GIGABYTE GA-H110M-S2 - Stuck at Debug Code C2 - don't like Patriot Viper Steel RAM but works with G.SKILL

[DynaxSC]

Hi,

I discovered a strange behaviour of GIGABYTE GA-H110M-S2 board.

General test conditions:
- Tested with Core i5 6500 and Core I7 6700K - no difference
- Default BIOS settings, but tried also altering several memory setting - does not change anything

What is wrong:

With BIOS F1 and F2 the board normally POST's and boots to Windows 11 with 2 x Patriot Viper Steel PVS416G440C9K RAM (4400MHZ, Samsung B-Die chips)

When BIOS upgraded to anything above F2 (ie. F20-F27b) the board POST stucks after 3 Debug Codes on Code C2 (OEM BDS initialization codes) - the full codes sequence is:

1. 30 - System is waking up from the S3 sleep state
2. CC - OEM BDS initialization codes
3. C2 - OEM BDS initialization codes -> TL631 test board lits DRAM LED here

When I insert some other RAM sticks, eg. some different GSKILL or kingston the board POST's without any problem.

The board is repaired by me, 2 ESD protection devices were exchanged - after this I tested the whole board comprehensively, all ports, slot's, eg: USB, SATA, Audio, VGA port, etc., etc, and it works 100% OK, there is completely nothing wrong with the board, except it does not POST with 2 x a.m. Patriot's.
Even more strange is, when I populate two mixed RAM sticks, eg. one G.SKILL and one Patriot memory, the board POST's normally, order of the sticks does not matter.

The strange thing it POST with Patriot's with BIOS F1 and F2, so GIGABYTE must have spoiled somrthing in the BIOS from F20 version on.
I believe this is second H110M GIGABYTE board that has same issue, so in my view it must be some BIOS bug GIGABYTE has integrated into newer BIOS'es, or maybe this has been done deliberately by GIGABYTE. Entering lightly some conspiration theory area, maybe they have some conflict with PATRIOT (PDP Systems) and have banned Patriot memories from their boards 🤭 ?

But coming back to seriousness, anybody any clue what could be the reason ?
Anybody has an idea if there is any way to correct this latest but crappy GIGABYTE BIOS ?

[FriedFred]

Probably it has something to do with timing of signals. Have you tried to load the default BIOS settings and try again?

[DynaxSC]

Hi, as I wrote default settings do not work, also tried several changes in BIOS memory settings, but to no avail. It's also quite difficult to test, as the board does not go into BIOS on Patriot mems, so need to change mems to different make, alter BIOS settings, and then change mems again to Patriots, and finally test. So Gigabyte messed up something in the BIOS, as these Patriot mems are very good mems based on Samsung B-Die chips, and they work in all other motherboards makes I repaired, and these are several hundred by now.

Also my experience with Gigabyte mobo repair is the worst compared to ASUS, MSI or ASROCK. Especially the quality of the pcb material including the soldermask is very low. I do a lot of cpu socket replacements, and only on Gigabyte's the soldermask under the cpu socket breakes away very easily during heating of the the socket area.

Referring to GPU's almost the only pcb's demaged due to inter-layer shorts and consequtive electrical arc fires I've seen were Gigabyte GPU's, same applies to motherboards (second in this ranking in MSI, but with much less cases). So in my opinion Gigabyte is not doing a good job regarding to firmware, as well as hardware quality.

Further regarding the Gigabyte GPU's, majority of the boards I've seen with broken vcore vrm's, you can almost be sure, the GPU chip is also broken. To my view this means also bad design of the vrm's, as in case of high-side mosfet failure (short) the vrm controller does not short the low side mosfets immediatelly to protect the GPU chip from getting full 12V from ATX power socket, so OV protection does not work at all, or at least good enough to protect GPU chip from over voltage. Another thing where Gigabyte has done not the best job. Thats the reason, why I do not buy broken Gigabyte GPU's for repair at all.

Another nice bad design job is the Gigabyte GTX1070 Mini-ITX GPU. The 12V ATX power line has a few smd mlcc filter caps before the vcore vrm power input. They are really only very few, 3 pcs in parallel. I've repaired maybe 15 of these cards with these filter caps shorted. The repair was always replacing the shorted caps and adding another 3-4 caps soldered on top of the original caps. The reason for the shorting of these caps is just current overload due to high voltage ripple from the high vrm load. These tiny 3 caps are just not able to deliver the needed current/energy and fail due to current overload. This is an evident design flaw (or maybe a savings job of the finacial controller). The only advantage of this failure, it's easy to repair, provided the pcb is still intact (some shorted caps lead to pcb/tracks melting). So this card is the only Gigabyte GPU exception to me, I can decide to buy for repair.

[stj]

the bios reads the timing from an eeprom on the ram,
when you mix ram the board will run ALL the ram at the lowest settings it reads.
thats why mixing them works.
but obviously there is a bios bug too.