Reply 20 of 176, by kool kitty89
The Quadtel labeled chip should just be a (semi) generic keyboard controller, mostly interchangeable between boards (sometimes NMOS vs CMOS and speed related issues depend and/or A20 handler implementations) and shouldn't be storing any BIOS parameters.
The realtime clock chip usually contains a small amount of static RAM for CMOS/BIOS settings, and the chipset itself might have some scratchpad memory for the same purpose, but there's odd cases where data remains even with no battery and all capacitors discharged, and I'm not sure how that works.
I have a PCChips M205 (not PCChips branded, but same board) that does that same thing, and I haven't tried swapping the BIOS to compare settings, but it retains disk parameters and EMS memory selection without power somehow (date/time as well as shadow RAM configuration get wiped, though).
The capacitors on that board also keep CMOS RAM powered for a surprisingly long time, which made the situation more confusing.
It could either be a small amount of EEPROM somewhere in the chipset or genuine NVRAM of some sort, maybe FERAM. (it had a short fad in the early 90s and ended up getting used by Sega briefly as the 8kB save RAM inside Sonic 3 carts, but I'm not sure if that could be the case: Sega's example also has a fairly high failure rate with unknown expiration date)
Anyway, I'd think a generic IDE/floppy or multi I/O ISA card would work fine for connecting IDE drives (and Compact Flash) to that board. It's worked fine for my 286 boards, though if you don't have old/small enough drives (or CF cards with known Cylinder/Head/Sector configurations) that would be less useful than XTIDE with large disk support.
IDE drives old enough to feature capacity limit jumpers should also work for any boards that don't have a BIOS with the 528 MB capacity limit (1024-16-63).
My 1990 vintage M205 worked fine with an old 2.5 GB drive I don't have anymore (it was Apolloboy's 286 build at the time and he kept the drive and since sold/traded it) and I swear it wasn't even in cap-limit mode and worked fine, and the settings were even still in the BIOS when I checked last year, but I didn't write them down or take a picture, so I'm not sure of the C/H/S count.
I think it may have been a Western Digital Caviar 32500 (those look like what I remember) and those are 4960/16/63. (which implies that BIOS might work all the way up to the 8GB limit ... though it lets you input values beyond that with a 255 Head limit and 65536kB total address/capacity limit, which seems common to a number of early 90s AMI BIOSs, 16-bit integer limit to the capacity description)
That's with a 1988 dated AMI BIOS (1990 date in the POST and set-up screens, but 1988 copyright on the sticker), and I have a Quadtel BIOS 286 I could try for comparison.
In any case, is you have any old IDE drives kicking around, it'd be worth a shot.
Did that machine come with an IDE drive and interface card or is that SCSI? It looks like a 40-wire IDE cable to me, but it's hard to tell from the photos.
Smaller CF cards (up to 1 GB) should work OK too with the correct C/H/S input, if you already have a CF to 40 pin IDE adapter. (I'd assume you would if you have the 40-pin IDE XTIDE card, but might not if it's the native CF adapter)
The color of that PCB really reminds me of some PCChips or Hedaka (Llion, etc) boards from that period, but did they build any with CHIPS and Technology chipsets? (I've seen Citygate based M209 boards by them with C&T DMA/clock ASICs, but not C&S chipsets)
It could just be coincidence, though, just superficial resemblance. (oddly enough, the Motorola based IDE/floppy interface card Apolloboy and I originally used with the M205 matched that PCB style, too)
Then you've got to get some chipset drivers working with this thing, especially assuming it has integrated EMS support. It looks like a fairly late, highly integrated chipset (1990-ish, that '9042' could be a 1990 date stamp too) and probably would run a good deal faster than that 12 MHz. You need a socket for that osccilator and a 286-16 or 20 to try with 32-50 MHz oscillators. If wait states are like my M205, that 70 ns RAM would work fine at 20 MHz anyway, maybe 22 or 25. (if it's zero wait state, maybe 16-20 MHz depending on the chipset's speed/tolerances)
Also neat to see an old XT clone box with an upgrade board inside like that. I swear it was one of those with a K6 badge stuck on it that I almost got as a freeby about 10 years ago, no idea what was inside it. (a guy was dropping it off at WeirdStuff Warehouse and offered it when I passed him in the parking lot)