820-00840 no GPU vcore.

[TCRScircuit]

Got a 00840 trolling me. Came in no liquid with a history of a dog biting it. First thing I checked was output on external. Nothing, Unplugged the screen and tried again. Nothing. This was with a genuine apple usb c to vga dongle. Checked CPU vcore... 0.94 volts. GPU vcore - 0v. Missing GTVR_FCCM. I thought it was U7100, replaced it with no change. PPBUS_S0_CPU drops to 0v every 30 secs then goes back up only near the GPU fets. I cant find a 00840 schematic so im using a 00875 schematic which is close. GPU is not shorted.

At this point, i dont know what more it can be. I'm guessing fets, but I dont want to put heat to that area if I dont have to. Anything else I can check before replacing the FETs and U7100 again?

 

[dukefawks]

Is a USB mouse lighting up? This is not looking good

 

[TCRScircuit]

I'll check for USB mouse. Part of me thinks the GPU isnt talking to U7100 to enable vcore... crap boards

 

[patrickribbsaeter]

I'll check for USB mouse. Part of me thinks the GPU isnt talking to U7100 to enable vcore... crap boards

Will send u a email now with information.

 

[TCRScircuit]

USB mouse flashes for a brief second so something is defiantly up with POST and the CPU. Im guessing its not posting because of GPU vcore being missing. Now to what FCCM means I have no idea. Im going to look around at similar intersil/renesas chips and see if I can figure out what that pin does. I know its suppose to be 5v. Maybe the FETs are messed up. I dont know, but if the fets are messed up, the GPU could have gotten killed. Hard to say on this one. Ill keep digging.

 

[TCRScircuit]

I give up. I checked the circuit high at low. Nothing. My only guess is the CPU isnt requesting GPU vcore. It has a SDA line that goes right to the processor. Anyway, this gets 4 min each side at 70c in the USC and if it doesnt work, RIP.

 

[TCRScircuit]

So, FCCM is a apple term for ZCD enable - Automatic skip mode operation (ZCD) for light load efficiency. Irrelevant because we arn't even getting core voltage. Going to look at the PWM side of things now

 

[TCRScircuit]

PWM 1 messed up. 5v to 0 and cycles. Cant get the SIC635 in the us so I will use SIC631s. Replacing all my fets plus the controller again.

 

[dukefawks]

You don't even know if the controller is being told to turn on. Serial data communication with CPU/GPU will enable the regulator. Maybe you get lucky but this is a bit more sophisticated than just measuring PWM output signals.

 

[TCRScircuit]

No Title

IT'S ALIVE!!! ....but with a cracked screen .

 

[TCRScircuit]

It croaked about a millisecond under gpu load. Board bubbled up from a internal short near the CPU and one of the GPU fets shorted ppbus to vcore. RIP.