Not a helpful answer, but it'll vary from board to board. On my KA7 the lowest resistance is on the output of the CPU voltage regulator with the CPU in place, 5.1 ohms. Make sure you take the resistance measurements with everything unplugged so you can know if the problem is with the motherboard. For example, on my KA7 the 3.3V on the ATX connector reads 10.6ohm, but that's actually through the PSU. If I unplug the power connector and measure the pins on the motherboard then it turns out they're not connected (motherboard uses its own 12V -> 3.3V regulator).

6.8 ohms doesn't sound completely unusual and doesn't sound like a short to me.

The PSU fan starting and stopping sounds like it's cutting out. Could be over current protection or could be ps_on signal from the motherboard turns off for some other reason. I'd probably start by unplugging everything from the motherboard then checking the resistance to ground for each of the power connector pins. Then put in CPU, RAM and VGA and recheck the resistances before plugging the PSU back in.

Ok, thank you on this measurement. But I have to explain that the measurement must be done with the CPU in, since the board with the issue is a soldered in SoC. So the measurement I took on another board (Athlon XP chip in) is 250ohm on the 3.3V rail. I'm hoping that since it is a much later ITX board (2014), that there isn't much on 3.3V, therefore high resistance.

The PSU I used for testing is a known working supply, so if it is an overcurrent protection, it is possible. But if low resistance on the 3.3V rail is possible, I will just have to go about this a bit different, since this test is not conclusive. I think next step is to pull out my oscilloscope and check the lines for the PSU when powering on. I'm not sure about PSU operation in PC so I will look it up.

It appears PS_ON is cycling off-on-off-on. I don't see anything wrong with the supply voltages.

the3dfxdude wrote on 2021-12-11, 22:32:

It appears PS_ON is cycling off-on-off-on. I don't see anything wrong with the supply voltages.

Hi! i have similar problems i have a asus b360 i motherboard that has low resistance on the 3.3v rail more particularly 31 ohms is that normal evry other mother board that i have has like 2 Mohms

Well, let's be careful. In my case it can still be a false lead that there is a short, because PS_ON is cycling, I can see it on my scope. I also see a power good signal. On another angle there seems to be a large number of gigabyte motherboards that have this issue over the past decade, which is attributed to a BIOS corruption. So I am going to get some tools to try to help me diagnose it. The various BIOS reset options don't work, but I think those kind of suppose the gigabyte BIOS code has to execute some self checks in order to do some of the reset or dual bios switching (where there isn't a physical switch).

In terms of design, which we don't know for sure what it is, I don't think we can be sure low resistance is a short. It's not a zero ohm short. The caps in question appear to be directly connected with the 3.3V line from PSU (doesn't go through voltage regulator). So that is why my early question on what could be the failure mode of solid caps, or are they very "leaky"? I only have one other motherboard with solid caps, and it is a gigabyte, but it is much older and still works perfectly and I use it alot so I don't want to pull that out.

Do you have solid caps?

For the bios chip, if it is an spi flash, there is the CH341A . I have a Zotac NM10ITX that after a cleaning didn't want to start, but I think it powered itself off after two seconds. I had to re flash the BIOS.

I was planning to use my tl866ii+. The manual says it supports 1.8V. I don't think I need a level shifter for it. CH341A does not support 1.8V, and needs a level shifter in between.
The chip on the board measured a 1.7V operating voltage, and the spec confirms it:
https://www.mxic.com.tw/Lists/Datasheet/Attac … 4Mb,%20v1.5.pdf

The tl866ii+ is a lot better, only mentioned the ch341a, because it's cheap, and I used it some times.

So I'm not quite ready to flash the bios, but with some basic probing, I think the bios chips are ok. They could be corrupted, but I won't know until I build a proper adapter to my tl866ii+ to handle programming reliably. My tl866ii+ recognizes both chips on board.

I tried sticking in a POST card into the PCI slot, and no code is present at all. I think the board is not sending the port 80 writes to the PCI bus? Or the CPU is truly dead. I don't think it's absolutely certain from this info. This board might not be old enough for my POST card. It's an old board, but not really that old. There is a DEBUG_PORT on it, but it's clear if it is used for such, that getting a diagnostic card going on a board by this point may be manufacturer specific.

Can someone check and tell me what the type of connection would work here? Is it LPC or SPI, and which card works with gigabyte? For some reason, the port is not listed in the manual, even though it is present on the board. A slightly older variant of the board has it in the manual (page 15):
https://download.gigabyte.com/FileList/Manual … 1800n-d2p_e.pdf

For reference, this is the model I have, and you can see it is pretty similar:
https://download.gigabyte.com/FileList/Manual … 1900n-d3v_e.pdf

I have the exact same board with the exact same issue. It died on New Year’s Eve 🙁

I have the original paper manual and it has the debug port listed on page 16. It is the same as documented on the GA-J1800N-D2P manual.

Did you get anywhere with fixing this? My 3.3 volt rail also shows a short to ground.

Steve

The debug port is some kind of LPC connector. I probed it and the bus is coming out of reset on power on, but nothing is being sent to it that I can see on the scope. There are post cards that could interface to it, but I am not sure I'd want to chance spending money for that right now. The only information on the post cards are in chinese and not online. Someone will have to tell me if they have one and can translate if it supports the debug port pinout on older gigabyte boards. But like I said, nothing comes up there so chance of reading a post code is low right now.

The short might not be the issue, it was just the first thing that seemed odd, but this is an advanced SoC design, so I don't know how "leaky" it might seem. However, after further investigating, this chip is part of a series of Intel Atom/Celeron/BayTrail chips that have bad I/O designs that cause them to short out over time with use (this is a known errata, I remember reading about it but my board seemed to be fine then). I am not sure which I/Os are affected, I guess, good luck asking for Intel for more information. This could be why our boards died. The information that came out first when it was discovered is that it affected the LPC clock specifically. But I don't know if the LPC clock is considered bad -- the waveform looks fine for 25mhz. The other side of this, is there doesn't seem to be any fetching of instructions from the BIOS to get far enough to utilize the super I/O or LPC. However, the BIOS serial line is active for about 1 second and then the board goes into reset. There could be some activity.

I am constructing a harness for the BIOS chip right now, so I should be able to read it later today. I might want to use a logic analyzer, but that will be for later, if the BIOS is reading fine.

But worst case, there are additional I/Os or clock pins on the SoC that deal with the peripherals (see board block diagram) that because of faulty I/O drivers the SoC can't handle and goes into a reboot loop. The best case it is just a corrupted BIOS.

Good (or should I say bad?) to know about the bad I/O design of these chips. I'd never come across this before. From the sound of it the CPU could well be dead which is a shame. It has been running 24/7 for seven years so I guess I can't be too mad. It is just annoying that I can't seem to find a replacement board at the moment that doesn't cost £££.

I was fortunate to borrow a Gigabyte GA-N3150N-D3V which uses the Braswell architecture. I checked 3.3V to ground on that board and no short.

Anyway from what you say I think this one is beyond my ability to fix. Unless it is a corrupt BIOS. I could swap the chip if that turns out to be the problem.

So I tried multiple ways to interface the BIOS chip to a tl866 progammer in circuit. Using a quality clip with real short leads works the best. Unfortunately I get random bits read incorrectly still. I am still back to the only theory that is most likely. The BIOS chip seems to be ok, and with alot more effort I think that will just be proven to be the case. The only thing left is the SoC, and considering we both had machines running 24/7 for 7 years die approximately the same time (I suppose our board does not depend on LPC like some of the other embedded devices do), we probably just hit the same problem, just on a different pin, which would mean this is simply unrecoverable.

It's quite a shame the manufacturers and Intel knew about the problem and kept information hidden, it's hard to know the extent this problem could be. Yes, cost for another similar board today is much higher, but I am now seeing a need so I might just have to pick up whatever seems to be the best deal and replace the entire board, but probably from a different vendor now.