Acer Spin 5 - Lenny_ICL (19771-1) -- Dead - PM_SLP_S3# held low

[cappernicus]

Laptop model: Sp513-54n-74V2


Testing power rails (separated by chip) reveals:

19V_DCBATOUT - (active)

5V_AUX_S5 - (active)
5V_S5 - (active)

3D3V_AUX_S5 - (active)
3D3V_S5 27Ω - (good) - enabled by Nuvoton chip

1D8V_VCCIN_AUX - (good) - enabled by 1D8V_S5 PG signal -- (dependent on 3D3V_S5)

1D8V_S5 41Ω - (good) - enabled by 3D3V_S5 PG signal

1D05V_VNN_BYPASS - (good) - enabled by 1D8V_VCCIN_AUX PG signal -- (dependent on 3D3V_S5)
1D05V_S5_BYPASS - (good) - enabled by 1D05V_VNN_BYPASS PG signal -- (dependent on 3D3V_S5)

1D1V_S3 160Ω - (disabled) - enabled by PM_SLP_S4# (from processor, detected by Nuvoton chip)
1D8V_S3 - (disabled) - enabled by PM_SLP_S4# (from processor, detected by Nuvoton chip)
0D6V_VREF_S0 - (disabled) - enabled by PWR_VDDQ_PG signal from 1D1V_S3 and 0D6V_VREF_S0 above

1V_CPU_CORE 200Ω - (disabled) - enabled by PWR_VDDQ_PG signal from 1D1V_S3 and 0D6V_VREF_S0 above -- (dependent on 3D3V_S5, and disabled by PM_SLP_S3#)

Load Switch - 5V_S0/3D3V_S0 - (disabled) - enabled by PM_SLP_S3#

Load Switch - 1D05V_VCCST - (enabled)
Load Switch - 1D05V_VCCSTG - (disabled) - (dependent on 3D3V_S0)
Load Switch - 1D8V_PCH_C10 - (disabled) - (dependent on 3D3V_S0)
Load Switch - 1D1V_VCCPLL_OC - (disabled) - (dependent on 3D3V_S0)

Errata: There are two mistakes on this schematic that I found. 1D8V_S3 is powered by 5V_S5, and 1D8V_S5 is powered by 3D3V_S5, not by 19V_DCBATOUT as noted on page 2.



Essentially, everything is dependent on those things in bold, and I cannot find anything wrong with any of the power rails, no apparent shorts, with one possible exception - the gates on both high side VCORE mosfets only have 1ohm resistance to source. I don't know what this should be but I know some regulators have internal pullups/pulldowns of similar value. All other mosfets test as expected.

As for how the board died - I was repairing the laptop which had no display and discovered it had a loose cable. After firmly seating the cable at the LCD everything worked fine, but during reassembly I had to unplug the other end of the cable from the motherboard. It caught going back in and went in slightly crooked. At that point I smelled the slightest bit of magic smoke (battery was connected) and it would no longer boot. To make matters worse, When I first started looking for the problem I'm pretty sure that VCORE had power, but now it doesn't. I can only assume that I accidentally bumped something with a probe tip somewhere and fried something else.

When I plug in the adapter (with or without battery) the 3D3V_S5 rail comes up for about 8-10 seconds, then shuts off. If I then press the power button it will pulse 3 times for ~3 seconds each, with 1 second between each. During each pulse the other rails turn on, which is why they are marked "good". Obviously it is trying to start the computer, but something doesn't work right so it shuts off. As far as I can tell that something is the PM_SLP_S3# and PM_SLP_S4#, but what I don't know is why. During each 3D3V_S5 pulse they attempt to go high, hovering at ~100mV, then drop when it shuts off, so something is holding them low. Both are driven by the processor/PCH and measured by the Nuvoton (and both have high resistance to ground), but PM_SLP_S3# is also connected to a few other chips which could be bad but I cannot find anything that seems unusual. The lowest resistance to any other pin on any of those chips (except the Nuvoton which I cannot easily measure) is 7.5k, and all the diodes connected to that line test normal as well.

Given that all the power components test ok and none of the mosfets are shorted, would it probably be safe to externally supply the missing PM_SLP_S3# and see what it does?



Files for this board are here:
https://www.badcaps.net/forum/troubl...ny-icl-19771-1

This is the first time I've done in-depth diagnosis of a motherboard before, as I didn't know laptop schematics and board files were obtainable until now. I'm not familiar with computer power sequencing or PCH/processor architecture except what I have pieced together from this analysis, so help is greatly appreciated. And I want to learn, so please support your claims. If you say my PCH is dead, explain why.

Thanks!

[cappernicus]

No takers? I've spent a couple weeks puzzling over this thing, tracing and measuring everything I could think of, but I just don't know much about computer architecture and signal sequencing.

[mon2]

3D3V_S5 is 27 ohms to ground? Something is wrong here. Too low of a resistance on this power rail.

Remove all power. Disconnect the battery. Meter in diode mode. Red meter probe to ground.

Black meter probe onto the battery pack's smbus scl and sda lines. Post each measurement. Testing to see if the smbus lines are shorted or not. If a low measurement, check for nearby esd TVS diodes which may be shunting one or both of these communication lines.

[cappernicus]

I thought so too, so when I discovered that I measured everything I could on that rail. I obviously couldn't measure the capacitors (unlikely), but every single resistor matched the datasheet and all the diodes and transistors tested good. That leaves the chips. The ones I can test via indirect means test ok, but I can't test most of them without removing stuff.

SCL measures 0.532V
SDA measures 0.478V

I tested all the TVS diodes, and for good measure I tested all the other diodes as well as the dual FET packages and I found two anomalies. One TVS line connected to THDB2_USB20_CON_N is 184ohms to ground, and one of the dual FETs (Q7405) connecting (CPU_I2C_SDA_SML1) to (CPU_SMB_SDA_P1_Q) has a diode drop of ~0.1V (~95ohms to ground and ~120ohms to 3D3V_S5 at its gate) in either direction on pin4 (CPU_I2C_SDA_SML1). This last one is odd though because when I tested it a couple days ago I noticed a low resistance on that pin, so I re-checked all the pins again just to be sure and compared it to another nearby, but when I came back to it it checked out normally. I just figured it was either due to capacitors charging/discharging as I tested things or I just messed up the first time. Now I'm not sure.

EDIT: Confirmed - it is not being pulled high like its SCL counterpart. I am assuming that since both of these are SDA lines the TVS is to blame, but from the looks of things could the TVS be the sole problem or do you think the processor/burnside-bridge may be damaged?

[cappernicus]

Update: I'm pretty sure I found it, but I need confirmation.

I don't know why I associated the TVS with the SDA in my last edit; they aren't connected. Probably because I wrote it at 4:00AM... :/

I pulled the two parts in question, and tested both them and the board. As I suspected (after I got some sleep) they are perfectly fine. The shorted line on the dual FET (CPU_I2C_SDA_SML1) is only connected to 3 relevant things, the CPU and both burnside-bridge chips. The short on the TVS, however, disappeared completely when I pulled it off. The chip tests good, and the pad shows no short (before I pulled the dual FET). Then after testing several other things the short on the dual FET disappeared as well. I re-installed both chips and both apparent shorts are still gone, with no change after applying power. Looks to me like a floating gate somewhere that was causing intermittent shorts. However, 3D3V_S5 is still ~27ohms, and there was no change to the voltage drops on the battery SMbus lines.

The real clincher though, is that following your logic I tested the touchscreen SDA line on the LCD connector and found that it is 2.3ohms to ground! No change after removing and re-installing both chips. And the only things it connects to is a resistor to 3D3V_S0 and the CPU. So the I2C line in the processor is fried. Possibly linked in the processor in some way to explain the floating gate symptoms.

Is my logic sound?

Given what I described, Why did you recommend testing the SMbus? Trying to learn.

Is there anything else I can test? Would it be pointless to try limited current injection on that line just to verify that the CPU is at fault (I have access to a thermal camera)? I know it would probably damage it more if it is, but at this point there's no repairing it anyway right?

[mcplslg123]

IS the bios chip,pin8 is powered by 3D3V_S5? If yes, then this low resistance to GND on 3V rail is almost a sure indication of shorted PCH part of SOC.

[cappernicus]

It is indeed, but why is that a strong indicator?

Not to be rude or doubt your experience, I just want to learn as much as I can from this.

[mcplslg123]

A shorted 3V rail is a nightmare on SOC platform.
I've observed that when bios chip pin8 is not powered by 3D3V_S5 directly,pch survives most of the time but gets shorted when its the other way.
These observations are not out of any textbook or guide and comes from experience.
However, you can consider other's opinion. If its me with this board,i wont waste my time any more.
Findout, if you have the same resistance to GND of 27 ohms on another coil like 1.05V/1.8V

[cappernicus]

1D8V_S5 is 41ohms to GND, The two others are 500 and higher, and the two 1.05V rails are 1k and 13k.

Experience is just as valuable as textbook knowledge.

Shorted rail(s) aside, What do you think of the sleep signal problem I asked about in my first post? I would assume that the SOC is detecting a problem (perhaps with the I2C communication) and keeps the system in sleep, except that I do see some voltage on those lines at the right times which I think means they are trying to turn on but something is holding them low. However, the resistances are fairly high on those lines, so are aren't apparently shorted. What do you think would happen if I supplied those missing signals?

[reformatt]

@mcpslg123 is 100% correct.

Even with a laptop powered off, you still have the main power rail present on the LCD connector. So if you have either battery or adaptor connected, regardless of off condition, that connector is considered live. Backlight control is done via a control signal to the inverter within the LCD, so from a designers point of view, there is no need to switch 19V on/off to the LCD connector. They are not designed to be hot plug.

So it depends on the pinout and design as to whether hot plugging this connector will create a problem. In this case, you have eDP_BKLTCTL_CPU on pin 4, 19V on pin 1 and 2. eDP_BKLTCTL_CPU goes directly to the CPU, so you likely put 19V up this line when it sparked. Game over. It some cases, the data line is shorted and the CPU survives, or it dies.

I always disconnect the battery before any disassembly or assembly. I don't know how many times I've dropped a screw off the end of my screwdriver onto the board. Having no battery connected is what saved my arse from creating magic smoke.

[mcplslg123]

1D8V_S5 resistance to GND is also quite low .So now i'm almost sure,its game over.