Fake cache on 486 motherboards

Navigation:

PcChips M912 with fake cache

PcChips M912 UMC UM8498F chipset

In spring 2019 I found on ebay a 486 motherboard with condition: NOS - new old stock to an acceptable price. I could not withstand... During the shipping time I already realized I have bought the most notorious 486 board: a PcChips M912 with fake cache (Write Back 61C328AH-15) which means that only empty plastic housing are placed on the board instead of real cache chips.

PcChips M912 with fake cache chips

I found some comments in the web - the chips are also not connected. My board has revision V1.7 and to all chips go real pcb tracks. It can be placed up to 1MB cache memory (8 pieces of 128kx8 SRAM e.g. ISSI IS61C1024, Winbond W241024AK or UMC UM611024AK). So, I de-soldered the fake cache, opened one part and found - only plastic...

Opened fake cache chip

Sockets with 32 pins and narrow pin distance (3/10" or 7.62mm) aren't available anymore today. I took two sockets with 16 pins each and placed it in series. Next time I will take the simpler version of sockets with double spring contact because put in a chip with 32 pins into a socket with small round holes needs really a lot of patience... The sockets with spring contacts offer the chip pins much more guidance.

Placed new sockets on the board

For first tests I bought ISSI IS61C1024AK-10N chips in China but I can't find a datasheet from ISSI with this speed rating... I assume these chips are relabeled fakes but up to now 19 of 20 pieces work. I tested the parts with my EEPROM-programmer TL866 II plus (only the lower 512kB are possible as part Winbond W24512) and the whole 486 system with the floppy version of memtest86+ Version 2.0 successfully. This version of memtest86+ runs well on my 486 machines, version 4 not.

PcChips M912 with real cache chips

AV7543 Revision 1.2 with (much better) fake cache: AA26257AK-15 (date code 9515)

This SiS 85C471 board from an unknown manufacturer (Made in Taiwan) has a very insidious fake cache: real IC housings with a plausible type name (sounds like Winbond) but the chip doesn't have a bonded die inside:

Fake cache AA26257AK-15

The chips were soldered without sockets and the cache configuration jumpers were replaced by fixed wire bridges.

I reworked this board: inserted sockets with double spring contact for the cache IC's and added real pin headers for the cache configuration:

Board after reworking

But I wouldn't buy this board again because:

The BIOS chip is blocking two ISA slots

There are four jumpers which are not explained on the board:

Jumper Function (reversed engineered)
J10 According to the SiS85C471 datasheet is pin 1 the output SMOUT1 (chip pin 58) and pin 3 the output SMOUT0 (chip pin 41). Pin 2 and 4 are ground - I would never set a jumper here...
J29 VESA local bus wait state: 1-2: 0WS, 2-3: 1WS
J30 VESA local bus speed: 1-2: <=33MHz, 2-3 >33MHz
J36 SIMM 30/72 selector: 1-2: 72pin bank 0 and 2 are active, 2-3: 30pin sockets are bank 0 and 2.

Last but not least: the turbo jumper must be set (or the turbo switch closed) to have full/turbo speed.


falk.richter*at*yandex.com
Last update: February 17, 2020