Hi guys,

I recently got this IBM 330 type 6571 machine. It was completely dead on arrival. No CPU (I put Am80486DX4-100). I changed the PSU to a regular AT, since the original had blown fuse.
On the attached below this post image, highlighted with Blue - power button conector, with Red - the undocumemted jumper. Here at Vogons I found another thread from where I found out that in order to use a regular AT PSU - I need to apply +5V and GRN to left and right pins of the blue connector on the attached above picture. Poking around I've discovered the following:

• Without that mode for regular AT PSU - mobo starts up when I close with a jumper the undocumented JP2. It's basically the same as applying +5V on the power switch connector (according to my multimeter measurments)

• Without that jumper - there's usually no PSGOOD signal on the AT connector, therefore mobo doesn't POST

What else have been done:

• I recaped the board completely just in case

• Most of the SMD chips were reflown

• All DIP chips legs including CPU socket and 1st memory slot - were reflown

• BIOS chip has been changed from original M28F101-150PI to the following: W29EE011-15, W29C011-15 (tried both) flashing it on the XGecu programmer. Tried few different bios binaries found on the internet (English and German), including the firmware dumped from the original chip

• Changed few coin batteries. Tried to provide 3-3.3v to the coin battery pins directly from the external laboratory power supply, with curren limit to 50mA

• Tried to change CPU multiplier and the bus frequency (cpu worked only on 75Mhz and 100Mhz, on other settings it didn't POST). In both cases in BIOS, CPU is detected as DX2 instead of DX4

• Tried to POST with lot's of different variants of memory configurations (1 slot / 2 slots / 4 slots populated with 8 / 16 Mb memory modules)

• Tried with 128Kb cache chips and a Tag chip, and without those - no matter what I do, motherboard doesn't detect cache

SO now about the issue. In all the above scenarios - the behaviour is the same: mobo POSTs, then starts to count the memory, then enters BIOS and error appears stating that CMOS battery is bad and asks if I want to autofix wrong settings.
If I confirm - I guess it loads default settings, If I decline - it does nothing allowing me to enter BIOS menu further (from where I can see the CPU speed and do whatever I want in BIOS).
Next when I exit BIOS without saving changes - the cycle repeats. However if I save changes - it reboots and I can boot from the boot devices (I'll say later about it)
But if I want to enter BIOS second time after changes were saved - mobo either:

• Reboots right away (sometimes)

• Hangs on the IBM logo after memory count (on cold start) or after message "F1 to enter setup, ESC to continue"

• Enters BIOS setup and freezes with the BIOS background and the text on top "IBM SurePath...blah blah"

• Enters BIOS setup and freezes with the BIOS background and the text on top "0000>" or "000000>"

• Enters BIOS setup and freezes with the BIOS message about the CMOS battery and corrupted menu"

Anyway on each next attempt to enter BIOS setup after initially saved settings mobo freezes untill I reset CMOS with a jumper - then everything starts form the beginning. After CMOS reset and accessing BIOS - it actually detects HDD (tried few and there were no issues with that).

Seems like saving BIOS settings (even if I don't load defaults, but just change date and save) - corrupts the CMOS memory which lives inside OPTI 82C602. To address the issue i did the following:

• Measured voltage from the coin battery on the 55 pin of OPTI 82C602 - it's present there (up to 3.3V, when I provided power from external PSU) and it doesn't depend if the system is powered on or off

• Changed all crystal oscilators and measured the correct sine wave on all of them - 16Mhz for devices, 34.38Mhz for CPU and 36Khz on the clock crystal

• Clock works and seconds are ticking correctly. However after clearing CMOS with a jumper, when I enter BIOS - date values are shown corrupted: i.e. it might appear as 34-51-2000 or in most cases 00-00-1900. When I move the cursor to the date it changes to 01-01-1900 or the max value which is applicable for day/month.

• Changed all crystal oscilators and measured the correct sine wave on all of them - 16Mhz for devices, 34.38Mhz for CPU and 36Khz on the clock crystal

• All input voltages on AT PSU are okay, changed PSU to the other one as well

• Tried to boot on internal video and using ISA Triden video card

• And of cause I looked through the traces a million times. The board was initially in it's original case, so it's was in quite good condition, no corrosion found at all. There was a coin battery, so no leaking acid

The other issue is that I can't boot the system at all. I tried like 10 different FDD 3.5" with different ribbon cables - result is the same: it tries to test FDD, but then fails, sometimes message appeared that FDD failed, but in most cases it continues to boot, starts to read from the floppy, but then fails to boot (tried number of different floppies with different boot version - DOS 6.22, boot floppy form Win95 and Win98). Only once I've goot Staring Windows 98 and then it froze. Even tried to install DOS on the HDD on other PC and then connect that HDD to this system - still doesn't boot (starts and then says that it's not system disk). Seems like floppy controller SMC FDC37C666GT doesn't work properly.

On the internet on github I found the same issue, and one guys said it's a known problem, but there was no solution.
I've been fighing with these issues during a week and there's nothing I can find!
If anyone faced this issue or have any kind of suggestions what else can be checked or done, I'd highly appreciate your help.