820-00840-A not turning on

Fanfwe

Member
Hi,
I am practising with another 820-00840-A board that does not turn on.
There is no sign of liquid damage.

My charger gives 20V so I have PPBUS_G3H.
PP3V3_G3H gives me 3.41V

Then my S5 rails are all missing.
I figured that PP3V3_S5 reads 7 Ohms to ground. I suppose this means a short to ground. I know 7 Ohms could be fine for some rails such as CPU VCore, but I think this is no good for that PP3V3_S5 rail, right ?

Is there anything else I should check before I try to inject some voltage into that rail in order to find the short ?
 

2informaticos

Administrator
Staff member
That is very low value for 3V3_S5.
First lift one side of the coil and see where the short remains.
If is on the filtaer cap, then you can try voltage injection.
7 ohm may doesn't heat anything, as you cannot inject more than 4v there.
But may have luck and some chip enters into thermal avalanche.
Remove CPU heatsink before...
 

Fanfwe

Member
Hi,
So, with the coil lifted, the short was remaining on the filter cap side.
So I tried to inject voltage.
At first I sent 1V, but not much happenned.
Then I went for 3.3V. I could see it was drawing between 200mA and 300mA for some time, but while I was trying to feel heat on the board, I saw the power supply current went down to just a few milliamps.
So, looks like the short fused by itself. Now it is gone, and I have about 0.6KOhms to ground on PP3V3_S5 which looks quite better.

However, I still do not have any S5 rails coming up. At the moment, I can only see PPBUS_G3H and PP3V3_G3H coming up.
 

Fanfwe

Member
I already had the coil resoldered from before.
Pin 23 has 13 volts
Pin 12 has 0 volts
Pin 13 has 0 volts
Pin 22 has 3.3 volts
Pin 29 has 0 volts
Pin 21 has 0 volts

Looks like I'm missing the enable signals for the rails. I went on and checked the U7890 gate that controls PM_EN_PVXS5, it has power from PPBUS_G3H however its two inputs are low. One of them comes from the SSD connector, so it's probably normal as I do not have an SSD connected. The other one comes from the PMIC U7800.
Looks like that chip has two power inputs, one that comes from PP3V3_S5, but it can't be that one, since that chip must turn on the S5 rails, so...
The other power input that I see is PP3V3_LDO3V, and on that one I only get 1.8V, so something is probably wrong there I guess ? It reads 27 Ohms to ground.
 

2informaticos

Administrator
Staff member
That voltage is only used by PMIC itself.
Remove C7805/24 and check again diode mode.
If still low, change U7800...
 

Fanfwe

Member
Allright, the caps were not causing that issue.
It took me a while to get a new PMIC in order to replace U7800, but I finally got it yesterday and replaced the chip.
The board still doesn't boot, however I am now getting a lot more power rails.
In fact, all rails are present, except for PPVCCCPU_S0G and PPVCCGT_S0G. So all I am missing now, is essentially CPU Vcore.

The boards draws 240mA at 20 Volts for a little while, then down to 20mA for a few seconds, and then back to 240mA and stays there.
The CPU get just a little warm, not hot.

PPVCCCPU_S0G measures 7 Ohms to ground
PPVCCGT_S0G measures 4 Ohms to ground

I do not have a reference board to compare this to, so I'm not sure if this is right or wrong.
 

Fanfwe

Member
Yeah, PM_EN_CPUVCC is high, I forgot to mention that.
OK, so I will try to play with the BIOS.
I'll start dumping the existing one, just to be safe.

But then, what should I be doing ? Can I just pick a dump from another board just as is ? I guess this will have ME issues, but maybe it's still good enough to start with, in order to see if board will boot ? Then if that works I could look at preparing a clean image for this board. Right ?
 

Fanfwe

Member
Didn't manage to find a clean ME BIOS. I took a dump from a dead CPU donor board and flashed it on this board.
Same problem. No PPVCCCPU_S0G and no PPVCCGT_S0G. When plugging in the charger, PM_EN_CPUVCC goes high, then low, then high and remains high.
 

2informaticos

Administrator
Staff member
Something had to cause that problem on 3V3_S5 rail.
The low value disappeard, but possibly there is a chip with internal damage; that could be CPU.

VCCGT doesn't come on untill graphics load goes high.
 
Top