Проблема с POST/загрузкой Gigabyte B450M S2H

Hi,

So I got a faulty Gigabyte B450M S2H motherboard off eBay. It was listed as having a potentially faulty CPU socket. When I got the board, I went straight to checking the CPU socket under a microscope with the plastic cover removed to verify if it was actually damaged. None of the pins looked broken or damaged.

So I started troubleshooting the board to see what's actually wrong. The 1st issue I found was there was 0.7V at the power button header. I found the cause was the 6-legged ESD chip that was next to the header. So I removed the chip which brought 5V back to the power button header.

I then tried to start the board and got no reaction. I checked the usual things like RSMRST, SLP S3, SLP S5, standby voltages and checking to see if the SIO was seeing the power button press. All checked out. All I found was PS_OUT wasn't dropping low when pressing the power button and the SIO_18V pin was reading 3V. 2_5 LEVEL was also reading 2.9V. Nothing else looked wrong with the SIO. There was no shorts or anything like that.

None of the main power rails had shorts. So I suspected the SIO was the issue. I wanted to be sure it was the SIO before replacing, so I kept checking but was getting nowhere. I decided to take an SIO off a donor board (a Gigabyte B450m DS3H V2) and soldered it to the B450M S2H board. The board uses an ITE8686E DXA and the donor I took was also an ITE8686E DXA.

After doing this, I checked all pins (giving them a nudge test and reflowing any that moved). The board now responds to the power button. So the previous SIO must've been faulty.

Now there's another issue. It seems that the board is not able to boot into anything. If I connect my SATA SSD or a bootable USB device, the Gigabyte splash screen just flickers on and off. If I remove the SATA SSD/drives or spam the F2 key to enter the BIOS, it goes into the BIOS and stays on. It's only when I try to go into Windows or similar that it stops working correctly. It's like it's resetting or something.

Another issue is when I do go into the BIOS, all sensor readings are off. It thinks the CPU voltage, RAM and VDDP voltage is at 3V. It thinks the 5v rail is at 7.8V and the 12V rail at 18V. All temp sensors show -1C.

From what I know, these sensor readings come from the SIO. So my suspicion now is the there may be an issue with the donor SIO I am using. I've tried reflashing the BIOS as it's a quick thing to do, but it did not help.

I notice the VCore voltage jumps between 0.9V and 1.4V. This seems to match the flickers. When in BIOS, VCore is at a stable 1V.

What I mainly want to know is does this sound like I maybe have another SIO problem and need to replace it again, or is there potentially another issue? I don't want to replace the SIO again if it's not needed. Not really sure where to go from here.

Appreciate any help.

SIO is likely ok, but probably some resistors missing around SIO, which are in voltage dividers on the analog measuring inputs of the SIO. that's why you are getting strange sensor readings. You either blown them off during SIO replacement, or they were knocked off before. Check all smd's around SIO for completeness to board view. Other components might be also broken, as your board probably got a electrostatic discharge, when ESD was shorted.

I've checked around the SIO and there's no missing components. Some areas aren't populated (can tell by how the pads look). I've looked at images/videos of the board online to try and compare. I don't see anything that's missing.

I have noticed that every time the board starts, it goes to a "CMOS has been reset" message. I tried changing some settings in the BIOS and rebooting to see if it saved anything. It did not. The message reappears after every boot and whatever I set gets reverted. If I connect a bootable device, the board just keeps resetting (gigabyte logo flashes on/off) as it tries to boot to the device. Basically all sensors are not being read. Fan shows 0 RPM even though it's spinning. All the board can currently do is sit in the BIOS. It can't boot to anything, save BIOS settings or read any sensors.

Under a thermal camera, if I have the CPU fan connected to the CPU FAN header, the SOP8 chip next to it gets really hot quite quickly. It quickly rises to around 75-80C where it settles there. I did try booting without the CPU fan connected to the header and the chip didn't heat up, but it made no difference to the boot issues. I also tried putting the fan into a SYS_FAN header to see what the SOP8 chip next to that did, it doesn't get as hot.

Something's really messed up on this board.

Tried another donor SIO (ITE8686E 1846 CXA) just to see if it was maybe the SIO I soldered. Unfortunately, the same exact symptoms occur. Really stumped here since there's no sensor data, no OS boot and the CMOS seems to keep resetting every time.

I have also found the 2nd RAM slot doesn't work (board only has 2 slots, so 1 works and the other one doesn't). I've thoroughly looked over the CPU socket pins under a microscope to ensure none are bent. They all look normal. I can't see any damage to the pins.

Seems like this board has a lot of issues. I have also noticed some brown residue on the board which I cleaned off.

Maybe the bios image is corrupted. Try to backup the image with a hardware programmer, to save the DMI data and MAC network card address, then program a stock image. Otherwise if this will not help, I would say the chipset is partially damaged. Try to boot from an USB stick, or if you have a M.2 pcie extension card, from such a card. Anyhow no boot device means internal PCI lanes not working or bios does not configure the board correctly. The pcie lanes are usually driven by the chipset, except the large pcie slot closest to CPU, which is driven by CPU directly, but still managed by the chipset too.

Mobos after electrostatic demages are often very difficult to repair. Usually there are several things damaged. Once I almost repaired such a mobo so it did finally start to POST and booted into Windows, at the cost of replacing 10 IC's including SIO and chipset. However despite that effort it was not stable and crashed often. Finally I gave up repairing it. Sometimes it's just not worth to spend the time.

Yeah, I know it's probably not worth the effort/time. However, I do like the practice and want to try and fix it. I'd only really give up if it's something like the chipset which I can't replace currently (don't have the tools to start doing it). When I got the board, it was completely dead so I am happy that I at least got it to turn on and display something. Before giving up, I want to be sure it's something unfixable like the chipset. Otherwise, there's maybe still a chance I can at least get into Windows and restore the sensor data.

2nd RAM slot issue has been fixed. Found some damaged traces near CPU socket. I restored them and 2nd slot is OK now. So this issue wasn't related to the initial problem I was having. It's possible I accidentally damaged them when trying to remove the CPU socket plastic cover. Luckily they could be restored.

So now only issue is no/wrong sensor data and inability to boot into anything other than the BIOS. Also I can't save any BIOS settings and every time I get a CMOS has been reset message on boot (both reboot and full power off).

I tried both my SATA SSD with Windows pre-installed and then a USB stick with the Windows 11 installer. The board does not like it and just continues to reset. All I see is the Gigabyte splash screen going on and off if I try to get it to boot into something (like my USB stick/SATA SSD). My keyboard LEDs also turn on/off (matching the flickers). VCore is jumping between 1V and 1.4V. This is only when I try to boot into something. If I go to the BIOS, it stays on.

Before replacing the SIO, I did a BIOS flash with my CH341A programmer. I saved the original BIOS, but flashed the F67 BIOS from Gigabyte's website. After replacing the SIO and getting into the BIOS, I did try flashing the BIOS again which was successful. Still same issue though.

Could it be the BIOS chip itself? I was wondering if it's possible for it to be faulty and causing problems even if it was able to read/write data using my programmer.

Very unlikely. Must be something else. If tracks were cut, so maybe it's worth to check, if there are no other mechanical demages, eg. knocked off smd's. Usually such demages are near the mounting holes and near both ends of the main pcie slot, or on the bottom side of the board. Also the socket might be slightly damaged, unfortunately it's not visible on AM4, unless cover is removed, but that's tricky to do. I wrote about how to do this in some neighbourhood post from recent days.

I did try another BIOS chip (took it from donor and flashed BIOS to it). Did not help as expected, but wanted to rule it out just in case. I have inspected the CPU pins with the plastic cover removed. I found it a bit difficult to get off. It uses a Lotes socket. I think trying to remove the plastic cover is what caused the trace damage to the RAM slots. I looked online how to remove the plastic cover with minimal damage but kept coming across how-to's for the Foxconn socket.

But yeah, there's no bent pins on the socket. I did see some scratches on the board and also some exposed copper in a few places. However, they do not look broken, just the masking has been scrapped off. For traces that looked potentially broken, I grinded each side and tested with my meter to confirm there's still continuity. So far all traces that have scratch marks nearby or exposed copper are fine.

The way this board POSTs even feels weird. On other boards I've worked on I always get some screen appearing that shows you a bit about the system (e.g. new CPU installed) before you're sent into the BIOS. I don't see anything like that. The board just jumps straight into the BIOS with the "CMOS has been reset" message if I have no bootable device hooked up. If I do have a bootable device hooked up, it just goes on a reset loop with the Gigabyte logo going on and off. I can spam the F2 key to enter the BIOS so even though it's flickering on/off, it is able to read my keyboard button presses. The "CMOS has been reset" message still appears before it lets me in the BIOS menu.

Not sure what issue to focus on 1st. The fact the CMOS resets all the time and BIOS settings never save because of it seems like the bigger issue. I've checked the CMOS battery and it has voltage (3V). I've checked that this voltage is reaching the SIO and everywhere else it's supposed to. I don't see any problems with the CMOS/RTC circuit itself. I have a boardview I am looking at but not sure what else to check for.

I would get back to the issue with wrong voltages sensing. This too high voltages might interrupt the configuration of the board by the Agesa firmware (part of Bios executed before handing over to uefi bios part, low level configuring the hardware devices on the board). Such interruption could leave the board not fully configured, what would explain no boot devices and other weird effects you have. SIO should have some reference voltage input derived probably from 3.3V. The ref. voltage, if not exactly as necessary, can lead to wrong voltage readings. Other possibility is that some resistors in the voltage dividers lost their correct values due to the electrostatic discharge energy or are mechanically cracked in two parts, what is very difficult to see, even with magnifier glass. Finally, not very likely, but some of the ground pins of the SIO might have broken GND connection, so the SIO or part of it hangs in the air, has no solid ground.

Another possibility is that the connection between SPI chip and CPU is partially interrupted, so CPU can read the chip, but cannot write to it. This would explain the always CMOS reset effect. The BIOS parameters are set and then written back to NVRAM region in the BIOS by the CPU.

The BIOS does see my boot devices. I just can't boot to them. I came across this https://www.reddit.com/r/gigabyte/co...dings_in_bios/ and other similar threads on Reddit about similar issues on various Gigabyte boards. The thread I linked is exactly how the voltage and temps reading show on my board. Looks to be the same values too. A comment there did say there could be an issue with the VIN circuit. So I looked at schematic and see VIN0-6, VREF, SYS_TEMP, PM_TEMP and CPU_TEMP relate to the hardware monitoring circuit. These all go to the SIO.

So I applied power without GPU to take some voltage measurements at the SIO pins. Probably don't need POST to get the right voltage at those SIO pins. GPU would be in the way so hopefully it's not needed to get the right voltages.

These are what I measured on each pin:

VIN0 = 1V

VIN1 = 2V

VIN2 = 1.973V

VIN3 = 2V

VIN4 = 1V

VIN5 = 0.925V

VIN6 = 1.130V

VREF = 2.8V

SYS_TEMP = 1.37V

PM_TEMP = 1V

CPU_TEMP = 2.7V

Not sure what each are supposed to be. Not sure if all VIN pins should be 2v or only some. That person that reported the same issue on the linked Reddit thread also said they're in a boot loop. So I am guessing the real problem are these voltage/temp readings in the BIOS. So maybe there's an issue with one or more of those voltage readings.

The VIN0-6 voltages look normal, they should always be between 0 and 2.0V, this is the input voltage range of the SIO A/D converters. That's why there are these voltage dividers before the VIN inputs, they normalize the voltages to the input range of the A/D converters. Then the BIOS converts these readings from SIO again to correct voltages by software to display them. I'm not sure what the other voltages should be, best would be to measure them on a working board with same or similar SIO model. The SIO's are very similar between the models, often the pinout is even the same or almost same. Most important should be Vref. This voltage gives the reference level for the A/D converters, a wrong level will directly result in an error in measurement.

I put the original SIO back on the B450M DS3H V2 board I have. I compared the voltage readings and they are exactly the same. So VREF being 2.8V is correct for this SIO. The sensors/temps read fine on this board. So the issue with the S2H must be somewhere else.

What would be the next thing to check?

I seem to have found the issue. I checked the LPC area on the SIO with pins such as LFRAME, LPC RESET and LAD0-3. I 1st checked voltages on the good board and then compared it against the non-working board. I found LAD2 had no voltage. So I looked around to see where it went. I saw it only really went to the CPU, so I checked the CPU socket area and found the LAD2 trace was broken. So I repaired the trace, did a quick power up and saw 3.3V on LAD2.

Then did a proper boot and it went into the BIOS. Now sensors read properly. I was able to boot into Windows which confirms the sensors were the reason it was not booting.

So I powered off and decided to re-solder the CPU FAN IC (had taken it off earlier when I was testing to see what happened because it was hitting like 80C very easily). Upon doing it, board started and everything but for some reason the IC burned within like 10 secs or so - probably when it tried to get into Windows as it now could. I powered it off immediately after seeing the smoke. So I took a donor IC and soldered it. CPU FAN was dead. Thinking I maybe killed the CPU FAN, I tested it in the SYS FAN header and it worked fine there.

So looked around and found the FANPWM pin for the CPU FAN IC was reading only 200 ohms which looked off compared to the 1KOhms I was seeing on the SYS FAN IC. Checked around and started suspecting the SIO. So I lifted the pin up to disconnect it and immediately saw the resistance went high. Then checked the SIO pin itself against its ground, got 13 ohms which confirms this pin is now shorted.

Not sure how the IC burned, but still. It's possible I accidentally put the wrong IC (don't think I did) on or it just finally failed as I did keep seeing it heat up to 80C quickly while I was troubleshooting the sensors issue. Not really sure. It caught me by surprise when I saw the smoke coming out.

With the SIO pin lifted, the CPU FAN header now works but can only run at max speed since there's no PWM signal any more. Bit of a shame this happened, but I'm at least satisfied that I got it to at least boot into Windows. I haven't checked if the BIOS saves settings yet. That's the next thing I'll probably check when I can. I take it there's no way to make the CPU fan header adjust its speed again without replacing the SIO?

You already know the answer. Anyway congrats to the success.