2021–10–28:
Fixing broken Pinephone, aka Pinephone repairability
One of the Pinephones I use for testing started having a parasitic power
consumption of additional 900mW after it blue-smoked a Pinephone keyboard MCU
during development of the Pinephone keyboard.
TL Lim sent me a replacement Pinephone Beta unit, because of this, but
I still dislike to throw otherwise perfectly fine HW to the trash, so
I decided to try to fix the issue.
Finding the culprit
Pinephone is disassemblable with just a screw driver and some fingernails
(tweezers help, too). That means extracting the mainboard was quite easy:
The presenting issue is that when the phone's mainboard is off, it will
still consume 900mW. This is quite a bit of power, that will surely create
noticeable heat, and it will likely be concentrated around the culprit.
With the mainboard out, I tried to find, where this heat is produced. I had
some ideas beforehand, so I focused on the PMIC and the circuits around the
USB-5V voltage rail.
Some unnamed youtube personalities use alcohol evaporation test to identify a
component that's heating up more than the rest of the board. Being the monkey
I am, I tried that test, too. I poured some alcohol on the PCB and connected
it to the battery.
PCB might have gotten a bit drunk, but otherwise nothing exceptional
happened. The reason was that the battery was completely dead. Multimeter showed
some 2.8V across its terminals, so it was providing no power to my test.
I fully charged the battery a week ago and let it sit on my table, unused, so
this was a bit unexpected.
Alcohol test is messy, so I gave up on that and just used my finger to find
the heating component. 1W is not easy to hide, not even from my big fingers.
I found that D600 was getting very hot. D600
serves as a rectifier for the boost DC-DC converter for USB-5V
power rail. It didn't make sense to me how it could be the the failing part,
because the DC-DC converter is off when the phone is off, and
USB-5V should not be loaded either in that state.
Something else connected to USB-5V must be heating up, too. So
I continued my search. It turned out to be U1302, which is a
DCIN to USB-5V switch. I measured a resistance across
its terminals and it was 14 Ohm. It was not a switch anymore. It
was loading the USB-5V power rail all by itself!
Quick calculation verified that the load matches with my power meter
measurements. USB-5V has roughly a voltage of
VBAT – 0.6V, when the DC-DC converter is off. That may be
approximately 3.6 V. This ammounts to
3.6^2 / 14 = 926 mW. That sealed the case.
Fixing the issue
Removing U1302 is safe. It will just make Pinephone unable to
supply power to external USB peripherals connected to the Type-C port.
So I decided to remove it. The more leads the part has, the more annoying it
is to remove with just a soldering iron. So I decided to cut the leads on one
side with pliers, and unsolder the leads on the other side.
And that fixed the issue. :) This Pinephone will continue serving some tasks
well into the future, instead of being thrown out. It's not the first time
I managed to fix some issue on one of my Pinephones in this way. It's all made
possible by having schematics and component placement maps publicly available
and by Pine64 making sure disassembly and re-assembly is easy and doesn't
require any special tools or risky maneuvers. If I'd want, I can buy the replacement part
on Aliexpress for $5 and solder it back.
Repairability at home with common EE hobbyist tools, is one of the other
things I like about Pinephone.