[Part 2] Sony WH-1000XM3 Not turining on

PART 2 Investigating the mainboard

Next, I took the left earpiece apart. Here you have to work very carefully because the NFC cable is very short and is soldered on both sides! After exposing the board I took a look at the bottom side, because there are many visible voltage supply areas. Mostly you can recognize them by the large inductors and capacitors. The Bluetooth chip from qualcomm seems to do the communication, the power management and the hardware buttons. I urgently need to investigate how the data gets from one side to the other, i.e. Atmega to the mainboard. But since all ICs are bga or with luck qfn packages, tracing is extremely difficult. I guess the board has at least 6 layers. Therefore a best guess approach is the best option for me.

I found the Datasheet after some search here.

So I first checked if the power supply reaches the other receiver without problems. This is actually just a connection via cable. Then I checked the voltages of the linear and switchmode controls from within the qualcomm ic. The chip has a total of 4 linear regulators and 2 dcdc internal power supplies. The corresponding voltages (1,8V 3,3V etc)  measured all fine on the board at the inductors and capacitors mentioned earlier.

If the headphones were again in the faulty mode I could see that the voltages increased but then collapsed again after a short time. This happens periodically with about 1.3 Hz. So where does this Signal to shutdown again come from. I started with the powerup cycle...
The datasheet gives the information that the hardware pin VREGENABLE boots the power supplies. This is usually connected to the battery voltage by a momentary switch. This seems to be the case with the WH-1000XM3. But there seems to be a diode in series to the signal, probably for ESD protection.

The voltage arrives at the output of the diode and is thus eliminated as source of error. Next I wanted to see if the power supplies are overloaded and therefore switch off. Therefore I tested all visible capacitors. Here I could not find any problems, as long as measurable in the circuit is usable at this point. So I could be half sure that everything is okay here. Then I tried to inject the voltages with my lab power supply into the board. To make a long story short this did not bring any improvement. What could stop the chip from starting now? And why only sometimes...

I looked at the chip's reset circuitry. The IC can get a reset from some sources. RST# pin, Power-on reset, USB charger attach reset, UART break character and Software configured watchdog timer. At this point I have ruled out a software malfunction for the time being and wanted to continue looking at the hardware. Furthermore the datasheet says: "If RST# is held low for >2.4s CSR8675 BGA turns off. A rising edge on VREGENABLE or VCHG is required to power on CSR8675 BGA". So this could well be my candidate.
As usual the reset pin is designed as active low. So it is either pulled high by a pullup resistor or by another external circuit. So quickly check where the pin for the reset line on this package is. Pad L1 for the reset is fortunately located directly at the edge of the chip and I could hardly believe my luck when I actually saw a track that meandered from this edge to a pad that could probably be measured.

Sorry for the bad image quality! I should' ve taken better images. You can get glorious images at the 52audio webpage that I've shown you in the previous post and here again. Absolute awesome teardown! I love the detailed work!

So I instantly grabbed the oscilloscope probe and examined the pin. And lo and behold, it is indeed the reset pin!!! Unfortunately, the headphones were now back in working mode and I waited a few hours and tried again and again to see if they finally had the error again.
Eventually I found out that as soon as I approach the board with my hands or the probe, the headphones are more likely to be switched on instead of beeing faulty, so we are definitely dealing with a floating pin or something like that!
Since I could not access the reset line properly without distorting the IC, I wanted to solder a thin copper wire to the pin. The problem with the idea was that the pad has an edge length of only about 0.3mm. This is a good demonstration on how small this is... This in combination with a high layercount of the board makes me suspect really bad things... To explain why this is so problematic: You try to heat a very small area so that the solder melts. To do this you have to use a lot of energy to overcompensate the heat dissipation into the other layers. This usually results in raising the temperature of the soldering iron, which in turn dramatically increases the chance of detaching the tiny pad from the board. The whole thing gets even worse when you consider that the wafer-thin copper layer on the board will expand faster than the substrate material underneath. And that in turn leads to pad detachment too. So extreme caution is required when working on such boards without preheating the whole board. But after about 10 minutes of soldering I had finally made a connection.

This is not the actual pad! Just to give you an idea on how tiny the pad is. That I was trying to solder to. The lower component is a SOT 23-3.

Not nice but in my defense: The insulation layer on the wire, which I unwound from a transformer, unfortunately contained more epoxy resin than I wanted (produces a lot of slag during soldering).

The green cable is a probe connection to one of the supply rails. Also only 0.5mm diameter...

Sneak peek through my microscope

Now I could make measurements without touching the board. The headphones were also nice enough to cooperate with me and entered their failed mode on the first try.

On the screen we see one of the power supplies (yellow trace) and the reset line (blue trace). There is a clear correlation between the two. So next I tried to overwrite the reset pin with the lab power supply to keep the IC permanently in non-reset mode. Now the headphones could not be switched on at all. Only when I disconnected the voltage I could switch them on again... So the reset pulse seems to be really important for the startup. At this point I unfortunately have to interrupt the repair because of lack of time... This time probably for a longer time until I have written my exams. In the meantime there is another project I am working on since the beginning of the year (Yes my projects take ages at the moment...).