I’m sure this 430FX board supports FP-SIMMs just fine.

Maybe the SIMM slots have a bad contact due corrosion, or a trace between these slots and the chipset is damaged?

This behaviour can also be caused by a corrupted BIOS. Can you make a BIOS dump with your EEPROM programmer and post it here? I should be able to tell you if the BIOS is valid or not and which BIOS update file you need from TRW.

Cheers, Jan

I have this board (in fact it was my first PC motherboard after coming from an Amiga 1200) and I never heard of this board having an AMI BIOS. Soyo always seemed to use Award and the 5TC is compatible with EDO. I would suggest flashing one of the Award bioses and see how you get on.

Chkcpu wrote on 2024-05-30, 13:27:

I’m sure this 430FX board supports FP-SIMMs just fine. […]
Show full quote

I’m sure this 430FX board supports FP-SIMMs just fine.

Maybe the SIMM slots have a bad contact due corrosion, or a trace between these slots and the chipset is damaged?

This behaviour can also be caused by a corrupted BIOS. Can you make a BIOS dump with your EEPROM programmer and post it here? I should be able to tell you if the BIOS is valid or not and which BIOS update file you need from TRW.

Cheers, Jan

Thank you for your suggestions, Jan!

Actually some time ago I've posted the original BIOS dump from this board in this thread: Re: Socket 4, 5, 7 BIOS image collection and someone already uploaded it on TRW page: https://theretroweb.com/motherboards/s/soyo-sy-5tc#bios (REV UA0).

I don't like to wash with soap and water old motherboards, but I may try this time to "bath" this mobo, since the board came to me covered in some dirt and dust.

I don't know if the RTC ODIN package needs 3.3 V battery power to not mess the normal boot of this board, but maybe I have to try to boot the mobo after adding the missing battery holder. I just don't have spare battery holder right now. I already modified the ODIN "chip" by adding the missing external pin for the ext. battery and disconnecting the in-package depleted battery by drilling a hole on the right spot.

Or if I know which of the other BIOS dumps on TRC page may work for my version of the motherboard, I'll try different BIOS as first possible and easiest solution.

Brightraven wrote on 2024-05-30, 15:22:

I have this board (in fact it was my first PC motherboard after coming from an Amiga 1200) and I never heard of this board having an AMI BIOS. Soyo always seemed to use Award and the 5TC is compatible with EDO. I would suggest flashing one of the Award bioses and see how you get on.

Thanks for the info, Brightraven! I messed the BIOS vendor with another s.7 board that I'm testing right now. It is Award BIOS. I'll correct my first post.

Unfortunately I don't have any 72-pin EDO-RAM sticks to test if the problem is caused by the FP-RAM modules I'm trying to use.

analog_programmer wrote on 2024-05-30, 15:38:

Actually some time ago I've posted the original BIOS dump from this board in this thread: Re: Socket 4, 5, 7 BIOS image collect […]
Show full quote

Actually some time ago I've posted the original BIOS dump from this board in this thread: Re: Socket 4, 5, 7 BIOS image collection and someone already uploaded it on TRW page: https://theretroweb.com/motherboards/s/soyo-sy-5tc#bios (REV UA0).

I don't like to wash with soap and water old motherboards, but I may try this time to "bath" this mobo, since the board came to me covered in some dirt and dust.

I don't know if the RTC ODIN package needs 3.3 V battery power to not mess the normal boot of this board, but maybe I have to try to boot the mobo after adding the missing battery holder. I just don't have spare battery holder right now. I already modified the ODIN "chip" by adding the missing external pin for the ext. battery and disconnecting the in-package depleted battery by drilling a hole on the right spot.

Or if I know which of the other BIOS dumps on TRC page may work for my version of the motherboard, I'll try different BIOS as first possible and easiest solution.

Your Rev UA0 BIOS checks out fine, so no corruption there.

I see on TRW that there are 2 BIOS flavors for this SY-5TC board, one with a 2A59CS2LC BIOS-ID and the other has 2A59CS2PC. Your UA0 BIOS has the later and looking at the picture of your board I see an UMC UM8669F super I/O chip. The ‘U’ in your BIOS version probably means it is for the board with UMC I/O, so the UA0 BIOS is correct for your 5TC.

The TCS BIOSes with the 2A59CS2LC ID are for boards with SMC I/O and are not suited for your board.
Therefore the latest 2A59CS2PC BIOS is the Ver. 2.3 and that should be the correct BIOS update for your board.

Because the BIOS hangs very early during POST step C6, it is still executing code in the bootblock which has only very basic video support. Did you try an ISA VGA card, to see if any error messages appear?

Seeing the condition of the board, my bet is on dirty and corroded SIMM slots. So I’m in favor of a good wash. 😉

Jan

Chkcpu wrote on 2024-05-30, 17:58:

Your Rev UA0 BIOS checks out fine, so no corruption there. […]
Show full quote

Your Rev UA0 BIOS checks out fine, so no corruption there.

I see on TRW that there are 2 BIOS flavors for this SY-5TC board, one with a 2A59CS2LC BIOS-ID and the other has 2A59CS2PC. Your UA0 BIOS has the later and looking at the picture of your board I see an UMC UM8669F super I/O chip. The ‘U’ in your BIOS version probably means it is for the board with UMC I/O, so the UA0 BIOS is correct for your 5TC.

The TCS BIOSes with the 2A59CS2LC ID are for boards with SMC I/O and are not suited for your board.
Therefore the latest 2A59CS2PC BIOS is the Ver. 2.3 and that should be the correct BIOS update for your board.

Because the BIOS hangs very early during POST step C6, it is still executing code in the bootblock which has only very basic video support. Did you try an ISA VGA card, to see if any error messages appear?

Seeing the condition of the board, my bet is on dirty and corroded SIMM slots. So I’m in favor of a good wash. 😉

Jan

Thank you very much for this detailed info! You've just confirmed my suggestion for the newer suitable BIOS dumps.

I've tested the board with both ISA and PCI perfectly working VGA cards and there is not a single blink on the display attached. The board just boots, the POST test card shows hang at "C6" code and "pc-speaker" starts cycle of one continuous long "beeeeeeeeeeeeeeeeep". Also I've checked all the voltages for RAM slots, CPU socket, PCI and ISA slots and they're all fine. Post test card also indicates good +5, +12 and -12 V voltages in both ISA and PCI slots. Unfortunately I do not have any oscilloscope for measurements above 200 kHz, so I think I can't test any RAM, ISA or PCI bus signals with my chineese cheapo signal tester.

I think, I'll give this board a good bath tomorrow, a day to dry out and then if it still "cries out" for RAM problem I may reflash it with the newest U-version BIOS available at TRW site. And if it still complains about the FP-RAM installed... I don't know what else I can do as next easy test step, except to search for some cheap EDO-RAM modules at local online "flea-markets".

Chkcpu wrote on 2024-05-30, 17:58:

Because the BIOS hangs very early during POST step C6, it is still executing code in the bootblock which has only very basic video support. Did you try an ISA VGA card, to see if any error messages appear?

I found this TRW blog page with some detailed information about Award BIOS versions POST error codes and POST routines: https://blog.theretroweb.com/2024/01/20/award … ost-codes-list/

For the continuous long "beep": "Repeating (endless loop)" - "Memory error" - "Check for improperly seated or missing memory." https://blog.theretroweb.com/2024/01/20/award … ist/#Beep_codes

For the "C1" POST code in v.4.51PG BIOS: "Memory presence test; OEM specific, test the size of on-board memory" https://blog.theretroweb.com/2024/01/20/award … /#Version_451PG

For the "C6" POST code in v.4.51PG BIOS: "Cache presence test; External cache-size detection test" https://blog.theretroweb.com/2024/01/20/award … /#Version_451PG

I don't know if the actual boot error is at step "C1" or "C6", but "C6" is the last POST code shown when the POST process stops.

As for the VGA tests, here I see that they're executed later than the memory and cache tests: https://blog.theretroweb.com/2024/01/20/award … versions_%3E_42

Now I'm wondering if the "C6, C1" ("C6" is the last one) codes on the POST test card indicate a problem with the RAM or a problem with the cache SRAM.

This should be easy to test: Remove all cache chips, the TAG chip and any inserted COAST module. The system would start without error - IF cache was the problem.

majestyk wrote on 2024-05-31, 06:34:

This should be easy to test: Remove all cache chips, the TAG chip and any inserted COAST module. The system would start without error - IF cache was the problem.

Yeah, I also thought about this kind of test, but I have only one 256 kB COASt module and I don't remember with what model s.7 mobo it was used almost three decades ago. As far as I know, these COASt modules are not standardized in their pinout. If I intall the "unknown" COASt module and it is not compatible with this particular motherboard, I think I may fry the board or at least the COASt module.

I may take some pictures of the COASt module I have, if this will help for someone to resolve its compatibility.

I will note again, that with or without SRAM cache and TAG chips installed the board's boot behavior doesn't change.

Hopefully the wrong coast stick wasn’t installed causing board damage. But usually the wrong stick will short stuff out.

Are your sticks 8 chip? I have ran into some boards caring about the configuration of the chips.

For example, I have a board which supports 32mb

But it can’t be single sided 16s or a single double sided 32 despite the board’s bus being 32 bits.

It must be two double sided 16 mb sicks.

Maybe you have a similar issue?

Try smaller ram with more chips.

Sphere478 wrote on 2024-05-31, 15:55:

Hopefully the wrong coast stick wasn’t installed causing board damage. But usually the wrong stick will short stuff out.

The board came to me equipped with UMC SRAM chips for 256 kB cache and TAG. I think this is factory configuration.

What if the JP1 for 3.3 V and JP29 for mixed voltage 3.3 and 5 V are reversed (JP2 is open, JP29 is shorted)?

Sphere478 wrote on 2024-05-31, 15:55:

Are your sticks 8 chip? I have ran into some boards caring about the configuration of the chips. […]
Show full quote

Are your sticks 8 chip? I have ran into some boards caring about the configuration of the chips.

For example, I have a board which supports 32mb

But it can’t be single sided 16s or a single double sided 32 despite the board’s bus being 32 bits.

It must be two double sided 16 mb sicks.

Maybe you have a similar issue?

Try smaller ram with more chips.

This is a good point. Yes, my only two 72-pin FP-RAM sticks are with 8 chips, single sided, 16 MB each. But they do work on late 486 boards and some slightly newer socket 7 boards. I'll try to find some EDO-RAM sticks - 8 MB single sided or 16 MB double sided.

I also have this mainboard and just dug it out for a test.

I can confirm it "eats everything" - FPM and EDO, all different sizes, one sided and double sided. Even my obscurest RAM sticks are being detected correctly.
Cache chips or / and COAST module are not needed for starting it, it just reports "cache memory -> none" in this case.

Thank you very much for this info, majestyk!
Do you remember what version of Award BIOS is on your board?
I see that your mobo has no cache installed, but JP1 (the manual says that JP1 is identical to JP2) and JP29 are reversed according to my board's settings. At first look my version of SY-5TC also lacks this factory battery holder (I'll add one soon) and the VRM near the cache sockets, otherwise it is pretty much identical to yours.

So, my next step will be giving a good "bath" to the board and finishing the ext. battery connection to the modified RTC module. And maybe I'll flash the newest version of the other two available at TRW site.

P.S. This is the cache configuration and JP2, JP29 settings on my board (don't mind the dust, the board is clean now):

Have you thoroughly inspected the board for any damage? It could be a cut trace or a shorted pin on the Intel 430FX chipset (much like the 3DFX Voodoo chips which always seem to get squished together).

Brightraven wrote on 2024-05-31, 17:58:

Have you thoroughly inspected the board for any damage? It could be a cut trace or a shorted pin on the Intel 430FX chipset (much like the 3DFX Voodoo chips which always seem to get squished together).

Yes, I've inspected it with magnifying glass after the first cleaning and I didn't notice any physically damaged elements or traces. Now the board is washed with soap and water. I'll give the board one day to dry out and may inspect it once more after it's completely clean and dry.

The manual shows it supports single sided simms with 64MB, but double sided only up to 32MB? Did the 430fx support modules bigger than 32MB.

JP 2 and JP29 are set correctly. Your board has no voltage regulator, so the chache chips run at 5V.
What about jumpers 4, 20, 16 and 17??

majestyk wrote on 2024-05-31, 21:41:

JP 2 and JP29 are set correctly. Your board has no voltage regulator, so the chache chips run at 5V.
What about jumpers 4, 20, 16 and 17??

JP4: 1-2
JP20: 2-3
JP16: 1-2
JP17: 1-2
JP24: open (default)
JP2: open
JP29: short

I think these are correct settings for 256 kB cache in write back mode.

Can you give me the markings/values for the missing VRM (U42) and the two missing SMD resistors near U42 (R106 and R107)?

Riikcakirds wrote on 2024-05-31, 21:18:

The manual shows it supports single sided simms with 64MB, but double sided only up to 32MB? Did the 430fx support modules bigger than 32MB.

Typically, 4MB, 16MB and 64MB SIMMs are single-sided; 8MB, 32MB and 128MB SIMMs are double-sided. If the manual was printed at a time at which 16MBit chips were the biggest available memory chips, a 64MB module had to have 32 (or 36) of them. Each of these chips drives one bit out of the 32/36 required bits. These chips were usually distributed over both sides, with 16/18 chips per side on a single-rank module. Each chip loads the address and data lines, so a 32-chip SIMM is quite heavy on the board. Now imagine how bulky and electrically demanding a 128MB module with 64/72 of those chips would have been.

WIth the advent of 64MBit chips, you could have them as 16M x 4 chips to build 128MB SIMMs using 8/9 of them per side to get to 16/18 chips in total. This is when usable 128MB SIMMs started to appear on the market, but likely after the manual has been printed.

As the 430FX chipset can only cache 64MB in its L2 cache, installing more than 64MB in that board is not that attractive anyway, though.