Saturday, December 19, 2015

Attempted repair of a FitBit Flex

I had an old Fitbit Flex that had stopped working so I decided to try to fix it. (Important note:  Fitbit had given me a free replacement - that's pretty great of them.  I tried this fix after losing the replacement).

The problem I had with this Flex was that the battery got low and then it would not recharge.  My idea was to charge it "manually" by somehow creating a better connection to the Flex from my DC power supply.  Using my multimeter I wasn't able to measure any voltage across any combination of the 3 pins, so my next step was to try to determine which pins should have power or be used to try to charge the Flex.

I started by reading this description of taking apart the Fitbit Flex, an IFixIt teardown:

From the above - and more importantly the comments (e.g. I was able to understand which pins on the Flex are connected to the battery:
"The center pin is positive. And while looking at the led side of the flex and the pins towards you, the right pin is ground." - theabsolut1
I connected a DC power supply set to ~3.2 V across those pins (also recommended from the above comments), but there was no appreciable current flow.

From that point, figuring I had nothing to lose, I decided to take the Flex apart.  Again, following comment in the above pointed me in the right direction:
"Im not sure why you guys sawed it apart, the cap by the led is held in place with something similar to hot glue, a few seconds in front of a hot air source and it comes out pretty easily, the circuit pack is also held in place with some sort of hot glue." - CalebG
I didn't have a hot air gun, so after reading this:

I decided I could use the toaster oven to melt the glue*.  I set it to 250 F, wrapped the Flex in a small piece of foil (mainly to keep it from falling between the grates) and put it in the oven for 4 minutes.  When the time was up took it out and used two pairs of pliers to pull it apart, one pair gripping the end with the lights / LED's - no more than 1 mm from the end (see steps 5 & 6 of the IFixIt for the analogous work but with cutting instead of melting the glue and pulling apart).

With the Flex apart, I started checking voltages and resistances.  The pins on the board had the same behavior as external pins - no voltages.  I then measured the voltage directly at the battery - it was putting out 2 V.  I then checked resistances between the battery and the pins - there was 0 resistance between the battery's negative terminal and the ground pin (good!), but there was infinite resistance between the positive battery terminal and the positive pin (center; bad!).

I decided to attempt to charge the battery by connecting a DC power supply (set to ~3.2 V) to the battery terminals instead of the pins:

It seemed to work, there was some measurable current reported by the DC power supply.  I disconnected it, and then measured the voltage across the battery terminal as ~3.2 V.  To see if charging the battery was going to help, I tried to detect its bluetooth signal with my phone - and it worked! (I used the Bluetooth Finder app for Droid:  I then tried to connect it to my phone / fitbit account, the LEDs came on, and I made it to the screen where it prompted me to tap the device to confirm its identity.  Tapping it did not work, and then it went dark.  Measuring the battery again, the voltage was down to less than 2 V.

I decided to try to do a more serious charging attempt, so I first soldered a wire to the positive terminal of the battery, to make it easier to charge and ultimately to possibly connect it to the positive terminal pin.  Based on the USB spec, I decided to try to charge at 5 V.  Initially, it appeared to be running at around ~100 mA of current, it didn't seem too hot so I left it to charge.  Unfortunately, when I came back I smelled cooked electronics and saw this:

I had thought that since the USB spec is to provide 5 V and the charger was just a USB adapter, that the battery component had built in circuitry to regulate the charging.  Based on the above that was almost certainly wrong!  In hindsight, it seems that the charging circuitry is probably inline between the positive battery terminal and the corresponding pin of the board, and the disconnect / infinite resistance between them was probably due to a failed component within that circuit.

In theory, if I hadn't blown the battery, I might have been able to recharge the battery properly and connect the device to my phone / account.  But then every week when the battery ran down I would have needed to do the manual recharge, which is slow and requires an unfortunate amount of human intervention.

* I actually originally was going to do it in our brand new oven, but the wife recommended using the toaster instead.

No comments:

Post a Comment