A1989 / 820-00850-A stuck in recovery

[cruiser_]

A friend replaced the display and it stopped booting. Then he did DFU/Revive... wants the data.

820-0850 is now stuck in 'Recovery'. Can be forced into DFU with a jumper but behavior almost the same.
The USBCs negotiate 20V and charge the battery.

Workin with only a battery/touchpad connected. During the revive the board starts the CPU for a few minutes but then fails, usually with error 9 (see attached log)
Nothing stands out on thermal imagery (battery coil/mosfet, CPU phases etc - photo shows U7220/U7210 - CPUCORE_PHASE1&2).

The only short I found is on EDP_INT_AUX_N.

Would think it should still work without a display.

Am I missing something?
Any way to get data off it?

[reformatt]

If EDP_INT_AUX_N is shorted, that usually happens when backlight voltage is shorted into that line. Usually its from liquid damage in the LCD connector, but alas, that line goes to the CPU and that board is toast for internal display anyway. The 15" models have a MUX chip which saves the CPU, but not the 13" models. I've seen these boards still boot despite that.

As with all T2 boards, you need working SSD's for them to boot. Check SSD rails for 2.5V for SSD0 and SSD1 etc. Most boards have some kind of corrosion or shorted capacitor around the fan inlets, it's the most common fault I see. Also check for pry damage on the sides as well as most novices use a screwdriver to pop the lid up the sides. Never underestimate the stupidity of people.

[cruiser_]

LOL great obervation, will keep it close to my heart

Yeah I'm only after the data at this point; A little bit of an update though.

I did succeed with a revive, although it took 20+ times - had to try different on/usb connect scenarios to keep the board working long enough for a revive.

As a side question - I noticed a few e people would get rid of CPU coil(s) to prevent it from starting at all for a successful DFU/revive. How does that work though, I would think the revive process would stop if CPU is not deemed 'started'?

Anyway, I think got back to what was the original problem. The board cycles. I have got all (?) the rails and signals up and on up to the PM_PCH_SYS_PWROK.
It stays on for about 60 seconds then drops and repeats. PPVCC_S0_CPU switches as expected so CPU seems to be somewhat ok.

HOWEVER... I am not getting PPVCCEDRAM_S0_CPU.
PVCCEDRAMS0_EN_FILT_BUF is high (EN@U7710), there is one little tiny spike on L7702 and that is it. U7710 pins' diode readings check out. So do R7702's.

Therefore, there is no PVCCEOPIO_EDRAM_PGOOD.

Could it be that PMIC waits that long (almost 60 seconds) and shuts down due to this missing rail...?
or maybe PMIC just checks that the rail is OK but it's not needed at this stage?
but then why would CPU_VR_READY go high at all? I'm puzzled...

[mon2]

How high is the fan spin during the on time? Normal or able to dry your hair from across the street?

[cruiser_]

How high is the fan spin during the on time? Normal or able to dry your hair from across the street?

they run calmly, relatively steadily but go full speed for a sec just before the cycle ends / current drops.
Put in in a case ... USB pendrive only blinks once or twice. Nothing on the USBc->HDMI adaptor.

[mon2]

Aha! Sounds like a similar case with the A2141 we had a month ago. Suggest to replace the trackpad or at least the cable that interconnects with the logic board. Believe you may have the same fault.

As a test, remove the trackpad - do you see the same sequence of events? There are sensors inside the trackpad that must be at a normal state else you will have this power cycling (like we saw on the A2141).

[cruiser_]

you asked about fans so I put it in a case to see how they operate

'naked' board behaves the same way current consumption / cycling wise.

[mon2]

So highly recommend to test with a known good trackpad and fresh interconnect cable. The naked board is missing the same sensors so goes into hyper drive fan spin before power cycling. Exactly the fault we wasted far too much time on with the A2141. With a good trackpad, the sensors kept the fan spin to a normal and quiet mode and the logic board booted. We too could not make use of an external monitor due to this cyclic event nor could we toggle the keycaps LED. The logic board would power cycle before we could reach this part of the boot cycle.