A couple of new Pi3B+ Not booting - 3V3 not working. Looking for further information/support.

[liamkennedy]

TL;DR : Batch of new Pi3B+'s - 2 out of 6 of them have the "known" 3v3 problem and are DOA.

EDIT: Adjusted the heading so as to not have this be such an alarming post. It turns out this is an extremely rare issue and I have just been one of the unlucky ones.

DETAIL:

I purchase quite a few Raspberry Pi's for my product (shipped more than 3000) and up until recently, I was sticking with the Pi3B as my application works just great on that. However.. supply of those seems to be diminishing from the distribution chain and now Pi3B+ are more available (e.g. Arrow electronics has 1000's of 3B+'s but now zero 3B's).

I have a very robust build and QA process for my product so I can be sure what I ship works when received.

On my most recent batch of six 3B+'s one was DOA.

The other 5 seemed to be OK... the assembly / test went fine.

At this point I just put that down to a random manufacturing defect. It's only the 2nd Pi I've ever received that was DOA.

Well tonight I randomly decided to pull a couple of the fully tested units out and use them.

The first one I tried now didn't want to boot. Of course I do the usual.. switch cables / power supplies .. still didn't boot. Tried just disconnecting the HDMI cable... and just plug in power. That then booted!! No idea if not plugging in the HDMI helped. I then unplugged/replugged everything again. and now the unit is booting seemingly reliably. (I hate these seemingly "no-repeatable fixes". I just don't want to ship something that's done that.

This had me pull out all of the other units I had built (and tested) today. I discovered one of those is also now not booting. Unplugging everything and reconnecting things didn't make any difference.

So now I have TWO of the six units today dead. And one.. that became dead.. and is miraculously now working.

That had me check the forums to see if anything like this was reported... and I found the sticky post that mentions the problem with some Pi3B+'s where the 3v3 stops working. I had not heard of this issue until today

I then used a DVM to check the 3V3 and sure enough, for the two units that are now dead, it's got nothing on it (well.. my DVM reads 6.8mV). 5V is fine.

So.. What the heck to do now? It seems to me 3B+'s have a very nasty reliability problem.

In looking back at the forum posts about this - it looks like this has been known about for some time (December 2018?). Is there any new official response from the foundation about this issue? The Sticky post that mentions this issue says "The issue is under investigation" - but it's not clear when that note was added.

I really want to be able to purchase 3B's for now.. but the supply chain seems to be a problem there - and I'm concerned it will get more problematic.

If anyone else has any further input on this I'd love to hear from you.

Thanks!

[LTolledo]

Yah, the "RPi3B+ 3.3v scare" put me off as well. While only having 3 units of RPi3B+ at the moment (been lucky so far), am putting off replacing my other RPi3Bs, as the RPi3B seems more "rock solid" and "reliable" compared to the RPi3B+.

Have not got my hands on the RPi3A+ yet, as it seems my favorite local and online suppliers seems not keen on carrying it at the moment...
The RPiA3+ is reported to carry the same PMIC as the RPi3B+.....

Regarding reliable supply of RPi3Bs, maybe some of the mods here can help you out....

[liamkennedy]

Yah, the "RPi3B+ 3.3v scare" put me off as well. While only having 3 units of RPi3B+ at the moment (been lucky so far), am putting off replacing my other RPi3Bs, as the RPi3B seems more "rock solid" and "reliable" compared to the RPi3B+.

OMG... I didn't realize "it" had a "name".... the "RPi3B+ 3.3v scare"?

Have not got my hands on the RPi3A+ yet, as it seems my favorite local and online suppliers seems not keen on carrying it at the moment...
The RPiA3+ is reported to carry the same PMIC as the RPi3B+.....

I have one.. and it seemed to work just fine.. maybe it was just luck.

Regarding reliable supply of RPi3Bs, maybe some of the mods here can help you out....

The 3B's seems to be in stock via Amazon. As a reseller (USA) I go through the distributor network usually - as that generally has a more reliable supply in quantity (and at a marginally lower price).. but I may just have to switch to Amazon for now. Crossing my fingers that 3B's will be in more plentiful supply.

[Burngate]

(I'm assuming you're referring to the ISS-Above, which appears to use Pimoroni's PiGlow)

If I were involved with producing / selling a commercial product, I would be thinking about adding a small LED-resistor across 3v3 and ground, as a go/no-go indicator for the 3v3 rail.

[liamkennedy]

(I'm assuming you're referring to the ISS-Above, which appears to use Pimoroni's PiGlow)

Yes. That's it. You mention the PiGlow... is that actually relevant to this issue (genuine question)?

If I were involved with producing / selling a commercial product, I would be thinking about adding a small LED-resistor across 3v3 and ground, as a go/no-go indicator for the 3v3 rail.

Adding? Or is this just a "QA/Test" tool?

Of course.. it would be much better to not have the issue at all. I can't imagine this is a good thing for anyone who is buying 3B+'s.

By the way.. that one 3b+ that seemed to "magically" fix itself after I was messing with the HDMI cable. It seems like THAT one is extremely picky about what HDMI cable I plug in. There are a couple of cables here that reliably cause it to fail (but still sometimes able to get it running after unplugging it). I opened a brand new set of HDMI cables and now that unit never has a problem. This has me think at least one of the units here has some sensitivity around the HDMI connector. By the way.. those cables that cause this problem on the 3B+ work 100% reliably in a bunch of 3B's and B+'s - so it seems to be JUST the 3B+. Very odd. I wonder if I am seeing two different failure modes here - or perhaps the issue has something to do with the HDMI port.

[LTolledo]

OMG... I didn't realize "it" had a "name".... the "RPi3B+ 3.3v scare"?

....well... that term just slipped off my mind when thinking about it...
its not IIRC the "official name" for it..... but seems to suit it based on the cause and effect

[mahjongg]

Well, you must be very unlucky, as only half a dozen of cases are known, and the rate of cases seemed to have dropped.

The number of failures appears to be low, and there is a feeling that the rate is falling. However, we are still interested to swap boards directly with users, particularly those with multiple failures in case we can establish a pattern that may help to root cause the problem, so please continue to point out any likely looking posts you see so we can "reach out".

Obviously your dealer should replace these defective units, but we still rather have you sending one or more of them to the RPF engineers, as one reason we cannot fix the problem is that we never (to my knowledge) got any defective units to examine, that really had this defect.

I reported your case, and there will be an RPF engineer reporting to you, perhaps you could also tell him exactly what you used to power these RPI's, as I have a feeling it could have something to do with this problem.

by the way, there is no name for this phenomenon known to me (as its so rare) but it seems it does have one now .

[k-pi]

So far so good, I have both RPi3B+ & RPi3A+, both of which seem to be working OK, but I am holding off from buying any more.

(I'll wait for the RPi4 - when ever that is going to make it's appearance.)

[liamkennedy]

Well, you must be very unlucky, as only half a dozen of cases are known, and the rate of cases seemed to have dropped.

The number of failures appears to be low, and there is a feeling that the rate is falling. However, we are still interested to swap boards directly with users, particularly those with multiple failures in case we can establish a pattern that may help to root cause the problem, so please continue to point out any likely looking posts you see so we can "reach out".

That is very useful to know.

In my case there is one that was DOA... just never switched on other than the RED power LED - which I eventually confirmed had the 3V3 issue.

Then the other two units that initially passed my assembly/test but then later exhibited the same "won't boot" issue. It looked like that was the SAME issue as the first one - however, the odd HDMI component to the problem/magical fix has me thinking that may be an indication of something else going on.

Could a small change in the HDMI port tolerances on the 3B+ connector actually cause some kind of short that results in the 3V3 problem? I somehow doubt that. HDMI connectors are such a basic component and if that was the case I am sure there would have been more issues.

Perhaps this was ACTUALLY caused by those two HDMI cables. I have tested the two cables on other Pi3B's and some non-RPi devices like between camera/Bluray and switcher/TV and there is no indication of any issue with the cables there.

I appreciate my experience here is just another "one" of the very few reported failures so it could very well be something on my side that is causing this. For now... I'm just not using those two HDMI cables in my assembly process.

Obviously, your dealer should replace these defective units, but we still rather have you sending one or more of them to the RPF engineers, as one reason we cannot fix the problem is that we never (to my knowledge) got any defective units to examine, that really had this defect.

I would be happy to send these to RPF for review. I've already purchased replacements (3B's) via Amazon (CanaKit) as they were in-stock and available for same-day delivery (need to ship these by Monday)

I reported your case, and there will be an RPF engineer reporting to you, perhaps you could also tell him exactly what you used to power these RPI's, as I have a feeling it could have something to do with this problem.

Thanks for reporting this. I'll look out for that contact from RPF

The power supply are ones I have been using for years - which are 2.5A 5.1V branded "pro-elec" which I would purchase through MCM-Electronics (now Newark/Element14) but are variously provided through other distributors as well. So - not the "official" power supplies - but ones I have come to view as extremely reliable over the years.

by the way, there is no name for this phenomenon known to me (as its so rare) but it seems it does have one now .

Haha. I did google search for that term and the only link that came up was to THIS discussion thread.

[Imperf3kt]

I recall one reported case where the user said his/her HAT worked fine on a 3b, but caused a 3b+ to lose the 3v3 supply.

Further investigation turned up the fact that the HAT was at fault and the issue was also happening on the Pi3b, but nobody noticed.

I cannot find the thread, but remembered it's conclusion.
Maybe you could check your hardware out just to be sure?

[liamkennedy]

I recall one reported case where the user said his/her HAT worked fine on a 3b, but caused a 3b+ to lose the 3v3 supply.

Further investigation turned up the fact that the HAT was at fault and the issue was also happening on the Pi3b, but nobody noticed.

I cannot find the thread, but remembered it's conclusion.
Maybe you could check your hardware out just to be sure?

Good idea. Yes. I've done a full test of everything involved here.

The RPi3B+ I have that is DOA had a PiGlow installed that's working just fine in another Pi3B+ and Pi3B... it remains DOA even without the PiGlow installed. I've also shipped 1000's of these PiGlows... and those are equally rare to fail. So far none of those failures ever resulted in causing the Pi to fail.

So it's looking to be unrelated to the PiGlow

[jamesh]

OK, so despite the title of this post attempting to screw up our sales, lets have some figures.

We have sold multiple millions of Pi3B+. If there was an inherent defect we would be clearly inundated with returns and problems. But we are not. And also note, every Pi leaving the factory works, as each one is individually tested by quite a cool robot machine thing.

So I would be extremely surprised for someone to received any Pi DOA, and expecially extremely highly surprised if you had two.

There is a known failure modeif the top two GPIO pins are shorted - this can cause the 3.3v failure. I suspect this is probably what has happened here.


In short there is no reason not to chose the 3B+, just take normal precautions when using.

[liamkennedy]

OK, so despite the title of this post attempting to screw up our sales, lets have some figures.

We have sold multiple millions of Pi3B+. If there was an inherent defect we would be clearly inundated with returns and problems. But we are not. And also note, every Pi leaving the factory works, as each one is individually tested by quite a cool robot machine thing.

So I would be extremely surprised for someone to received any Pi DOA, and expecially extremely highly surprised if you had two.

Sorry that my post has been so upsetting to you. I'm very happy to send this one back to you so you can check it out and to provide any further diagnostic detail that would be helpful. I'll see if I can edit the post title to make it seem less alarming.

The first one that was DOA on first boot - I can assure you all I did was place a PiGlow on the GPIO and plugged in the power supply / HDMI. I've done that thousands of times and know what I am doing. It's not USER ERROR.

There is a known failure mode if the top two GPIO pins are shorted - this can cause the 3.3v failure. I suspect this is probably what has happened here.

Yes I understand that - however the Sticky post mentions a "spontaneous" case too.

There have been some reports of the 3V3 supply suddenly stopping working, often after shorting the 3V3 to GND, but in a few cases also spontaneously

The first one in this new shipment was DOA... so perhaps that was one of those "spontaneous" cases. Possible? Maybe not user error?

In short there is no reason not to chose the 3B+, just take normal precautions when using.

Understood - and I appreciate this perspective.

What would have helped is to have some further information in the Sticky post about this issue. If there was a note like the one above from the forum moderator that only half a dozen cases were known - I would have been much less likely to feel such a reluctance to switch my product to be using the 3B+

I have not yet received any contact from an engineer to request the DOA unit back (as mentioned by the other forum mod - mahjongg). Do you want it? I'd be happy to help in any way if that would be appreciated.

[jamesh]

OK, so despite the title of this post attempting to screw up our sales, lets have some figures.

We have sold multiple millions of Pi3B+. If there was an inherent defect we would be clearly inundated with returns and problems. But we are not. And also note, every Pi leaving the factory works, as each one is individually tested by quite a cool robot machine thing.

So I would be extremely surprised for someone to received any Pi DOA, and expecially extremely highly surprised if you had two.

Sorry that my post has been so upsetting to you. I'm very happy to send this one back to you so you can check it out and to provide any further diagnostic detail that would be helpful. I'll see if I can edit the post title to make it seem less alarming.

The first one that was DOA on first boot - I can assure you all I did was place a PiGlow on the GPIO and plugged in the power supply / HDMI. I've done that thousands of times and know what I am doing. It's not USER ERROR.

There is a known failure mode if the top two GPIO pins are shorted - this can cause the 3.3v failure. I suspect this is probably what has happened here.

Yes I understand that - however the Sticky post mentions a "spontaneous" case too.

There have been some reports of the 3V3 supply suddenly stopping working, often after shorting the 3V3 to GND, but in a few cases also spontaneously

The first one in this new shipment was DOA... so perhaps that was one of those "spontaneous" cases. Possible? Maybe not user error?

In short there is no reason not to chose the 3B+, just take normal precautions when using.

Understood - and I appreciate this perspective.

What would have helped is to have some further information in the Sticky post about this issue. If there was a note like the one above from the forum moderator that only half a dozen cases were known - I would have been much less likely to feel such a reluctance to switch my product to be using the 3B+

I have not yet received any contact from an engineer to request the DOA unit back (as mentioned by the other forum mod - mahjongg). Do you want it? I'd be happy to help in any way if that would be appreciated.

I think that we are fairly sure of the cause of this 9as above), so probably don't need any more back, but I will check with the HW team.

[hippy]

Well, you must be very unlucky, as only half a dozen of cases are known, and the rate of cases seemed to have dropped.

The number of failures appears to be low, and there is a feeling that the rate is falling. However, we are still interested to swap boards directly with users, particularly those with multiple failures in case we can establish a pattern that may help to root cause the problem, so please continue to point out any likely looking posts you see so we can "reach out".

That is very useful to know.

Having caught my attention I started to track those 'unexplained failure' reports posted to the forum. There have been about 60 so far, which, as said is low, isn't many compared to how many have been sold. To save repeating myself, my previous comments on the issue are here -

https://www.raspberrypi.org/forums/view ... 5#p1412622

As noted there; the number of reports peaked in August 2018, and that still holds true, but, to update the figures, there has been a dozen or so reports since Christmas - 2 in Mar 2018, 3 Apr, 2 May, 7 Jun, 9 Jul, 14 Aug, 3 Sep, 2 Oct, 4 Nov, 1 Dec, 8 in Jan 2019, 4 Feb, 2 Mar.

[liamkennedy]

So now I have a reliable way of inducing this 3V3 problem without even touching the GPIO - it has to do with which device I have plugged into the HDMI port.

This is unrelated to that DOA Pi3B+. That one is totally dead and doesn't boot under any circumstance.

I have the Pi's plugged into a Blackmagic Design TVS (BMD TVS) switcher. This allows me to boot up to four of my devices simultaneously and have the output displayed on a multi-view monitor (that helps me speed my build/test process for my product). It seems connecting to THAT switcher is THE factor that causes the Pi3B+'s to not boot. This only happens with Pi3B+'s - not the Pi3B.

Simply unplugging the HDMI port (connected to my BMD TVS) with the Pi powered and not booting allows the Pi3B+ to continue to boot up.

I did some more investigation using my DVM and have a curious result to share.

With the Pi's power supply DISCONNECTED but the HDMI port connected to my BMD TVS the DVM shows a voltage of 0.6V on the 3V3 GPIO port.

If I instead plug the HDMI directly into a monitor the DVM shows a floating but very low mV range value (I assume that is significant).

I'm not any kind of expert in these issues - but it does seem conclusive in my own environment that my own video switcher is CAUSING this issue for the 3B+'s (except that one which was DOA).

I fully acknowledge that my output device (that BMD TVS) is an unusual factor in this. I am curious however if this might explain some of the other reports of issues out there. Could other HDMI display devices (TV's/Monitors) have the same kind of thing going on?

What would even cause such a "back voltage" is way beyond my expertise. I have no idea what is happening with that.

I hope that is helpful.

[fruitoftheloom]

So now I have a reliable way of inducing this 3V3 problem without even touching the GPIO - it has to do with which device I have plugged into the HDMI port.

This is unrelated to that DOA Pi3B+. That one is totally dead and doesn't boot under any circumstance.

I have the Pi's plugged into a Blackmagic Design TVS (BMD TVS) switcher. This allows me to boot up to four of my devices simultaneously and have the output displayed on a multi-view monitor (that helps me speed my build/test process for my product). It seems connecting to THAT switcher is THE factor that causes the Pi3B+'s to not boot. This only happens with Pi3B+'s - not the Pi3B.

Simply unplugging the HDMI port (connected to my BMD TVS) with the Pi powered and not booting allows the Pi3B+ to continue to boot up.

I did some more investigation using my DVM and have a curious result to share.

With the Pi's power supply DISCONNECTED but the HDMI port connected to my BMD TVS the DVM shows a voltage of 0.6V on the 3V3 GPIO port.

If I instead plug the HDMI directly into a monitor the DVM shows a floating but very low mV range value (I assume that is significant).

I'm not any kind of expert in these issues - but it does seem conclusive in my own environment that my own video switcher is CAUSING this issue for the 3B+'s (except that one which was DOA).

I fully acknowledge that my output device (that BMD TVS) is an unusual factor in this. I am curious however if this might explain some of the other reports of issues out there. Could other HDMI display devices (TV's/Monitors) have the same kind of thing going on?

What would even cause such a "back voltage" is way beyond my expertise. I have no idea what is happening with that.

I hope that is helpful.

Yes HDMi carries voltage but very low:

https://en.wikipedia.org/wiki/HDMI

Though the RPi has circuit protection:

https://www.raspberrypi.org/forums/view ... 8#p1258837

[dannyk6]

I'm wondering if I'm seeing this issue as well. Similar to OP, we use Rpis in our products as well.Typically assemble 10-20 per day and have been using Pi for about 2 months now and have never had a dead pi in our 3+ year history of using Pi's. I just recently bought a new batch of 200 3b+ from Arrow and so far 17 out of about 50 of the ones I've went though will not boot, just a solid red LED. Using the same SD card in a different pi works fine.

I will be bringing a multimeter to the office tomorrow to test for 3v3 issue and will report back, although everything else is pointing to this.

[liamkennedy]

I'm wondering if I'm seeing this issue as well. Similar to OP, we use Rpis in our products as well.Typically assemble 10-20 per day and have been using Pi for about 2 months now and have never had a dead pi in our 3+ year history of using Pi's. I just recently bought a new batch of 200 3b+ from Arrow and so far 17 out of about 50 of the ones I've went though will not boot, just a solid red LED. Using the same SD card in a different pi works fine.

I will be bringing a multimeter to the office tomorrow to test for 3v3 issue and will report back, although everything else is pointing to this.

Very interested to hear what you find out. Test#1 is - unplug the HDMI cable. See if they boot then.

[MicroTransactions]

I've been having similar issues, although I don't have a multimetre on hand to check the GPIO. Flashes the ACT light once, then just sits with the red light on. Mind you, mine worked initially, but the next day wouldn't function with HDMI in. Boots fine without it. https://www.raspberrypi.org/forums/view ... &p=1437588

[dannyk6]

3v3 issue confirmed!! There is no voltage on the 3v3 pin on our dead Pi's (3b+ from Arrow). Red light is on, and there IS voltage on the 5v pin. We currently have 17 failed Pis and still have about 150 to check. I will update again once we've checked all of our in-stock Pis.

Also just a note: these are exhibiting this behavior without the HDMI cable plugged in. Only power is plugged in. Tried powering on with HDMI cable plugged in, then unplugging HDMI cable and still no boot.

[liamkennedy]

3v3 issue confirmed!! There is no voltage on the 3v3 pin on our dead Pi's (3b+ from Arrow). Red light is on, and there IS voltage on the 5v pin. We currently have 17 failed Pis and still have about 150 to check. I will update again once we've checked all of our in-stock Pis.

Also just a note: these are exhibiting this behavior without the HDMI cable plugged in. Only power is plugged in. Tried powering on with HDMI cable plugged in, then unplugging HDMI cable and still no boot.

Those do sound more like the DOA category rather than HDMI induced failure. Although 17 out of 50. Ouch. Does your product have anything plugged into the GPIO?

In my case with the HDMI failure mode the 3v3 GPIO had 0.6V shown on the DVM when the power (microUSB) was unplugged.

[dannyk6]

Nothing plugged into the GPIO. No chance of shorting the GPIO pins as the Pi's are placed in a plastic case before ever powering them up.

We went through another 25 Pis day and had 0 failures. I don't know if we hit a bad batch yesterday or what?

The failed Pis we saw yesterday weren't initially DOA. They all booted up fine, we ran our normal configuration steps, rebooted and that's when the 3v3 issue started and the Pi wouldn't boot up. Nothing has changed with our config setup. Same monitors, power supplies, CLI commands etc. that we've been using for months.

[Imperf3kt]

Was anything ever connected to the GPIO, not just right now, but at any point in the Pi's life?

[liamkennedy]

Was anything ever connected to the GPIO, not just right now, but at any point in the Pi's life?

Who are you asking that question of?