Powerboost 500 C smoking switch (10+ units)

[AdaNaturebytes]

Hi Adafruit,

We're a small startup who purchased 300 Powerboost 500 C's to power our maker kits. We didn't find any issues with the boosts or switches during testing. We recently shipped our kits and they have been performing well, but we're starting to see a number of users (10 so far) contact us stating that their switch is smoking, or the switch is failing after X days of use. We are using your official #805, Breadboard-friendly SPDT Slide Switch and the solder joints are intact and clean.

UPDATE: Two users simultanoulsy reported the following a moment ago, which suggests that it's not the switch but a transistor situated nearby that is smoking, This is 10 - 15 units in 300. Can you confirm a known production run of the 500 C where you have seen similar issues Ada? These three users report a smell of burning.

User 1 - "After powering it up I just saw smoke and it smells like it has burnt out"

User 2 - "To my surprise the Power Boost started smoking and even switching it off didn’t help, only unplugging the battery helped. I gave it another try, but the same result and it has completely failed now. The smoke came from the transistor, next to the switch.

User 3 "The Powerboost intermittently will not provide power. My battery is fully charged but when I flick the switch the blue light doesn't come on and nothing happens. If I disconnect the batteries from the powerboost and then reconnect, then it will work again. Once or twice I've had a whiff or burning plastic from the Powerboost."

[adafruit_support_mike]

That almost always means someone connected a battery with the wrong polarity.

There are two ways to arrange the positive and negative wires in a JST-PH2 connector, and unfortunately you can find batteries wired either way.

Ours have the wires connected like this:

1578-00.jpg (23.34 KiB) Viewed 787 times

If the wires enter the JST-PH2 connector from its left side and the keyed face of the connector is up, the red/positive wire is above the black/negative wire.

A battery that has black above red will kill any Adafruit board you plug it into.

[AdaNaturebytes]

Hi Adafruit_support_mike,

We're using PKCELLs (the same you supply) and the JST connector can only mate the correct way as the keyed face prevents this. We've been talking to our users and many report a few days of use, everything working fine, then they turn it off or on and the unit smokes or they smell burning and it fails.

We're collecting the damaged boards as we suspect a general issue, but would like Ada to take a look due to the quantity involved here.

[adafruit_support_mike]

I agree that the failure rate is unusually high. We haven't seen any other batches with that kind of problem.

I'll talk to the folks in production and QA about what we can do to identify the problem. In the meantime, any other information you can collect about the context surrounding the failures will help.. what kind of circuit did the user have it connected to, did the board fail while running, while starting up, while stopping, etc. The more information we have, the better the odds of finding some common factor.

Please do confirm that people were using the LiPo you shipped when the PowerBoost failed. Even eliminating that variable is a step forward.

[AdaNaturebytes]

Hi Mike,

We're collecting together as much information as possible. At all times the Powerboost was connected to a Raspberry Pi (as part of the Kit) via the standard USB connection from the Boost.

We'll reply with our findings shortly.

[adafruit2]

common issues are: wrong polarity on the battery, wrong kind of battery, too-long wires on the input or output, or spikes of current that burn out the amp, shorting input or output, too high voltages on input or output. if you can indicate which exact component is dying and how its wired exactly, that will help.

[AdaNaturebytes]

Hi MIke / Site Admin,

Whilst we collect together the full picture, here's a photo sent from one of our users identifiying the damaged component.

You can see bubbling / a burnt connection in the component adjacent to the ER in POWER written on the PCB. We must now have 10 + boards with the same damaged component and issue, with users reporting smoke and a hot to touch switch. Note - we're using the official Adafruit switch too.

powerboost_500c_damaged.jpeg (406.41 KiB) Viewed 746 times

[adafruit2]

i wonder if the switch is a make-before-break type or if during switching they linger in a middle position - that would cause VBAT to short to GND and possibly destroy the charger chip (which is what is destroyed here)

Try having them solder only the ENABLE and GND leads of the switch and not have the other pin connect to VBAT, they'll need to bend the switch out of the way so there's no risk of accidentally having the third wire touch VBAT

[AdaNaturebytes]

Hi Admin,

This makes sense in terms of the switch position. We used the very same switch stated in the official guide. I just checked the feedback and one user reported;

Day 1 - "Noticed smoke coming from / near the switch after approx 45 mins use so turned it off. The switch was incredibly hot"

Next day - "Turned it on again and it ran fine for several hours. The switch was cool to the touch".

Following day - "Turned it on and it smoked straight away".

Can we email you directly with the full report of our investigation (and detailed photos).

[adafruit2]

we don't do email support, you can post here including images, or link to an imgur or other file. you may need to find another solution if this board isnt suited for your setup

[AdaNaturebytes]

Hi Adafruit2,

We have way too many photos to post all with the same issue. We reproduced the issue ourselves to confirm a moment ago. We literally turned the Powerboost on, waited, turned it off, waited, turned it on - boom. The charge controller exploded. There are 300 Powerboosts with our users, all with cleanly soldered Adafruit switches on as we followed the official instructions (https://learn.adafruit.com/adafruit-pow ... off-switch) and used your own recommended switch (https://learn.adafruit.com/adafruit-pow ... off-switch), however this results in the Powerboost failing.

We will now no longer use the switch, but our users have no working Powerboost. Can we return the failed units for replacements? We can inform our users to desolder the switch in the meantime.

[adafruit2]

you can put the images in an imgur or other thing. its not like email is great for attaching dozens of photos either :)

also, *its not clear* what the issue is. this is just one guess. it could be something else! we recommend really figuring out what the problem is first.

[AdaNaturebytes]

We're continuing to build the case of evidence. Here's a photo received moments ago showing the exact same component damaged and the same connection to the board looks charred.

ada_.jpg (194.44 KiB) Viewed 657 times

[adafruit2]

hiya, we don't see any damage to that board or charring?
if you're still receiving reports, have you told your kit builders to not do the thing that might causing them to get damaged?
might be a good idea to notify them to avoid more problems!

[AdaNaturebytes]

Yes, we're telling all of our customers to desolder or remove the switch. The switch is causing the issue - without it they perform fine, but with it, randomly turning it on will cause the charge controller to pop or smoke and the switch itself gets hot. Again, this is using the recommened switch.

I've posted the photo again but indicated via a red dot where the middle component connection to the PCB. In all cases has charred. KD4C is reported to smoke, smell of burnt plastic, or physically pops and bubbles. The middle connection (KD4C) to the PCB is charred black in both of these cases below and both units died.

We have recreated the issue on our desk. Using the switch is a random game - sometimes it's fine, sometimes it smokes and pops instantly, or after 30 mins. We're basically collecting (in the post) the damaged units from our users and would like to replace them, this time without the switch. We would understand if we had used our own switch, but using the official instructions we expected it to be fine.

compare1.jpg (668.2 KiB) Viewed 649 times