Diagnosing South Bridge (FH82H370) on Asus H370-Pro Gaming Motherboard

[Ross Kohler]

Hi,

I'm learning about motherboard repair and I believe I've traced the issue down for the South Bridge on the particular motherboard I'm working on. My path of concluding that:
- There's no sign of any attempt to power on when pushing the power button.
- No PSON signal is reaching the 24-pin ATX connector.
- The Super IO chip is receiving the PS_SWITCH signal from the power button and is relaying this signal back to the South Bridge.
- The Super IO is not receiving the S_SLPS#_R signal back from the South Bridge
- If I pull the Super IO S_SLPS#_R pin high, the power supply turns on suggesting that the Super IO chip is working fine.
- The capacitors around the South Bridge all seem to be fine and all have voltages that look reasonable (1,8V, 3.3V and 1V).

At this point, I've run out of ideas on what to test. I would like to dive deeper into figuring out how the South Bridge works and what I should be expecting in it's proper operation but I can't find a data sheet for it. I'm leaning towards concluding that the South Bridge died but that seems like a lazy conclusion unless I can find hard evidence of it. Can someone help me in figuring out what else I could look at?

This post here goes into some more detail about what I've been looking at and someone was nice enough to share a board view to help me.

Any help would be greatly appreciated!

[прямо]

Not surprised at all if the PCH is dead. Asus boards seem to always have a problem with overheating PCH. Got a couple Asus boards here that have heatsinks on the PCH, but when they are running, I can't even touch them with my finger.

[piernov]

300-series chipset datasheet is available but it doesn't help that much, the relevant info is left out.
First need to check power rails, depends on the board but should have a 3V RTC rail, a 3V standby rail and a 1V standby/main rail going to the PCH at least.
Then some signals, in particular the RTCRST#, SRTCRST# and RSMRST# signals (input to the PCH) and SLP_SUS# (output from the PCH).
The 32 kHz RTC clock should also be present.
Finally need to confirm if SLP_S4# is completely missing or at least pulsing.

[Ross Kohler]

Not surprised at all if the PCH is dead. Asus boards seem to always have a problem with overheating PCH. Got a couple Asus boards here that have heatsinks on the PCH, but when they are running, I can't even touch them with my finger.

I'm nearing this conclusion. The one thing that gives me hope is that the PCH doesn't get hot while it's plugged in to the ATX. Even if I force the ATX on, it doesn't get hot. Does this happen? The PCH dies without a short circuit through it?

If I touch the PCH I do think I can feel some warmth, but it's difficult to tell for sure.

[Ross Kohler]

300-series chipset datasheet is available but it doesn't help that much, the relevant info is left out.
First need to check power rails, depends on the board but should have a 3V RTC rail, a 3V standby rail and a 1V standby/main rail going to the PCH at least.
Then some signals, in particular the RTCRST#, SRTCRST# and RSMRST# signals (input to the PCH) and SLP_SUS# (output from the PCH).
The 32 kHz RTC clock should also be present.
Finally need to confirm if SLP_S4# is completely missing or at least pulsing.

piernov voltage rails look good. I've measured 1.8, 3.3 and 1.05V all around the PCH on the capacitors on top of the chip. I'll check these other signals. Thanks for the tips.

[DynaxSC]

Hi, PCH's have generally two types of failures: 1. short on any or all of the power supply lines - this is easy to diagnose, 2. functinality failures due to errors in the silicon structure - they lead to fatal failures (no boot, no start) or strange behavior of the chipset/board during boot and operation if any possible. I would estimate first failures type is 70%, type 2 is 30% approximately. The type 2 is much more difficult to diagnose, and can be finally confirmed only by replacing the chipset. Some things can be tested however, eg. If pch is reading the bios on Power Button push (scope eg. on pin 2 of the bios), clocks present on the cristal (scope with probe x10 to have low parasitic capacity), eg. voltages present on the cristal pins (0,2 - 0,4V). The supply voltage for the chipsets are usually higher than nominal, if the voltage exceeds too much the nominal, if often leads to overheating of the chipsets, and reducing of their life span. This can happen due to aging of the vrms and eg. changes in the reference voltages generated by the vrm chips, or due to bad resistances of the resistors used in divider network defining the output voltage of the vrms. I have often seen much too high supply voltages for chipsets on board, eg. 3.45V instead of 3,30V. This is enough to stress the pch and reduce the chip life span significantly, leading to failures. It seems to me that mobo vendors to not care about this issue deliberately. Another issue is that the chip quality of intel chipsets seems to be much worse than that of processors, as chipsets fail much more often than processors, which usually do not fail, if not something special happens to them.
Chipsets have also some internal vrm generating 1,2V most probably during operation only, the voltage is connected to an external cap for stabilisation, this vrm may also fail, leading to non functioning chipset, dont remember the pin/signal name now.
Also the DMI bus to the CPU can be checked in the CPU socket (use reverse diode mode, voltages should be 0,2-0,4V). This is tricky, socket pins can be bent easily).
Normal operating temperature of chipset silicon without a heatsink should be 50-55 degree Celsius. If it is higher this is also an indication for internal chipset failure.
Resistance on the 1,05V power line should be some 25-70 ohms (chip must have room temperature, resistance lowers significantly with rising silicon temperature), if lower the chipset is also suspect to internal failure. The other lines should be in single kohm range.

[DynaxSC]

Btw, check also if the bios image is correct, seen many board with corrupted bios image, not booting, or even not starting the board. Backup first the image, and then write a stock image to the bios - you need a hw programmer, eg. CH-341 (must be modded before used to not destroy 3,3V bios chips, as it has a design flaw already leaving factory - instructions to be googled). The backup is important, as each bios has some personalisation data for each particular board, eg. UUID, serial no or MAC LAN address. This data is not present in the stock image, and if bios is corrupted must be transferred to the stock bios manually (DMI data section, and network data section in Bios). Usually bios can be programmed in-circuit with a special clip (board must be disconnected from ATX power). On some boards however its not possible to program the bios in-circuit, so then it must be desoldered for programming.

And a tip, do not force power to the mobo by Atx pin shorting. This has no value for diagnosis (no board will start correctly that way) and can destroy some mobos - speaking from my own experience.

[DynaxSC]

Checked again, and the 1.2V generated by the PCH is called VCCDPHY_1P24 (Gigabyte boardview naming) - so its 1.24V exactly, I believe this has something to do with the clock generation on the PCH, but I'm not fully sure.

Regarding the sections in the BIOS I wrote about, they are called exatly "GbE region" (LAN MAC Adres is usually stored here) and what I called DMI section, is correctly "NVRAM" section located in the BIOS region.

And another hint, the most important signal from the PCH starting the whole power sequence is SLP_S3# (or N-SLP_S3 on Gigabyte boardviews). PCH should activate this signal on pressing the Power Button, where the Power Button signal is relayed usually by the SIO to the PCH only when SIO is in correct state, eg. the monitored stand-by voltages are present (VIN inputs).