What is it supposed to generate, 5V from 12V? I would definitely jumper over.

It generates 5.27V from the PSU 12V for early P66 CPUs, later revisions of the board did not include this regulator circuit as Intel got newer P5 steppings to work fine at 5.0V and 66MHz. Socket 5 Plato boards use a similar circuit in order to generate 3.3V for the P54C processors.

So you should be able to bypass the regulator and feed 5V to the CPU if you're using a P60 or a newer P66.

TheMobRules wrote on 2024-10-24, 03:27:

It generates 5.27V from the PSU 12V for early P66 CPUs, later revisions of the board did not include this regulator circuit as Intel got newer P5 steppings to work fine at 5.0V and 66MHz.

that's not at all what i'm seeing in the march '95 intel specification update. according to page 4 of that, on the contrary, the final D1 stepping (with the FDIV bug fix) was the first to introduce some s-specs that didn't include 5.0V in their voltage range, for both 60 and 66 mhz variants.

also, the original batman board didn't even include the spot for the regulator to begin with. and according to page 6 of the batman's revenge manual, only boards with the regulator can reliably use 66 mhz processors, with no mention of steppings.

Thanks for your answers !
I've removed damaged parts and resoldered MANY loose legs.

Fortunately none of the crocked legs broke.

Two problems ahead: power up CPU and settle it in the socket...

auron wrote on 2024-10-24, 07:33:

that's not at all what i'm seeing in the march '95 intel specification update. according to page 4 of that, on the contrary, the final D1 stepping (with the FDIV bug fix) was the first to introduce some s-specs that didn't include 5.0V in their voltage range, for both 60 and 66 mhz variants.

also, the original batman board didn't even include the spot for the regulator to begin with. and according to page 6 of the batman's revenge manual, only boards with the regulator can reliably use 66 mhz processors, with no mention of steppings.

Boards without the regulator work just fine with 66 MHz CPUs, I asked that same question on this forum a while back. I think it’s just a case of Intel being unclear on purpose to avoid details about manufacturing defects/limitations.

Locutus wrote on 2024-10-24, 07:57:

Thanks for your answers !
I've removed damaged parts and resoldered MANY loose legs.

The fix on that chip looks great!

Not sure what to do with the socket, I don’t think replacements can be easily found. Maybe you can replace it with a regular PGA LIF socket with the proper number of pins?

auron wrote on 2024-10-24, 07:33:

that's not at all what i'm seeing in the march '95 intel specification update. according to page 4 of that, on the contrary, the final D1 stepping (with the FDIV bug fix) was the first to introduce some s-specs that didn't include 5.0V in their voltage range, for both 60 and 66 mhz variants.

also, the original batman board didn't even include the spot for the regulator to begin with. and according to page 6 of the batman's revenge manual, only boards with the regulator can reliably use 66 mhz processors, with no mention of steppings.

Looks like the case of soft recall. "Oh btw our 66MHz CPUs arent 5V after all, they are 5.25V. Yes your system was crashing because you were holding it wrong".

TheMobRules wrote on 2024-10-24, 11:08:

The fix on that chip looks great!
Not sure what to do with the socket, I don’t think replacements can be easily found. Maybe you can replace it with a regular PGA LIF socket with the proper number of pins?

Thanks, me and my "chinascope" did our best 😉

Indeed sockets 4 are available but the only sources I found are from outside the EU.
I won't buy it anyway until it becomes clear if I can fix this motherboard.

Here is my improvised solution with the broken socket... does the job well.

I've also managed to 'bypass' the 12 to 5.2V regulator (shorts instead of CR4A1 and CR4A2 diodes).
The motherboard gives signs of live (it gets into sort of infinite loop).

https://www.youtube.com/watch?v=Lyucn4K5_-g

A couple of observations:

By comparing with my own board, it seems you're missing SMD electrolytic capacitors at positions C7E4 (near the ISA slot you removed) and C1J3 (top right, near the SIMM slot) and an SMD tantalum at C6F2 (below the crystal). Probably none of those is critical, but you never know.

More importantly, what happens if you try to boot with the RCVR jumper set to the other position? If you do that, the board will run in recovery mode (no video) and will try to reflash the BIOS from the floppy drive. I am suggesting this because the BIOS on these Intel boards are known to become corrupt after some time and a reflash becomes necessary. The infinite loop you experience may be a symptom of BIOS corruption.

TheMobRules wrote on 2024-10-25, 02:19:

A couple of observations:
By comparing with my own board, it seems you're missing SMD electrolytic capacitors at positions C7E4 (near the ISA slot you removed) and C1J3 (top right, near the SIMM slot) and an SMD tantalum at C6F2 (below the crystal). Probably none of those is critical, but you never know.
More importantly, what happens if you try to boot with the RCVR jumper set to the other position? If you do that, the board will run in recovery mode (no video) and will try to reflash the BIOS from the floppy drive. I am suggesting this because the BIOS on these Intel boards are known to become corrupt after some time and a reflash becomes necessary. The infinite loop you experience may be a symptom of BIOS corruption.

Thanks for you answers!
Yes, I'm aware of those missing capacitors, during inspection I've seen remains of solder.
All are bypass caps, not critical.

I was suspecting BIOS corruption but was not aware of the existence of 'recovery' one.
You are right !
The motherboard successfully enters recovery mode (no VGA signal, but I'm sure it loads OS).

https://youtu.be/X6XFTnuE3Lo

The story continues....

I've downloaded latest BIOS (1.00.13.AF2 ) from retroweb and put it into a bootable floppy with DOS 6.22.

I set RCVR jumper to 'ON', power on the MB, and the process of programming flash seems to pass successfully.
System boots DOS, beeps once, floppy continues to read data and after some time beeps twice and the floppy stops reading.
Contents of the autoexec.bat seem to confirm that everything goes as planned:
@echo off
beep.com <- first beep
fmup.exe < - reading data from floppy for programming flash and two beeps at the end

I power MB down, set RCVR jumper to 'normal' and power it up.
Unfortunately the POST stops with 'EE' code...

Maybe flash chip itself is bad ?
Any suggestions ?

The flashing process normally includes a verification that the BIOS was written successfully, so given that the recovery process appears to finish as expected the flash chip wouldn't be my first suspect.

Now, since the recovery works we can assume that the critical parts of the boot process are working, that is basic chipset initialization, CMOS/RTC (in the Dallas chip), floppy (handled by the SMC I/O chip) and some other stuff. Let's think about what is not needed on the recovery process: since no video can be displayed in the recovery process due to limited ROM space, and no peripherals other than the floppy drive are used, maybe initializing the PCI to ISA bridge is not necessary during recovery.

Now, my knowledge about this is very limited, but it could point to the 82378 bridge chip where you had to repair the legs. Maybe there's still something wrong with that one? Short, unsoldered legs, or something like that. Given the condition it was in when you got the board that chip must have been hit pretty bad.

Just my speculation, of course it would be much more useful to know what the 'EE' POST code means...

try isa VGA?

TheMobRules wrote on 2024-10-26, 01:03:

Just my speculation, of course it would be much more useful to know what the 'EE' POST code means...

https://web.archive.org/web/20230925005155/ht … des/amibios.htm nothing
https://github.com/cirosantilli/ralf-brown-in … nter61d/PORTS.A "E1h-EFh setup utility pages 1-15"

TheMobRules wrote on 2024-10-26, 01:03:

The flashing process normally includes a verification that the BIOS was written successfully, so given that the recovery process appears to finish as expected the flash chip wouldn't be my first suspect.

Trying to rule out 'bad flash chip' without desoldering it, I've made a little experiment:

Ran recovery mode, powered MB down, disabled 'Flash update' via jumper and powered it up again...
Since recovery jumper was still 'on', recovery process begins and seems to run normally.
At the end, instead of two beeps I got series of beeps (probably sign of verification failure).

My doubts about the chip remain sustained... the flash should contain exactly the same code.

Locutus wrote on 2024-10-26, 23:10:

At the end, instead of two beeps I got series of beeps (probably sign of verification failure).

or "same version detected, aborting"

I have a small... or maybe a great success.

I've tested some available BIOS versions.
After performing the recovery with version 1.0.0.4, the board started in 'normal' mode!
However POST ended with error 63h but I could enter the setup in which I disabled the cache and... voila - MB POSTs normally !

The victory is closer and closer.

Happy ending !
The problem with the cache memory was related to cold solder joints, which were not visible from the outside.
After removing the cache chips , they were obvious.
BIOS upgraded to the latest version, cache enabled - the board works like a charm.

https://youtube.com/shorts/R4gKWVgz11g?feature=share

DC-DC is there to boost 5V to a slightly higher 5V for 66 MHz CPUs.

Cheers,

Good work Locutus ! Glad you found the problem and fixed it.

Locutus wrote on 2024-10-25, 01:18:

I've also managed to 'bypass' the 12 to 5.2V regulator (shorts instead of CR4A1 and CR4A2 diodes).
The motherboard gives signs of live (it gets into sort of infinite loop).

Interesting, I thought that the 12V to 5.2V regulator would be bypassed when the voltage jumper is set to 60MHz (5.0V). I also have this board with the assembled regulator, and when I turned it on, the CR4A2 diode burned out. Are you sure that I can simply bypass it? Why are these present?