820-00244 - PP5V_S5 woes

[Atomrepair]

This board has no visible damage. It seemed like a straightforward issue, as I found PP5V_S5 to be shorted to GND. I traced it to C8270, which had shorted for unknown reasons. After removing it, the short was gone, but still no power on PP5V_S5. I replaced U7650 with a good looking one from a donor board. Machine turned on and I got an image on screen, but it shut down mid-boot and wont turn back on.

Measured: PP5V_S5 is 'jumping around'. It has a sawtooth pattern on my scope.
I first suspected U7650 - may have been a bad chip or poor soldering. I took it off and replaced it with another one. No change.

Other components on the PP5V_S5 rail look fine. No liquid or other damage found anywhere.

Measurements:
PM_SLP_S5_L is mostly 0V but has little spikes up to ~1.5V

P5VP3V3_VREF2, S5_PWR_EN, P3V3_S5_EN and S5_PWRGD are all showing a block pattern on scope.

PPBUS_G3H is at 8.7V

Where to from here?

 

[2informaticos]

Try first to start in SMC bypass.

"PM_SLP_S5_L is mostly 0V but has little spikes up to ~1.5V"
Then xV_S4 voltages should pulse too.
Check if PM_SLP_S4_L and xV_S3 voltages pulse.
Do the same for PM_SLP_S3_L and all xV_S0.
Any pulse at ALL_SYS_PWRGD?

BTW, check the board alone, w/o tp/kb connected.

 

[Atomrepair]

Yeah, I'm working with just the board.
Found out PP5V_S4 is shorted to GND.. I hooked up lab supply to it; Nothing getting hot with 4A at 1V.
Short to GND inside the board, I guess?

 

[JohnB8812]

Inject 5 volts into 5v_S4 and see what heats up 1v is too low.

 

[Atomrepair]

Nothing on the board heats up, at least not quickly; my (fairly thick) wire soldered to the board does smoke and its insulation melts at 5V / 10A (my power supply's max current)
If I leave it on 1V for a longer period a large area around my jumper's location heats up, but nothing specific. (I've attached my wire at pad 1 of C7657)

 

[2informaticos]

An IR cam can help you in this case.
FLIR is the best option, but Seek Cam (phone adapter) could be good enough too.
If your desk power supply is 10A limit, then start with 1V and increase voltage until get 9-9.5A.
If power supply indicates 10A can get stuck and doesn't really inject.

 

[KC Micro Tronix]

I have a FLIR pro on a ipad on my bench. I consider it one of my most important tools. I grab that thing multiple times a day. It will pay for it's self the first week you own it. The software interface glitches and it is a pain in the ass at times but I would not part with it.

You defiantly want a "Pro" the resolution is so important. FLIR also overlays visual data with IR data so you can pinpoint the exact part. I can literally identify a specific cap on a phone board and pluck it off. I use it on iPads all the time and MacBooks when trying to find which Tantalum is dead. I have made thousands off the thing!

 

[Atomrepair]

I'll think about it, but not in such a hurry that I'll get one just for this repair. I'll lift stuff off the board for this rail and see where that gets me.

 

[Atomrepair]

Frustrating board.. I found out why 5V_S4 was short to GND; Q7660 was faulty. internal short, probably. I replaced it, Short gone, but no voltage on the rail still. Fed the rail 5V through my lab supply, no issues there.

Replaced U7650 again to rule that out. Got 8.7V on 5V_S4. Q7660 burnt up again.

Found that R7672, R7673, R7693 and R7692 are way off from what the schematic says they should be:

R7672 was 25K, schematic says 10K
R7673 was 2K, schem says 1K

R7692 was 17K, schem says 15K
R7693 was 2.7K, schem says 1K

This seems a little dubious to me, for all those resistors to be off like that, but I dont have a donor boards for this model to verify. Anyone know?

Replaced Q7660, U7650, R7672, R7673, R7692, R7693, C7673 and C7693 all at once.
Hooked up adapter, heard a high pitched whine, notice amperage going up - whine stops, current draw back to 0.01 (voltage 20). Hooking it up again results in no whine and 0.01 Amps. So, it seems that U7650 is dying every time I replace it, but I'm out of ideas on what might be causing it.

 

[2informaticos]

It is watse of time.
Will end with dead CPU, as all A1534 machines...

 

[Atomrepair]

Alright, too bad