See attached photo of 486 board which was sold to me as 'working'.
When powering on, it fails to post. Results are Beep Beep... beep (x8), tick tick tick tick, Beep Beep.
Tried the following:
Multiple different types of 30 pin and 72 pin RAM.
Three video cards
Reset the CMOS
I note the cache is soldered and the jumpers are fixed to 256k. I don't think it is a fake cache board despite the cache being soldered. The build quality looks too good.
Oh, it definitely has fake cache. The "WRITE BACK" label on the chips is a dead giveaway of that, in addition to the chips being soldered to the board.
It seems to be the Ver. 7 of the Biostar MB-1433/50UIV, but I don't recall Biostar boards having fake cache. Maybe it's a clone, can you show the back of the board? There should be a Biostar logo somewhere if it's original.
Regarding your problem: can you try powering on with only a single FPM 72-pin SIMM in the "BANK 0/2" slot? What video cards are you trying? ISA or VLB?
Well this is odd..... I tried it with one stick of 72pin and it posted. Put the 2nd stick back in and it registered 8MB as expected.
It uses a fancy BIOS that looks a little like windows 3.1
Shame if it is a fake cache board. POST screen reads GREEN-486VL with WRITE-BACK SRAM.
Will plug in an IO card, boot to dos and run sysutil or something.
Absolutely no biostar logo or any other meaningful info that I can see anywhere. If it is a clone, it is a good one. Fake cache boards I have seen in the past tend to use very flimsy PCB and other crappy components....
Got it to POST. Switched to yet another video card. Simple old 512k trident. Previously been using a 1MB trident, a couple of cirrus cards and a VLB cirrus
Got it to boot after swapping VLB IO card for simple ISA one.
No L2 cache reported in HWINFO despite it being enabled in the bios.
CPU speed is reported as 100mhz but sysinfo shows it as slower than a DX2/66
HWINFO does report is as a biostar board, so guess it is a knock off with fake cache? Which again is ODD given the apparent quality of the board.... Though no way a DX4/100 should report as slower than a DX2 /66.
I'm curious how 9 chips can amount to 256k of cache, since 256 doesn't divide to 9.
8 + tag ?
Likely another way to tell its fake . .the tag chip sis identical tot eh cache chips 😒
8 + tag is correct. Tag is often identical on those boards. I'm still pretty sure this is fake cache.
The question is wether it can be removed and replaced with real cache chips.
I'm curious how 9 chips can amount to 256k of cache, since 256 doesn't divide to 9.
8 + tag ?
Likely another way to tell its fake . .the tag chip sis identical tot eh cache chips 😒
8 + tag is correct. Tag is often identical on those boards. I'm still pretty sure this is fake cache.
The question is wether it can be removed and replaced with real cache chips.
I keep reading your name as Door Kaat
Meow !
Still better than Basement Cat.
all the traces under them look to be heading back to the CPU area so its likely wired up but they cheaped out with the cache.
Likely another way to tell its fake . .the tag chip sis identical tot eh cache chips 😒
8 + tag is correct. Tag is often identical on those boards. I'm still pretty sure this is fake cache.
The question is wether it can be removed and replaced with real cache chips.
I keep reading your name as Door Kaat
Meow !
Still better than Basement Cat.
all the traces under them look to be heading back to the CPU area so its likely wired up but they cheaped out with the cache.
The name is that of a liquor brand. Not to be confused with user dormcat
I'm not that creative when picking out online names - I saw an advert while registering and used the name. 😅
The missing 74f245 transcievers are probably also required for cache to work?
Maybe JP26 has something to do with memory configurations?
Doornkaatwrote on 2022-01-28, 10:10:The name is that of a liquor brand. Not to be confused with user dormcat
I'm not that creative when picking out online names - I […] Show full quote
8 + tag is correct. Tag is often identical on those boards. I'm still pretty sure this is fake cache.
The question is wether it can be removed and replaced with real cache chips.
I keep reading your name as Door Kaat
Meow !
Still better than Basement Cat.
all the traces under them look to be heading back to the CPU area so its likely wired up but they cheaped out with the cache.
The name is that of a liquor brand. Not to be confused with user dormcat
I'm not that creative when picking out online names - I saw an advert while registering and used the name. 😅
The missing 74f245 transcievers are probably also required for cache to work?
Very likely, guess they skimped on a lot of the smaller cache related components too, but at least the traces are present which should make it possible to get working cache .. assuming they haven't nixxed it elsewhere on the board. Ive seen it before where all traces were present and going to the right places but it didnt matter how many cache components were added the board refused to detect the cache.
Going to see if I can get the DX4 working in my Soyo board as the performance of this one is just going to be woeful. Shame I got conned into a fake cache board but ah well.
I was dubious that the cache was soldered but the traces were there so I thought I'd try my luck. Didn't pay too much for it. Enough, but not too much hehe.
I had a look around on Ultimate Retro and there are a bunch of boards with a very similar layout, with as you already found the Biostar the closest match https://www.ultimateretro.net/en/motherboards/9737
As you say, the traces all look to be there, so maybe you can add some real cache. The differences I can see:
RN1,5,6,4 and 7,8,9 are all bigger than need be. Possibly not a problem, except that 1,5,4,6 all block RP19&20.
RP19&20 are fitted on the Biostar boards (and another board that has space on the PCB for them).
R66 should be removed and R65 fitted (looks to be 10k).
JP7,33,9,10 all need to be fitted.
U21,22 need to be fitted (and those two link resistors removed).
Main problem is not knowing what RP19 and 20 should be. I assume they're some sort of bus termination resistor pack. Maybe someone with a similar board could see if they have any markings on the side.
I took snippets of the boards area and aligned them all, so it should be possible to flick between them to see what changes (there are a couple more in the .zip file):
[edit: just to add, the clock generator UM9515-01 (which I can't find a datasheet for, along with most UMC stuff) looks like it's probably compatible with the MX8315, which there is a datasheet for and it gives a few more possible clock options: https://www.datasheetarchive.com/pdf/download … e=O&term=MX8315 . So the board can (maybe) do 20,25,33,40,50,60,66 and 80MHz.
I did compare a bit more with the Biostar board and it looks like the handful of other changes are mostly to do with changing from using an external RTC to using what I guess is an internal one on one of the UMC chips. For which there are no datasheets]
So how did stuff like this happen back in the day? A competing factory in china decided to do almost a 1:1 clone of a Biostar board to save a few bucks? Surely that couldn't be cost effective? Biostar themselves decided to sell a few 'no name' boards based off their design using surplus stock? To me, this seems like the most logical answer, but I'm keen to know more about how this might have happened. As I understand it, PCChips designed their own fake cache board (with no traces etc) so I guess other manufactures just decided to follow suit since the fake modules were now available to purchase presumably at a fraction of the price of real cache. Because they already had their PCB layouts and other cheap supporting components (resistors etc), it was likely easier to just keep everything else the same and not fit the cache. Obviously a BIOS change was also required so the system didn't expect cache...
Seems like a lot of stuffing around to save a few bucks on cache chips. Keen to hear what others think?
I don't have the skills to fit real cache (as much as I would love to give it a go) so I might see about selling it on to someone who does!
