall wart claims. Never trust that, always use a completely different known good supply.

I think, in this case, it is highly likely that the first Pi is "screwed". However, I think the second one was affected by the SD card that the first one had damaged (see footnote). I have used 3 (THREE) different PSUs for testing these computers: all of which are high quality ones. I only use Stontronics PSUs and have used tens of thousands of them over the past 10 years in industrial equipment which I manufacture. So, I am reasonably confident that the PSUs are up to the job, particularly when it is considered that one of those PSUs had already been running a Pi 3 continuously for nearly two weeks without problem.

2gig cards can be used with Jessie Lite and I have a 128meg card I'm using with Tiny Core. Any testing should be done with the least valuable cards you have.

Agreed, but I am not going to do this now. You will see why later

But I think you should try to figure out what is going on with the sdcards that now don't work right. Obviously you don't want to damage anything else of value, do you have any sdcard reading device that isn't really important, like an old slow netbook you don't care that much about?

If you read my original post you will see that the SD cards have already been tested in a PC running Ubuntu Linux. The one which reads 32 MB in a Windows PC, also reads 32 MB in the Ubuntu PC and a Raspberry Pi 2. The other SD card, which does not get hot, is not even detected by any computer.

I can't image this trashing the entire thing but I recall plugging in flash drives into ports and the flash drive gets hot and the port stops working right, perhaps only for a period of time. If you think about it, the amount of current that goes into the flash drive is at least partly defined by the drive itself and not the source. If the flash drive negotiates a high current that it can't deal with and then draws that current for no reason, where is it going to go but into getting hot?

I am a qualified electrical engineer with many years experience in power engineering and electronics. I understand the heating effect of an electric current. With all due respect, the current drawn by a device (assuming a fixed voltage source) is ENTIRELY defined by the device (load) so long as the source VOLTAGE remains constant. If there is a "negotiable" current limit for a USB port, then any attempt to exceed that current will result in the VOLTAGE being reduced (overcurrent protection) and as a RESULT the current drawn by the load will also reduce. If, as you say, "a flash drive negotiates a high current" (presumably this means that it negotiates for the current LIMIT on the USB port to be raised) and actually draws current to the negotiated limit, it will get HOT, whether that current is being used for a "designed for purpose" or otherwise ("no reason"). That is the very nature of electronic devices. They do not, of themselves, perform WORK, hence all of the energy consumed ultimately is turned into heat energy.

Another thing is that I was trying to reflash some flash drives and I put them into a port on a netbook and they kept claiming the drives were, I think it was 242 meg. The drives were 4 gig or 8 gig. Nothing worked right until I rebooted the netbook, which was running x86 Ubuntu 14.04 LTS. So it is possible to get incorrect size readings that clearly are coming from the host hardware/software in some sense.

I accept your point regarding mis-reading of the size of solid state storage devices. However, these are NOT changing size under different conditions, after re-boots or being inserted in different machines. I have checked the SD cards in other Raspberry Pies, Windows XP PCs and Ubuntu Linux PCs. In each and every case the SD cards perform identically, i.e. showing 32 MB for the one which is detected and gets hot and the other SD card which remains cold is NEVER detected by any computer into which it is inserted.

In your case, the sdcards show up as 32meg no matter where you put them. Assuming that they aren't getting hot just putting them in any reader, I would use a second sdcard reader, or boot to Tiny Core Linux and use the built in one, and with dd write zeros to the drive. Then I'd remove the drive and put it somewhere and see if its size was correct again.

See my above comment. One gets HOT in any SD card reader on any machine. The other is NOT DETECTED by any machine and remains cold.

Again, thank you for your response. My present thoughts on this matter are that the PSUs are OK.

FOOTNOTE: I have just tested the Pi 3 which was not involved in the initial catastrophic problem and it boots OK with the original PSU. This is after a 36 hour recovery period, which leads to me to suspect that it suffered a thermal overcurrent protection failure and in typical Pi style required a lengthy recovery. This confirms my view that there is NO problem with any of my PSUs and that the "fault" was, in each case, with the Pies. The SD cards are SanDisk Ultra purchased from Tesco and have both been in service for more than 12 months in Pi 2s. In response to those suggesting a need for heatsinks on the chips of the Pi 3 I can confirm that the chips on this particular one are barely getting warm.

I think I will return the "defective" Pi to RS for investigation, as this seems the safest and least time-consuming option, especially as I can now confirm that the other Pi 3 is OK.

Thanks, again for your help.

[pik33]

My Pi3 burned one of three SD cards I put in it. Two other card works OK.

[MarkDaniels]

by pik33 » Sun Mar 20, 2016 7:21 pm

"My Pi3 burned one of three SD cards I put in it. Two other card works OK."

Please tell me more about what happened with your Pi 3.

[stderr]

I only use Stontronics PSUs and have used tens of thousands of them over the past 10 years in industrial equipment which I manufacture. So, I am reasonably confident that the PSUs are up to the job, particularly when it is considered that one of those PSUs had already been running a Pi 3 continuously for nearly two weeks without problem.

You didn't say in the original post, I don't think you said, that you had a lot of known high quality power supplies and you'd tried them variously. All the time around here, the problem is, however, power, so that is the first thing to mention when someone posts with a problem like you posted about, right before implicating the sdcard. This doesn't mean that it is impossible for a pi to be bad, someone was complaining recently that parts were falling off their pi, that can't be good.

> If you think about it, the amount of current that
> goes into the flash drive is at least partly defined
> by the drive itself and not the source. If the
> flash drive negotiates a high current that it can't
> deal with and then draws that current for no reason,
> where is it going to go but into getting hot?
>
I am a qualified electrical engineer with many years experience in power engineering and electronics. I understand the heating effect of an electric current.

OK.

With all due respect, the current drawn by a device (assuming a fixed voltage source) is ENTIRELY defined by the device (load) so long as the source VOLTAGE remains constant.

I'm not arguing against Ohm's law, obviously if the source keeps the voltage constant, the load is defined by the load. I'm talking about which circuit is responsible for limiting the current, is it the $2000 laptop or the $2 USB flash device? The responsibility for limiting the current isn't a question of Ohm's law, it's a question of design. I shouldn't be afraid to plug *any* USB device into any host because part of the design should be that you can do that without fear. That was the design of the original RS-232, you could touch any pin to any other pin. (Now, of course, "RS-232" is highly risky with different voltage tolerances and all that.)

If I stick a fork into the electrical outlet I will create a direct short. Under the USB design philosophy, the fact that the fork doesn't limit the current is a problem of the fork. That's not how house wiring works, depending on the country, the house might have circuit breakers or fuses on the line or even in the plug box itself, the latter being similar to what one would think negotiating the maximum current available in usb means. And to current design, it might have ground fault and arc fault interrupters, all sorts of high tech designed to prevent house fires.

In many USB host circuits, it appears that the only physical limitation on the current that it will allow is its capability and some sort of polyfuse or fuses. The rest of it appears to be simply software based. For example, if you have five flash drives attached to a device and they are all low current, say that's 100ma, then depending on other software settings, the operating system will be happy or not by simply adding up what the loads *claim* to be.

Supposedly USB wasn't to allow any one device access to more than 500ma. This was a problem quickly because of wifi and mobile phone modems and people wanting to run their hard drives without a wall wart, just attach to USB. Because people have been able to often get away with this, it is clear that the average port isn't really limiting the current to some set amount based on the specs. There would surely be some high current limit but it is likely higher, and more scarily, it seems to be limited in some cases by what the power supply can physically put out, causing brown out rebooting and possibly damage to the host.

If, as you say, "a flash drive negotiates a high current" (presumably this means that it negotiates for the current LIMIT on the USB port to be raised) and actually draws current to the negotiated limit, it will get HOT, whether that current is being used for a "designed for purpose" or otherwise ("no reason"). That is the very nature of electronic devices. They do not, of themselves, perform WORK, hence all of the energy consumed ultimately is turned into heat energy.

We can use this by looking at it in reverse to show that clearly the flash drive wouldn't normally draw that level of current since it normally, even on constant writing, never gets *that* warm. So we might even conclude that because the flash drive would never need that much power, it isn't even negotiating a higher amount, it is just going nuts and loading things down without even asking. *And the host is letting it do that.*

Of course this isn't completely true, the host has its polyfuse, say, or it might have some other means, the newer pi have something that monitors at 600ma and 1.2 amps and supposedly shuts things off. But even 600ma is a lot into something that is supposedly to never need more than 100ma.

The next version of USB will fix the which side is up insertion problem, I hope it also addresses the need for more control of the loads that will be allowed on each port, including off into compliment hubs.

[MarkDaniels]

Thank you everyone for taking the time to post. I can now confirm that the problem is with the Pi 3 and NOT the power supplies. I have one dead Pi 3 which caused the initial problems and two dead SD cards. The power supplies are all fine, and always were. With both Pi 3s down (one dead and the other "unconscious" due to it's poly fuses having taken a beating) it was difficult to determine the nature of the problem. RS have agreed to exchange the dead Pi and I should have the replacement tomorrow morning.

I think this issue is now closed.

[jamesh]

It does sound like a bad Pi. This sort of fault usually goes back to the Foundation for testing to see what went wrong - hopefully will happen in this case.

[mnfisher]

I have had a similar problem.

First Pi 3 - installed Raspbian and powered up. No image on screen and the memory card got very hot very quickly (I burnt my finger on it). 64GB SanDisk card cooked.

Second Pi3 - installed Raspbian and all ran smoothly. Installed OpenElec on the same memory card and restarted. After the second of the OpenElec setup screens a "Storage removed" message was displayed and again the memory card became hot - and again destroyed.

I might be doing something wrong (though I'm not a newbie to Pi, micro SD etc) - the temperature reached and speed suggest a short circuit.
The power supply might be faulty - I did run Raspbian correctly for several hours.
It might be connected to the brand of memory (mine were both SanDisk too)
It might pose a fire risk - the cards did get very hot.

It hopefully is an isolated case - CPC's excellent customer support department will replace and I will try again.

Martin

[bksutherland]

I have to admit I'm having the same experience, it's the Pi3, not the cards. I purchased 4 RPi3's and trying to find a combination of Pi and SDCard that works is more difficult that usual. I too had the display stick on the coloured boot screen, the card got super-hot. It's basically destroyed two that way (but the Pi was not damaged, it then worked with other cards). Different current supplies don't seem to make any difference, tried 1A and 3A. The cards that were destroyed were premium and I've been using them with no probs on Pi2 - Kingston Ultimate 90MB/s read / 45 MB/s write. I've also noticed the Pi3 spontaneously crash and reboot after a while, pretty sure it's disk I/O failing. The Sandisk Ultra SD I have seems to work the best.

[RaTTuS]

if the RPI is melting SDcards then RMA so the Foundation can get it back to see what is wrong

[SheepWillPrevail]

It doesn't sound too unreasonable the RPI 3 destroys SD cards. The RPI 3 can get much hotter than most SD's cards operational temperature range; for example check out SanDisk and Samsung specifications.

Both list the maximum operational temperature as 85 degrees Celsius. If you check the thermal images some parts of the board can get up to ~100 degrees Celsius. The SD card is quite close to the CPU.

I'm having some problems too with my RPI 3 & Samsung EVO card, it doesn't pass the H2testw test. Time to order a new card.

[cdemont]

Hello,

I exactly had the same problem. Raspberry Pi 3 model B. All component were certified and bought in a kit (with boxe, power supply, SD card, hdmi cable).
First SD card with software pre-installed -> directly cooked, never booted up. Red led.
I prepared another SD card, could boot on Noobs / Menu / Installation of distribution. Reboot. OK I could use the machine for a few hours.
At next switch on the following day, red led, SD card almost burnt, was very hot. Dead.

Could you please make some answers about this recurrent issue ?

Regards,
Christophe

[W_____]

Hi,

I'm having the same issue - ordered the starter kit off Amazon - came with an SD card with NOOBS pre-installed. Came on ok first time around, shortly after the card was unusable.

Bought a SanDisk card, worked for a couple of days- same thing happened. Now the card becomes extremely hot when plugged into my PC, etc. and shows total size of 31 mb.

Both times this seems to have occurred when the power cable was pulled out of the Pi without turning it off properly (I know, my bad) - but do we expect that to have this much of an effect?

Cheers,
W

[newtron]

Hi,

exactly the same problem here:

• Brand-new Raspberry Pi 3 Model B
• Brand-new Sandisk 64 GB Ultra Class 10 SDXC card
• Original Raspberry Pi power supply (5.1V/2.5A)

Performed steps:

1. Checked SD card for compatibility using list http://elinux.org/RPi_SD_cards , looked good
2. Unzipped NOOBS onto SD card
3. Put RPi into original plastic case (no fault due to shorts!)
4. Put SD card into Pi
5. Connected HDMI cable, no keyboard
6. Connected original power supply 5.1V/2.5A

After that, the following happened:

1. No HDMI output
2. Red LED lit solidly
3. Slightly burnt electronics smell
4. SD card began to overhead severely, was hard to remove from socket due to bending
5. SD card exhibited slight crack due to thermal stress
6. Inserted SD card into Windows notebook, was recognized, but only with 32 MB (exactly as observed by another poster in this thread) - also began to heat up in computer's SD slot
7. SD card died completely. Fortunately, Windows notebook SD card reader was not damaged (afaik, the notebook has business grade overcurrent protection on all USB/SD sockets)

I have now filed an RMA claim to exchange the Pi and/or power supply; I have not tried out other SD cards. This is my first instance with a flash memory failing BTW - the Pi completely "toasted" it; I'm surprised this can happen (without any DIY peripherals connected). Never had any problem with Arduino kit for example.

As said, all components were new, this was their first use.

[wallville]

I was sitting here on my mac, then looked over and saw my brand new Raspberry Pi 3 lose power...then wouldn't restart. I decided to change the power source, after requesting an RMA from the Amazon Vendor, which is running from an APC mounted USB charging port (don't have my current tester).

I determined the supplied power which claims 5V with 2.5A wasn't pushing enough power to the Rpi3. I'm actually doing a burn-in test...this is my 3rd RPi3.

[6L20]

Similar problem here.
I bought 5 kits with rpi, noobs preinstalled sd cards, boxes and psu.
I set them up for the art piece I work on (very simple python script), tested for a few days, working smoothly.
I then unplugged (I always shut them down properly) them all and plugged them 5 meters away in the gallery, each of the five connected to a different power socket (but on the same breaker ). 2 out of the 5 did not boot at all. Nothing on screen, no Act led.
I took the sd cards out, and tried to format and reinstall them from both a mac and a pc. They are both seen as 127Mb, I tried many different methods to format them (command line, disk utility, sdformatter...), no way they get back to their original 16Gb.
Did something happen on the ac power of the building that destroyed them? Reading this post it seems that it's not such an uncommon issue. I don't know if I can dare trying to use brand new ones in the same rpis again.

[pcmanbob]

Hi.

My favourite way to format suspect SD cards is to use a digital camera , they tend not to care what's already on them and just do a format anyway.

Another way is to just download a copy of raspbian lite and use etcher to flash the SD card directly , it will not need to formatted as the image will just be written over what is already there.

https://www.raspberrypi.org/downloads/raspbian/
https://www.raspberrypi.org/documentati ... /README.md

[fruitoftheloom]

Similar problem here.
I bought 5 kits with rpi, noobs preinstalled sd cards, boxes and psu.
I set them up for the art piece I work on (very simple python script), tested for a few days, working smoothly.
I then unplugged (I always shut them down properly) them all and plugged them 5 meters away in the gallery, each of the five connected to a different power socket (but on the same breaker ). 2 out of the 5 did not boot at all. Nothing on screen, no Act led.
I took the sd cards out, and tried to format and reinstall them from both a mac and a pc. They are both seen as 127Mb, I tried many different methods to format them (command line, disk utility, sdformatter...), no way they get back to their original 16Gb.
Did something happen on the ac power of the building that destroyed them? Reading this post it seems that it's not such an uncommon issue. I don't know if I can dare trying to use brand new ones in the same rpis again.

If you do not have a Digital Camera, gparted is a good Linux Tool:

https://gparted.org/

[cspan]

Similar problem here today with 3 B+, but first noticed heat issue in 3 B.

First thing I did was update/upgrade within Raspbian. Second thing I did was to clone my card using SD card copier. Had already done it so had to go back and check something about new UUID.

Then I moved my 3B boot card to the 3B+ and tried to boot. It would not. Fearing it was corrupted, upon moving it back to boot in the 3 B, it did not boot, and it also got very hot. Later I tried to reformat it and it got very hot again. RMA for the card, had to take picture, picture revealed crack in card.

Failure to boot may have been because I had a dual boot Raspbian and LibreElec. There seems to be a sense that you can't just move a card from the 3 B to the 3 B+ if you have that setup, even if you did update/upgrade within Raspbian, because NOOBS is unaffected (and may remain outdated) and thus the initial boot file or whatever isn't current enough - not what the 3 B+ is looking for. It's unclear if the latest NOOBS would work that way.

In any case, it's pretty startling to see that either the card or the Pi itself is not fault-tolerant for these things. For an inexpensive computer you kind of don't want to go through multiple $15 MicroSD cards, esp. when you're not sure what caused the excess heat in the first place.

[bhav]

I'm curious about a similar issue I have had with a Pi 3+.

How are you handling the card when you insert it into adapters or the PI itself? I find the card is very difficult to pull out of the pi and even more difficult when it's in a case through a slot. You can easily physically damage the card if it's bending if you try gripping it out with your fingernail. I think I damaged my card in this manner so I will test again when I can source another card.

[cspan]

FWIW, I too had trouble getting my microSD card out of the slot when the Pi was in the official case. But that was the card that got damaged. I put another card in temporarily just to check that the Pi itself wasn't damaged (it wasn't) - the card I used was a Class 6 card from an old standard-def (480p) action camera. Well, that card isn't hard to get out, even with the case on. So maybe it's just damaged cards that are hard to remove? Maybe they swell up a bit from the heat?