Page 1 of 1

Pi 3 B+ fries if 3V3 rail is shorted to 5V

Posted: Fri May 18, 2018 9:44 pm
by imi207
Has anyone noticed that Pi 3 B+ fries if 3V3 rail is shorted to 5V. This did not happen in Pi 3 B. Seems to because of the upgrade to MXL7704 integrated regulator chip. You can read more about it here: https://www.raspberrypi.org/blog/pi-power-supply-chip/.

Is there a place where this hardware bug can be reported?

Re: Pi 3 B+ fries if 3V3 rail is shorted to 5V

Posted: Sat May 19, 2018 3:51 am
by wh7qq
There was another instance of this reported recently. If shorting 5V to 3.3V doesn't fry the cpu and other things, it is sheer, blind luck and not a bug. Don't blame the 3B+ or cry bug when doing something really dumb fries it. If your 3B didn't fry with the same screw-up, it is just luck. 3.3v is the cpu supply.

Re: Pi 3 B+ fries if 3V3 rail is shorted to 5V

Posted: Sat May 19, 2018 8:20 am
by mfa298
as wh7qq says its not a hardware bug. Its something you shouldn't do.

By connecting the 5v to 3v3 rails you're pushing 5V into things that are only designed to handle 3v3 so expect them to break. If you were lucky on earlier models and they survived something like that then that's pure luck rather than any design.

Re: Pi 3 B+ fries if 3V3 rail is shorted to 5V

Posted: Sat May 19, 2018 8:34 am
by bensimmo
Following on from this, is there any protection they could have placed on the 3v3 lines, mainly one sat next to the 5v pins.
Given it an easy wire or screwdriver or component away from a short.

Now if they had thought many years ahead and placed them right down the other end, the PoE hat's would have been much smaller.

Re: Pi 3 B+ fries if 3V3 rail is shorted to 5V

Posted: Sat May 19, 2018 9:21 am
by HawaiianPi
bensimmo wrote:
Sat May 19, 2018 8:34 am
... is there any protection they could have placed on the 3v3 lines, mainly one sat next to the 5v pins.
There is a lot of protections they could have built in... if they were selling a much more expensive computer.

bensimmo wrote:
Sat May 19, 2018 8:34 am
Given it an easy wire or screwdriver or component away from a short.
Here's a thought, don't do that when the system is powered up.

Re: Pi 3 B+ fries if 3V3 rail is shorted to 5V

Posted: Sat May 19, 2018 12:09 pm
by Z80 Refugee
imi207 wrote:
Fri May 18, 2018 9:44 pm
Has anyone noticed that Pi 3 B+ fries if 3V3 rail is shorted to 5V. This did not happen in Pi 3 B. Seems to because of the upgrade to MXL7704 integrated regulator chip. You can read more about it here: https://www.raspberrypi.org/blog/pi-power-supply-chip/.

Is there a place where this hardware bug can be reported?
I find it hard to believe any RPi wouldn't fry if you did that, and I can't see why a 3B should be any different.

Re: Pi 3 B+ fries if 3V3 rail is shorted to 5V

Posted: Sat May 19, 2018 2:31 pm
by bensimmo
HawaiianPi wrote:
Sat May 19, 2018 9:21 am
bensimmo wrote:
Sat May 19, 2018 8:34 am
... is there any protection they could have placed on the 3v3 lines, mainly one sat next to the 5v pins.
There is a lot of protections they could have built in... if they were selling a much more expensive computer.

bensimmo wrote:
Sat May 19, 2018 8:34 am
Given it an easy wire or screwdriver or component away from a short.
Here's a thought, don't do that when the system is powered up.
Cheaply done of course...

It's all well and good saying that but the pins come as is uncovered and unprotected.

Re: Pi 3 B+ fries if 3V3 rail is shorted to 5V

Posted: Sat May 19, 2018 5:21 pm
by wh7qq
The real protection of a more costly computer is that the OP probably would not have bought one if it cost more. Just wanting a cheap computer to play with and then crying when it gets broken by rough use....

Actually the RPis have been very robust. I expected death many times after my clumsy handling of the tiny boards and all the exposed pins...but they live on in spite of me. I have 6 RPis...zero thru 3B...and most live as bare boards. I watched that video of the Woz talking about static protection and I believe him but still manhandle the RPis. Hard not to.

Re: Pi 3 B+ fries if 3V3 rail is shorted to 5V

Posted: Sat May 19, 2018 10:55 pm
by LTolledo
"Mishandle/misuse an equipment, blame the manufacturer" -- truly a mindset some of us still have...

one proposed solution to prevent GPIO pin shorts, if not usually using the GPIO pins
Just MAKE SURE do this procedure with the Rpi NOT CONNECTED to ANY power source.

1. get a 40-pin header (that's in an single 2x20 configuration, though a two 1x40 configuration or two 2x10 configurations will do as well)
2. remove all the metal pins, just leaving the plastic part. (throw away the removed metallic pins immediately, dont let it lying around!)
3. insert the plastic part of the 40-pin header to the GPIO port
4. secure the header to the Rpi with hot-glue (or whatever non-metallic adhesive you prefer)

Just MAKE SURE do the above procedure with the Rpi NOT CONNECTED to ANY power source.

When you have confirmed that the hot-glue (or any non-metallic adhesive) have set and dried, you may apply power back to the RPi

Better yet encase the RPi in a non-metallic casing, just provide ample ventilation for the heatsinks.

Re: Pi 3 B+ fries if 3V3 rail is shorted to 5V

Posted: Sun May 20, 2018 8:41 am
by HawaiianPi
LTolledo wrote:
Sat May 19, 2018 10:55 pm
Better yet encase the RPi in a non-metallic casing, just provide ample ventilation for the heatsinks.
Or even better still, enclose the Pi in a metallic casing that acts as a heatsink itself.
Image

bensimmo wrote:
Sat May 19, 2018 2:31 pm
It's all well and good saying that but the pins come as is uncovered and unprotected.
The entire computer comes uncovered and unprotected. What's your point?

The OP shorted pins when messing with the GPIO while the system was powered up. Common sense should tell you that is a bad idea. And if he didn't know that before, he does now. The Pi is all about education, so consider that a $35 lesson learned, one that could save his life!

Imagine if the OP was messing around inside a mains powered device while turned on.
Image

Re: Pi 3 B+ fries if 3V3 rail is shorted to 5V

Posted: Sun May 20, 2018 9:03 am
by bensimmo
It was just a bloody suggestion to incite talk of ways it could be improved.
I know it is the way it is now, flippity flips sake.
:-)

How could you improve it from what we get now, on the 3B(+)
Education is thinking of way to improve it, not just working around it.
RPF/T are constantly looking to improve the product as far as I know.

Sticking the 5V out the way, over near the PoE pins not only makes PoE HATs simpler, smaller and leaving the gpio alone but they could then have a small cover placed over just the PoE/5V pins.
USB is over there and that's 5V, so there are lines.
I appreciate that then leaves few hats that use 5V useless.
Ah but room for a new model to make a bit more money ;-)

As a side, if these go pop in a school, they tend not to get replaced, they don't have the money.
We've not managed to replace our aging B's, they don't get used now.
They use the PCs.
When our project A+ went, it has not been replaced. Just the two left to use now.

But in the bin they go.
Back to the PCs to do the programming and other lessons.


Stuff happens when large amounts of young children are involved, if you cover the gpio up, then they cannot use them.
So why not think of a cheap protection system for the most dangerous to the Pi connections.


P.S. I have no idea how many go pop, I doubt anyone does and it may well be a bit more fragile on the 3B+ than previous setups as this topic was trying to say.

Re: Pi 3 B+ fries if 3V3 rail is shorted to 5V

Posted: Sun May 20, 2018 9:26 am
by Imperf3kt
imi207 wrote:
Fri May 18, 2018 9:44 pm
Has anyone noticed that Pi 3 B+ fries if 3V3 rail is shorted to 5V. This did not happen in Pi 3 B. Seems to because of the upgrade to MXL7704 integrated regulator chip. You can read more about it here: https://www.raspberrypi.org/blog/pi-power-supply-chip/.

Is there a place where this hardware bug can be reported?
[sarcasm]
Yes, I put diesel in my petroleum car all the time. It seems to work okay. One day I tried to put it in my shiny new LPG model. Things didn't go so well.
[/sarcasm]

Perhaps you should read the manual instead? Shorting 5v to 3v3 or any other GPIO, is guaranteed to kill your Pi.
Are you perhaps referring the the pin next to Run which on a Pi3B, used to be ground but on a Pi3B+ is now a 3v3 connection?
Pi3B
Image

Pi3B+
Image

Re: Pi 3 B+ fries if 3V3 rail is shorted to 5V

Posted: Sun May 20, 2018 9:34 am
by Z80 Refugee
bensimmo wrote:
Sun May 20, 2018 9:03 am
It was just a bloody suggestion to incite talk of ways it could be improved.
I know it is the way it is now, flippity flips sake.
:-)

How could you improve it from what we get now, on the 3B(+)
Education is thinking of way to improve it, not just working around it.
RPF/T are constantly looking to improve the product as far as I know.

Sticking the 5V out the way, over near the PoE pins not only makes PoE HATs simpler, smaller and leaving the gpio alone but they could then have a small cover placed over just the PoE/5V pins.
USB is over there and that's 5V, so there are lines.
I appreciate that then leaves few hats that use 5V useless.
Ah but room for a new model to make a bit more money ;-)

As a side, if these go pop in a school, they tend not to get replaced, they don't have the money.
We've not managed to replace our aging B's, they don't get used now.
They use the PCs.
When our project A+ went, it has not been replaced. Just the two left to use now.

But in the bin they go.
Back to the PCs to do the programming and other lessons.

Stuff happens when large amounts of young children are involved, if you cover the gpio up, then they cannot use them.
So why not think of a cheap protection system for the most dangerous to the Pi connections.

P.S. I have no idea how many go pop, I doubt anyone does and it may well be a bit more fragile on the 3B+ than previous setups as this topic was trying to say.
I agree, there was a poor design decision at the outset, but it's very difficult to decide on a change to the pinout now.

Access to the main +5V supply is very useful for expansion circuits, so I am happy to defend the presence of +5V pins on the expansion connector. What could have been done is to have made them pins 1 & 2 with 0V on pins 3 & 4 - that way an accidental short to an adjacent pin would not be fatal (the 5V current could be limited/protected to protect the PCB tracks, and the PSU should protect itself anyway).

However, it is what it is. For RPi's that need to be "hardened" against accidents, it would be straightforward to remove pins 2 & 4 from the GPIO pin strip (and alternative arrangements made if 5V is needed externally).

Re: Pi 3 B+ fries if 3V3 rail is shorted to 5V

Posted: Sun May 20, 2018 9:40 am
by n67
Actually, a lot of consumer product safety work is protecting the user against himself. That is what it means to make a product "safe".

I.e., it is all fun and games for us crochety types to be snotty on a forum here - about poor, demented users who short pins together getting what they deserve - but it shows that, deep down, we know they are right.

Meta: If I were willing to think about it long enough, I'm sure I could come up with some suitable far out analogy of some situation where some product were to be sold in such a way that an innocent user action would cause it to self-destruct - well, you get the idea..

Meta 2: And so it goes. Just being my usual contrarian self...

P.S. I'm not saying it would be easy to make the RPi protected against user error - obviously, it isn't.
Just that, as usual, blaming the victim is easy, attractive, and wrong.

Re: Pi 3 B+ fries if 3V3 rail is shorted to 5V

Posted: Sun May 20, 2018 9:47 am
by Z80 Refugee
bensimmo wrote:
Sat May 19, 2018 8:34 am
Following on from this, is there any protection they could have placed on the 3v3 lines, mainly one sat next to the 5v pins.
Yes:
Z80 Refugee wrote:
Sun May 20, 2018 9:34 am
What could have been done is to have made them pins 1 & 2 with 0V on pins 3 & 4 - that way an accidental short to an adjacent pin would not be fatal
What could be done in a future design is to provide the 3.3V GPIO pins from a completely separate supply than the main RPi 3.3V supply rail. If it only supplies expansion, accidental shorts can only affect expansion circuits. But that wouldn't protect the GPIO 2 line from an accidental short to 5V (pins 3 & 4).

Putting the 5V pins at the other end of the GPIO wouldn't have been possible from the outset - those pins did not exist on the early RPi's. Doing it now, as a redesign, would mean they are no longer backwards-compatible, with all the consequences of invalidating existing HATs etc etc.

Re: Pi 3 B+ fries if 3V3 rail is shorted to 5V

Posted: Sun May 20, 2018 9:56 am
by Z80 Refugee
n67 wrote:
Sun May 20, 2018 9:40 am
Actually, a lot of consumer product safety work is protecting the user against himself. That is what it means to make a product "safe".

I.e., it is all fun and games for us crochety types to be snotty on a forum here - about poor, demented users who short pins together getting what they deserve - but it shows that, deep down, we know they are right.
Sorry, I don't agree. Yes, there could have been design improvements (as I have posted above), but in principle the RPi is a component not a product. It is not possible to design a component to be proof against all forms of accident or deliberate idiocy, only once it has been packaged as a product can you reasonably expect that.

Even when packaged, there are things you can expect a user not to be allowed to do: such as poking screwdrivers where they shouldn't go. Product safety (quite reasonably) only requires that you can't get fingers where they shouldn't be - one cannot legislate for determined meddlers with a tool kit.

Yes, the RPi has faults; no, even if it didn't have those faults it would still require care to use it responsibly. Make things as good and reliable as they can reasonably be within the price point, but do not neglect the need to educate the user. The problem in this day and age is that people think they can do anything without reading a word in the instructions.

Re: Pi 3 B+ fries if 3V3 rail is shorted to 5V

Posted: Sun May 20, 2018 11:17 am
by Mortimer
Crikey...
Stick one of these over the 5V GPIO pins, and problem solved...
Image
:lol:

Re: Pi 3 B+ fries if 3V3 rail is shorted to 5V

Posted: Sun May 20, 2018 11:38 am
by DougieLawson
Mortimer wrote:
Sun May 20, 2018 11:17 am
Crikey...
Stick one of these over the 5V GPIO pins, and problem solved...
Image
Nice, simple, pragmatic option.

If I were designing the header for the RPi I'd have kept 5V away off the header on seperate pins, but that would have added some pennies to the factory build cost. I guess when they were only expecting to sell 10,000 units it wasn't seen as a problem. The header can't be changed now, it would ruin too many things.

Re: Pi 3 B+ fries if 3V3 rail is shorted to 5V

Posted: Sun May 20, 2018 11:43 am
by mfa298
bensimmo wrote:
Sun May 20, 2018 9:03 am
Sticking the 5V out the way, over near the PoE pins not only makes PoE HATs simpler, smaller and leaving the gpio alone but they could then have a small cover placed over just the PoE/5V pins.
USB is over there and that's 5V, so there are lines.
I appreciate that then leaves few hats that use 5V useless.
Ah but room for a new model to make a bit more money ;-)
Moving things around on the header breaks backwards compatibility and I think that's one of the things that makes the Pi great. Stuff I designed for the original B should still work on the 3B+ 5 years later - that's pretty impressive.

As for the PoE hat I'm not sure the change your suggesting would lead to that conclusion. AIUI the PoE hat has some extra stuff on it and does utilise some of the other pins as well. Until we see some pictures of the underside (maybe only possible once they're released) we'll know more of what's going on. The 6 bin header under the fan looks suspiciously like a micro-controller programming header...

bensimmo wrote:
Sun May 20, 2018 9:03 am
As a side, if these go pop in a school, they tend not to get replaced, they don't have the money.
We've not managed to replace our aging B's, they don't get used now.
They use the PCs.
When our project A+ went, it has not been replaced. Just the two left to use now.
I suspect there are two issues that maybe need addressing here.

The first would be choosing suitable cases for the Pi's and schools seeing than as an integral part of purchasing a pi. IMost of my cases still allow for using the GPIO. either through providing a slot through which the cables can pass. The other is a three part case there the top section removes easily to allow access to the gpio/csi/dsi but the rest of the board is still covered by the middle section.

The second is that in any electronics project you should only make changes with it switched off. As you get more experienced you then learn the risks of what can go wrong and can choose to take the risk. Even as an experienced hobbiest I occasionally make mistakes when making connections which can lead to broken things if I choose to work on something that's powered.

Sometimes this comes down to educators thinking about the risks for their students doing something vs the risks of them (as a hopefully more experienced person) doing them. I might be able to connect things to a running pi successfully 90% of the time, I doubt an inexperienced student could do that (they might be lucky to get 50%)

Re: Pi 3 B+ fries if 3V3 rail is shorted to 5V

Posted: Sun May 20, 2018 12:38 pm
by mahjongg
its all a learning experience, just as a TV repairman of old learned not to put his hand inside a working (CRT) TV, or (god forbid) throw a metal spanner in there, any electronic engineer worth his salary should know not to short two power supplies together.
Yes, there could have been added electronics to prevent disasters, but you know what they say "make it idiot proof, and a better idiot will still screw it up".

Locking now as this discussion really has run its course, and leaving it open will just inside more nastiness.

Has the current generation really become this dumb and lazy??? no, don't try to answer that... sigh.. :roll: :roll: :cry:

P.S. unlike what the OP said, it doesn't matter what kind of 5V to 3.3V regulator you have, in all situations something will get damaged if you put 5V on a 3.3V rail, if its not the regulator it will be the 3.3V logic. If the 3.3V logic could be powered with 5V, the 5V to 3.3V regulator would not be needed. So his remark that the old regulator could have survived this, even if true, is just a red-herring.