HP RP7 RP7800 from 2012/2013 worst fault - has "character"

[Rayblaster]

Hello friends, I've got a nice toy to play with - it is an HP RP7800 All-in-one POS system. I got it for free and it has one problem - it turns on very rarely, and just for a few seconds. After that it dies and doesn't come on if I don't leave it unplugged for at least 2 hours. I've repaired motherboards before, this is not my first one, but problem with these types of faults is that everything you check on the board will usually be fine. This is the case for me here.

I dissassembled it completely, and got my hands on the motherboard. Model No. is PIQ67H/Higgins 11047-I 48.3GD02.011. Has a socketed desktop Celeron G518 and uses laptop DDR3 SODIMM (2 slots).

Equipment I have right now - Cheap electronic microscope, rather good adjustable lab power supply and a few multimeters. Soon I will have oscilloscope and thermal camera (don't need them for now at least).

This motherboard uses 19.5V 180w HP charger/adapter with center pin (big barrel connector). I double checked it on a working HP laptop that indeed it is good and charger is detected as 180w. Power adapter problems out of the question for now. Tried different, known working CPU and known working RAM - no changes in behaviour. Changed RTC bat with new one also.

If I try to supply 19V from my bench power supply nothing comes on and draws 0.01A (10mA) (there is usually 2 leds that light up in sequence) it needs the signal from the center pin. If I dont use the HP adapter one of them lights but the other led doesn't, so clearly I have to use only the HP adapter if I want to start it.

Of course as I expected there is nothing shorted on the board (it is working intermittently after all). There is always 3V on the power button and 3V goes to 0V when pressed. This means EC/Super IO got power. What is interesting is that there is no voltage on BIOS chip. BIOS chip is mx25l12872f and needs 3.3V on pin2. I know some boards power BIOS chip later in power-up sequence. Super IO is Nuvoton npcd379hakfx. I could not find any datasheet, but I see it gets exactly 1.8V supply. What is even more interesting is that sometimes when I plug the power to the board for a few seconds I DO have 3.3V on BIOS chip and board actually tries to start. Something is cutting off the 3.3V supply for BIOS.

Tried to inject 3.3V there - no shorts, about 20mA current draw, but board still dont start.

My logic told me to start probing the power management ICs and I got to a TI TPS51220A. Everyrthing there looks normal all LDO and reference voltages are present.

As per datasheet there are two EN signals, they should be over 0.5V and less than 6V. EN1 is missing, EN2 is there always at 2V. Tried to inject forcefully 2V on the other enable signal - something sinks 20mA and I don't like it.


Maybe some chip pull it intentionally low? Even with this huge current for enable signals, some more 5V supplies started to show up around the board, but board don't come on and I hear high pitched click noise from the TPS chip itself /it also heats a little/. Still no 3.3V on BIOS chip.

Today I tried to start the board and it actually started for just under 1 minute with picture output and everything working! While working, I measured stable 3.3V on BIOS and 3.3V ot the previously missing EN1 signal on TPS. I, with my finger also did not find anything abnormally hot on the board while working.

Board shut down and same story again. This board has character!

After the holidays I will buy a freeze spray, because I suspect that this board is working only when cold and some component got very reactive to temperature.

I couldn't find any schematics to this board, they would definetly help track what is happening in the early power sequence.

A little tiny voice in my head tells me it is a defective PCH, I can't explain why, it is just a feeling with these types of faults.

Best wishes to everyone here!

[Rayblaster]

Anybody?

[rogfanther]

So, can you post some pictures of the board ? Does it have electrolytic caps that could be going bad ?

[Rayblaster]

Here are photos of both sides. I am sure there are no bad caps. I tried using freezing spray all around the board, but to my surprise behaviour does not change in any way. I have to leave the board unpowered for hours so it will power up for under one minute. Again, nothing heats up abnormally while working or not working. A thermal camera may be actually beneficial, because it is much more sebsitive than our skin and may show some abnormalities to give at least some direction where is the fault.

What I found is actually interesting. Under the EC are some bipolar transistors that are current-driven. There are always 3.3V on their collectors and if they are opened 3.3V goes directly to BIOS chip through their emitters. The signal that comes to their bases is through a 150ohm resistor. When the board is working there is 3.3V before this resistor. When it is not - no 3.3V, no current for the NPN transistors gates. These NPNs are clearly used as electronic switches and nothing else.

Now, the very interesting part - if I inject 3.3V there with the AC adapter OFF I get a current of 0.5A and the PCH slowly gets hot on one side only. If the AC adapter is plugged in and ON current draw is about 0.022A which is the gate current of the transistor.

With 19.5V AC adapter OFF I tried heating the PCH with hot air to ~200degC while injecting the 3.3V before the gate resistor and consumption suddenly drops again to 0.022А as if the AC adapter was plugged and on. I suppose this is the fault and is indeed in the PCH. Am I on right direction or is this normal behaviour of some PCHs? Like if they intentionally pulldown 3.3V signal to ground when they want it disabled? But is this protection-like mechanism (opens circuit) normal also then?

What I am now thinking is replacing the PCH as a last try to fix this board or write it out for spares (as long as I find a good one, can buy multiple from multiple places, since they are not too expensive and I do not do this for any profit, its for fun). I could not find any schematics anywhere, even asked on some paid places. Looks like these were never leaked which is very unfortunate and sad at the same time. If anyone has them it will be great if they can share them here. This board was probably made at Foxconn since I saw their markings around the metal chasis inside. It is a high quality board with excellent factory soldering for sure.


I would be grateful if anyone can at least give me tips what to check next before jumping into the PCH rabbit hole.

Best wishes to everyone!