820-3787-A mid 2015 DG power cycle

[eclipsevl]

Hi,

I've got a 820-3787-A board which is stuck in power cycle. PP3V3_S5 is fine. Got orange led on the magsafe plug.
With multimeter I could see that PP1V5_S0, PPVCC_S0_CPU, PP1V5R1V35_GPU_REG_R, P1V05_S0GPU_REG_R, PP1V05_S0_REG_R, PP1V35_S3, PP5V_S4 are trying to start up, there is a voltage for a fraction of a second.

Couple of times after powering the board up, I saw PP1V5R1V35_GPU_REG_R at 1.5V and fans started to spin. It happened twice.
Another two suspicious things:
1. There was a blown ceramic cap at PP1V5R1V35_GPU_REG_R line, however without making short circuit. After removal I found that pad connected to PP1V5R1V35_GPU_REG_R was not making contact - pcb pad was ripped off. Cap itself was shorted. Didn't look like it was mechanical damage.
2. SPI flash was soldered - there is residue of a flux. Not sure if it is relevant to problem with startup sequence.

Resistance at PPVCC_S0_CPU is ~35 Ohm, P1V05_S0GPU_REG_R ~200Ohm, PP1V5R1V35_GPU_REG_R ~150Ohm. Could not find any visually bad component/short circuit.

Need some help with further debug.

 

[2informaticos]

First of all, welcome to the forum!

Be sure to correctly rebuild the broken trace; where you've found the shorted cap.
Test if machine turns on in SMC bypass mode.
Check all SPI Bus Termination resistors and their traces.
PLT_RESET_L pulsing too?
Just in case, flash known good BIOS, clean ME.

 

[eclipsevl]

Thank you!

> Be sure to correctly rebuild the broken trace; where you've found the shorted cap.
There were no broken traces, looks like cap was connected only to via.

> Check all SPI Bus Termination resistors and their traces.
There is 3.3k resistor at CS, no external resistors on other lines.

> PLT_RESET_L pulsing too?
Yes, there is a pulse.

> Just in case, flash known good BIOS, clean ME.
Ok, this is new to me. I will figure out how to program SPI flash, but have no idea where to get bios image. Hope it is possible to google it somewhere.

 

[eclipsevl]

Ok, got weird results in SMC bypass.

1. First attempt it started right away, fans spinning 100%
2. Second attempt - It started at like after ~4-5 cycles (power up cycles).

PPVCC_S0_CPU = 1.8
PP1V5R1V35_GPU_REG_R = 1.5
P1V05_S0GPU_REG_R = 1.05
PP1V05_S0_REG_R = 1.05
PP1V35_S3,
PP5V_S4 = 5

PPVOUT_S0_LCDBKLT = 0 (should it be high in SMC bypass?)

After that - it always need several cycles to start up. Feels like there is a bad cap somewhere.

 

[2informaticos]

You've said ripped off capacitor pad.
Not sure if was in the middle of the via; was the reason I mentioned trace continuity checking.

Starting in SMC bypass suppose a problem with a SMC sensor (pages 44-46); check AVREF_SMC first.
When started, check for USB activity; don't waste time checking backlight yet.

A1398 820-3787-A Clean.rar

drive.google.com

 

[eclipsevl]

>Starting in SMC bypass suppose a problem with a SMC sensor (pages 44-46); check AVREF_SMC first.
AVREF_SMC = 3.3V

>When started, check for USB activity; don't waste time checking backlight yet.
Did not check USB activity yet; checked activity on SPI CS line - always high.

New observation: even without SMC bypass board eventually starts up for ~5 seconds (CPU, GPU voltages), shuts down. Then cycle repeats.
The only difference with SMC bypass: after startup it does not shuts down.

Not sure if it says something.

Thank you so much for the bios!

 

[2informaticos]

Did you try the BIOS file?