VOGONS


First post, by gnif

User metadata
Rank Newbie
Rank
Newbie

Hi All,

Been reading this forum for a while now, an excellent source of information, and really have enjoyed my stay here.

Recently I built a Retro 486 Socket 3 SiS496 machine, however, I made a mistake and damaged it, blowing up the southbridge (dropped a screw into the open case while playing around with jumpers). This motherboard has now become a parts board unless I can obtain another 85C497 for a decent price (I am capable of swapping the chip).

Because I wanted to continue to play with this particular chipset I purchased another "tested" board from eBay with the same chipset, the KM-S4-1 and this is where this problem started. At this point I know that all my hardware is good and compatible, FP RAM, CPU, etc. However the BIOS won't proceed past the RAM test at power on, it gets to 640K on the memory test and hangs. If I enter the bios before the test starts I can see the BIOS is reporting that I have 2x the memory installed than I actually have. I have verified there are no shorts with the SIMM presence detect pins which as it turns out these boards do not even use as the SiS496 datasheet reports that memory is probed for size by writing and reading test patterns.

Further to this, if I swap out the BIOS from my damaged board (AMIBIOS) into this one (originally AWARD BIOS), I get resource conflict warnings, however, the system now detects the memory correctly and boots. This confirmes it's not a hardware fault, but likely a bad BIOS, however, I do not at present have a spare EEPROM to flash or the equipment to erase and program the UV erasable EPROMS these two boards came with.

I have ordered a programmer and eraser (I have wanted these for a long time now anyway), however as this community is full of very smart people I figured I would ask in case someone else has encountered this and knows of a possible solution.

Reply 1 of 2, by gnif

User metadata
Rank Newbie
Rank
Newbie

Ok, a follow-up, it seems the motherboard itself has a fault. I took one of my SIMMs and insulated the RAS & A10 pins to make it appear as a 4MB SIMM which now the system can post and boot from. Now it's booting I dumped the BIOS and compared it to one I found on the internet... they are identical.

Further tracing shows a potential fault with the SIS 85C496 itself, a faulty octal buffer was discovered that buffers the DRAM address lines, including A11. I replaced the chip and verified with a DSO that things were now working, however, this is the signaling I am seeing coming from the A11 line now.

DS1Z_QuickPrint1.png
Filename
DS1Z_QuickPrint1.png
File size
28.34 KiB
Views
294 views
File license
Public domain

This is after reset for the first ~100ms during which time the BIOS is probing the RAM for it's capacity. It's pretty ugly (I call it a squawtooth signal 😜) and leads me to believe the output in the 85C496 is damaged.

Reply 2 of 2, by gnif

User metadata
Rank Newbie
Rank
Newbie

SUCCESS!

I gave up on this board and used it as a parts board to repair the other one, swapping the northbridge over. Using the AWARD bios as I damaged the original AMIBIOS from this MB, I was able to run the system, not only does it work correctly, it detects ALL the ram properly! This confirms that the DRAM controller in the northbridge was faulty!