That sounds weird to me. Do you have any pictures?
Is it possible that Vcc is connected directly to a +5V plane on the through hole?

I would first remove the bridge (since it´s nonsense), insert a known good cache chip and solder a bridge on the backside of the motherboard from some 5V power supply rail to pin 28.

Without the TAG RAM working correctly the whole cache won´t be detected and addressed by the BIOS. And probably POST will hang at this point.

Maybe there´s a design flaw in this mainboard revision or the trace got interrupted due to age or overload.

And I agree 100% that pictures would be essential here.

Photos attached (apparently can't post in-line without https...) .

The contact in the 28pin DIL socket is missing.
Can you measure +5V at pin 28 on the backside? If you can, replace the socket.

majestyk wrote on 2021-04-06, 19:11:

The contact in the 28pin DIL socket is missing.
Can you measure +5V at pin 28 on the backside? If you can, replace the socket.

3.48V on the pad at the back

The pad looks to have never had a pin installed - just the small blob of solder.

I have jumpered the TAG chip's pin over to 5V and it stalls out at "starting ms-dos". Same with both the included 256KB cache set, and my known good 128KB set from another board. Jumpers set to match.

Could it be that a filter in beteween has gone bad and is going toward the resistance, high ohm, instead of 0 ohm?
This could explain the 5 to 3.48V

You´re booting DOS from SSD?
Just to be on the safe side I would give booting from a real Floppy a try.

Any BIOS settings for cache mode, speed etc.?

majestyk wrote on 2021-04-06, 20:12:

You´re booting DOS from SSD?
Just to be on the safe side I would give booting from a real Floppy a try.

Any BIOS settings for cache mode, speed etc.?

IDE drive removed, floppy starts to load, but system stops doing anything; floppy continues to spin without any access. Locked up - cannot CTRL+ALT+DRL. Disable L2 cache, boots to prompt without a problem.

All I have in the BIOS is write thru/back and tag bit settings. I've tried changing every setting in the BIOS already, and changing the cache/RAM settings to the slowest values available.

Again, this is not acceptable as you need solid 5V source that is truly 5V somewhere nearby on the *motherboard's*, solder that wire's end to this then run the wire jumper and solder this to the 15ns 32K x 8 bit pin VCC 28 this means solder this to this tag cache IC's leg of the pin 28. And another issue and this is a requirement, you need also to attach a small capacitor to this pin 28 to ground nearby.

This means using multi-meter to locate ground and 5V on the motherboard's.

Cheers,

Can you post a picture of the entire board ? Also: what CPU are you using ?
Something is very odd when they jump a wire on the TAG chip + missing the full pin 28 in the socket.
What happens in the BIOS when you try to change: L2 cache tag bits: 8 bits ?

Horun wrote on 2021-04-06, 23:28:

Can you post a picture of the entire board ? Also: what CPU are you using ?
Something is very odd when they jump a wire on the TAG chip + missing the full pin 28 in the socket.
What happens in the BIOS when you try to change: L2 cache tag bits: 8 bits ?

Attached - Jamicon KM-S4-1, BIOS 4.2

No change in behavior whether I have the cache tag bits set to 7 or 8. Same issue with a DX and DX2 CPUs, rejumpered.

I would think the open capacitor spot just to the side of the CPU socket would also be 5V - it's the same Voltage as the VCC pin, just under 3.5V. Not sure if a coincidence with the 3.45V CPU.

Interestingly, I put a 486DX in there and switched the CPU Voltage to 5V. The capacitor pads next to the CPU, and the TAG pin28 pad on the motherboard are now 5V.

Now make sense. May need to have a 3.3V and 5V tolerant SRAM chips.

Cheers,

pentiumspeed wrote on 2021-04-07, 00:08:

Now make sense. May need to have a 3.3V and 5V tolerant SRAM chips.

Cheers,

You have that right - these all appear to be 4.5-5.5VCC range.

Still stuck "starting ms-dos" with cap and jumper wire. Same with the um61256fk-15 and 256KB group installed.

Thanks for the picture ! I have nothing that comes to mind that would help other than a bad board revision, it is a good looking board !

I imagine that you need to make the cache timings slowest possible, since you have very slow cache chips for a (fast) 486 board. It might help to lower bus to 25MHz for testing things too.

It´s running with just one RAM stick in slot 3 now, I don´t know if tis SIS chipset supports that. Did you try with a pair of identical sticks in slots 1 +2?

Th cache chip you´r using is 32Kx8. According to the manual this should be 8Kx8 for 128K cache, 16Kx8 for 256K of cache and 32Kx8 for 512K of cache.
Some mainboards / chipsets don´t tolerate wrong sizes of TAG-RAM.

majestyk wrote on 2021-04-07, 05:16:

It´s running with just one RAM stick in slot 3 now, I don´t know if tis SIS chipset supports that. Did you try with a pair of identical sticks in slots 1 +2?

Th cache chip you´r using is 32Kx8. According to the manual this should be 8Kx8 for 128K cache, 16Kx8 for 256K of cache and 32Kx8 for 512K of cache.
Some mainboards / chipsets don´t tolerate wrong sizes of TAG-RAM.

Yes - you will see photos of simms in slots 1-2, and others in 3. The board doesn't seem to care about that, so I've gone back and forth today, to keep ruling that out. Has been easier to work in the tight spot with my hands by removing the simms (that TAG jumper wire + cache jumpers near the simms are a bit tight, even with tweezers).

I saw the manual mentioning TAG SRAM sizes/ratios. I'm trying to find a combo here that matches the ratio they show. I don't think I have a set with close enough timings, and with the inability to configure them any slower any timings what the BIOS photos already show, not sure it's possible without finding another source.

This evening I've been messing with this other pictured combo I received today - 486DX with 256KB L2, 8MB RAM, trident isa, soundblaster pro 2, IO card, desktop case and linear psu. Worked great out of case, and time to remove and neutralize that battery. Will test the old psu a bit before I potentially put it on a system.