Repair battery powered supply circuit in RF thermostat valve

[zzattack]

Overnight one of my thermostat valves seems to have died. It was about due for new batteries so I was not surprised, but installing new batteries resulted in the unit seemingly browning out before it properly initializes its display. I've opened it up for inspection and I own several identical but functional valves so I can get reference measurements with ease.

There's a micro USB port that isn't visible unless you open it up, but powering from there still works fine on the broken unit. That suggests there's an issue only with the supply circuitry that regulates the battery input. If I power the board on the bench from the battery terminals with 2V, a good device seems to regulate this to 3.0V using circuitry I do not understand. The broken device bumps this to over 7V, which can't be right. I'm looking for some help on deciphering how this regulation works, and how to pinpoint the likely culprit. Detailed scan is attached, hopefully the filesize isn't prohibitive.

[goaty]

Looks like TLV61224 step-up converter with fixed 3V output.
Guess circuit thereafter could probably be dead due to 7V.

Unlikely, but check if FB (Pin 2) is connected to VOUT (Pin 4).

[zzattack]

On the SOT6 marked 'QXC' there is indeed continuity (on working device) between pins 2 and 4, and also between 1, 5 and 6 (VIN, L, EN), so your identification may very well be correct.

I have some hopes the 7V didn't do much damage, as all functionality seems to be fine when powering from USB.

[zzattack]

Swapping the regulator and inductor makes little difference. The previously working unit still works fine and regulator output sits at 3.0V. The defective unit with exchanged regulator and inductor now outputs about 5.7V. Does that mean the problem develops further down?

[goaty]

You swapped the chip and the inductor at the same time from good too bad unit ?

If so, then they might be ok, and the capacitors might be bad.

Very strange it´s outputting more than the 3V.

Do you have an oscilloscope ? Or measure in AC Voltage if there is any AC before or after the Regulator (Ripple)

[zzattack]

Indeed. There's about 200mV ripple on the input and about 1V on the output from the TLV61224. Swapping the 3 larger surrounding capacitors made no difference.

[goaty]

So you´re powering it with a battery ? Tried another battery ?
Not sure if the chip needs minimum load. You could check current consumption. If it´s lower than on good one, maby this is reason the voltage goes over 3V ?
Not sure.

In datasheet curves go only down to 10µA, and Dave told us to read datasheets. So if for some reason the circuit behind it is disconnected...

[zzattack]

Powering with bench makes measuring the current easy, but it behaves just the same with batteries. On working device the draw is about 30mA during startup (display init, backlight, RF comms init..) and a bit later it drops to maybe 1mA, whereas on the broken device it jumps between 250-40mA.

The LCD and backlight turn on for a bit, so there's definitely some draw from the step-up converter.

[goaty]

Allright, so it´s not too little, but too much current (250mA) after regulator. You measure resistance from Vout (Pin4) to Ground (Pin 3), maby a short or low resistance somewhere.

You could power the thing with 3V (current limited bench supply) after the regulator and see what is getting warm, but that´s risky

Fun sidenote: I did not find the switching frequency anywhere in the datasheet.

[zzattack]

No short there. It's hard to get a stable reading, but on the working device it's about 8.7k Ohms and upward of 40k on the defective device. So clearly not a short. I'm powering it up with a bench PSU. Since the batteries got drained overnight without any form of current limiting, I don't think I'll be doing much more damage with a bench PSU now. I've had it on for over 30 minutes with PSU set to 2.5V/300mA but nothing gets hot at all. I guess it averages too little for any heat to become noticable. I've also removed all ic's besides the ATMEGA, and the current drain and regulator output voltage remain roughly equal.

[goaty]

But you stated something about 250mA, so that´s more than the regulator can deliver.

If you desoldered everything besides the controller, then whats left - however improbable - must be the culprit ;-)

[zzattack]

I've taken off the mcu, but that wasn't the culprit either. So there's only a bunch of passives that I haven't swapped thus far.
On the working board, input VCC to MCU is about 2.5V so half the 5V in from USB. On the broken board it sits closer to 4.3V, but functions okay with that.

[zzattack]

Putting the MCU back on was a righteous pain, but I think I got it done so that just means I'm back where I started. Powered over USB all still works.

Maybe a better starting points is to figure out how the TLV61224 boosts up to 7V. Understanding this might reveal the underlying problem, but I have no idea how to go about this yet.

[goaty]

In my humble opinion it´s only possible if the FB is not connected or pulled down, so the boost regulator ramps up the voltage trying to get 3V.
Just replace the chip !

[zzattack]

Looks like you were correct right from the bat. One of via's of the miniscule trace connecting VOUT to FB is mostly burnt resulting in intermittent contact. Could this have happened due to the batteries running very low? After repairing the trace, the power issue seems to be fixed.
Meanwhile I've reassembled both my reference device and the previously defective one, but both mention a 'temperature sense failure' and show a temperature near -15°C (whereas room is ~22°C). Probably, I've exchanged a component between the two that is not directly compatible, or destroyed the NTC.