A1398 - 820-00426 Fan spins for 3 seconds then turns off

[acowan]

So this board initially had corrosion around the Thunderbolt boost regulator L3095 and Q3088.

After an ultrasonic the board seems to come to life .. fan spins for a few seconds (between 3 and 5) then it's back into an S5 state.

This seems a very different problem to the 1/4 fan spin mentioned frequently in the forums.

 

[2informaticos]

Try to start in SMC bypass first.
Also try removing Q3080.

 

[acowan]

Oh wow Q3080 has disintegrated - ill clean up that mess and report back

 

[acowan]

Hmmm no change ... Q3080 gone and starting in SMC bypass makes no difference. It's running for a good 6 seconds then hard fan stop.

 

[acowan]

Probing on ALL_SYS_PWRGD and CPUVR_GOOD are falling to 0v at pretty much the same time

 

[2informaticos]

Monitor all power rails; big coils and pages 64/66 too.

 

[acowan]

Found another area of crud on the board - the hall effect pads. R5250 has jumped off the board. SMC_LID should be high yes ?

 

[acowan]

SMC_LID is now fixed .. still no change in behavior

 

[2informaticos]

You should have more corroded components.
Check all voltages, as above mentioned.

You may need to reball SMC; maybe heat PCH and CPU with quality flux too.

 

[acowan]

Going through all of this now - I may be a little while.

Wish I knew more about the start-up sequence after the S0 state.

 

[acowan]

More fun and games !

Having spent some time with the scope this is where I am.

- On start up the fan stutters, makes a spin, stops then starts up for approx 6 seconds. PP5V5_S5 drops to 0v for approx 100ms before raising back to 5.5v
- While the system is running for the 6 seconds or so all voltages are rock solid EXCEPT anything GPU/LCD related - but i'm not sure at what point in the start up sequence we would expect to see this enabled

e.g.

L8760 0v
L8710 0v
LCD_PWR_EN 0v
LCD_BKLT_EN 0v

 

[2informaticos]

Try known good BIOS now...

 

[acowan]

And for that i'm going to need to spend the ??? (or Euros) on a Medusa -

 

[2informaticos]

Don't you have a normal USB programmer???
Of course, you need to remove the chip from the board.

If you really plan to stay in business with Apple stuff, then spend that money; because time is money...

 

[acowan]

Amazing .. a bios flash worked! ... took me way too long to work out how to do this. Wrong tools for the job really.

 

[2informaticos]

Be sure the BIOS flashed had clean ME region; to avoid any future issue.
I suppose you didn't change serial number with original from back cover; fix checksum of Fsys area in BIOS is required in such case.