feipoa wrote:

I removed a UM8663BF Super I/O IC from a revision 3.1 Biostar MB-8433UUD motherboard. The silkscreen shows UM8663AF/BF, so perhaps you can use either of these. This particular MB had a defective floppy controller, so I am hoping the replacement fixes the issue.

Thread Hijack Warning

Hello feipoa, I just acquired a revision 3.1 Biostar MB-8433UUD motherboard recently but cannot find any documentation online explaining what if anything makes this board different from a rev 2.
Are the BIOS updates compatible between revisions, and do they support the same CPU models ?

Thanks

Off the top of my head,

1) Revision 3.1 has a socket for the RTC module.
2) Revision 3.1 always has three jumpers for the FSB. Some version 2's have three, some have two.
3) Revision 3.1's came with later datecodes on the chipsets
4) I noticed some difference between the impedance matching series resistors between the models
5) Revision 1 has the terminator pack socket allowing you to run the UMC CPU.

All seem to have the same CPU support, except for revision 1, which supports more CPUs (UMC U5S). I've had no problem using newer BIOS revisions on the older board revisions.

Damn that was fast.
Thanks feipoa.

Now, for your take on an age old question.
Which CPU would you chose to run on this board (assuming all setups are stable) with preference to passive or near silent cooling:

• Am5x86 40x3 120MHz

• Am5x86 33x4 133MHz

• Am5x86 40x4 160MHz

• Cy5x86 33x3 100MHz

• Cy5x86 40x3 120MHz

Given your avatar, the answer may be obvious but i still wonder...
Keep in mind this machine is being geared towards flexibility for early 486 to early pentium-era DOS games.
From Wing Commander to Wing Commander 3, with some Sierra and Lucasarts peppered in there for good measure.

Do you intend to run a 3D accelerator? If main objective is for passive cooling, then you might want to offload some task/heat to the video card. The Cx5x86-120 or Am5x86-160 run well on this board, provided you stick with 256K for cache. Not many boards run well with a POD, but this board is the exception.

If you are interested in the diode mod, I recommend the POD100 with Voodoo or Voodoo2. Modifying the POD83's voltage regulator for overclocking However, that isn't passive.

For passive, the Am5x86 is probably the cooler running chip, however, I haven't run it passive before. Though, the QFP version of the Am5x86-133 came with a tiny heatsink on an Evergreen module. It got damn hot. I assume Evergreen cleared it for this passive environment. So if you run it at 160 MHz with a socket 370 cooler, that might be the best solution. perhaps run some tests with the Cx5x86 under the same heatsink to see which can handle passive.

I may slap in a voodoo1 with an S3Virge.
I was hoping to put in 512kb of cache. Is that a problem on this board (regardless of speed) ?
Not sure how the cache on this board correlates with cachable ram. I was hoping to use 96mb EDO, for the odd case when I'd load windows 95.

I previously was able to use an AMD 5x86 at 133 with passive cooling but with a fairly large heatsink glued on. It still got quite hot though.

512KB single-banked works fine on the fastest cache settings if your FSB is 33 MHz, not 40 MHz. If using 40 MHz, you'll need to use a slower cache timing in the BIOS, perhaps even as slow as 3-2-2-2 for reliable operation. EDO RAM does not help speed on this motherboard. EDO RAM seems to demand slower cache and or RAM timings in the BIOS. I do not recommend using EDO on this board.

512KB double-banked works fine with 40 MHz and the fastest timings, but you will need to modify the board slightly to use 512K double-banked. I tested an Am5x86-160 at 2111 with one 128 MB FPM stick and 512K double-banked; it ran fine. When getting close to the limit of reliable timings, it is best to use as few sticks as possible. I also tested 1024K with Am5x86-160, which seems to work with 2x128 MB sticks, but I had to use 3222 for the cache timing. When I say "work", I mean Quake didn't crash and Memtest passed one round.

S3 + Voodoo1 sounds like a nice combination.

Thanks for the advice feipoa.

I must admit that I really like this motherboard so far.
I want to benchmark how it does against my HOT-433v4 and M919 cacheless.

Not that I care for absolute speed, but rather long-term stability. It would be nice to try and build the ultimate stable-but fast 486.

Hi,

i also got a UB ~ 1 Year ago and now trying to restore it.

It had one broken ISA Slot, one ripped off the RTC with a Screwdriver and ripped off the Memory.
So 2 PS/2 Slots are bend and flaky.
Also Clock Generator IC where missing and BIOS.
Now it boots straight to MS-DOS.

Have still to add a ISA Slot and replace PS/2 Banks 😀

Makes me wonder what happened to that poor lil' MB.

Someone tried to desolder the RTC without proper Equipment... i guess.
He used a Screwdriver to force the RTC off the Board while heating it up. 😒
It seems the Screwdriver slipped of and he hit the ISA Slot..

Now need to find some decent CacheChips 😀

Re-Owning the thread 😀

I dug the mobo back from its cave.
If I disable the floppy seek at boot, I do not get a "Floppy Drive Fail (C0)" error. --> This means the controller is aware something is wrong. The floppy drive still does not work though.
Also, I am unable to disable the COM ports. Although they are set to disable in the BIOS, they still appear as active on the post screen.
I can disable Parallel port and Floppy Controller.

If replacing the super i/o didn't work for @Feipoa and me, I wonder what else talks to the floppy drive ? Could it be the UM8667 that is messing things up ?
All traces from the floppy connector go to either GND, a resistor array or to the super IO chip. I don't see anything else. Traces on my motherboard look good, not broken.

@feipoa Did you get to fix your FDC controller since ?

Yes, my FDC works fine on this board after replacing the floppy IC and ensuring the leads are making electrical contact with the solder pads.

Synoptic wrote on 2020-11-15, 02:41:

If replacing the super i/o didn't work for @Feipoa and me, I wonder what else talks to the floppy drive ? Could it be the UM8667 that is messing things up ?
All traces from the floppy connector go to either GND, a resistor array or to the super IO chip. I don't see anything else. Traces on my motherboard look good, not broken.

I'm not of any help for this issue but it would be really interesting to hear of the equipment and process you or anyone else who'd like to join in on what it takes for something like trying to replace one of those QFP i/o chips or an entire ISA slot. I wouldn't even know where to begin. Replacing just a capacitor or resistor is enough difficulty for me.

You can use quickchip, which is a special type of solder that stays molten long enough for all 4 sides to stay liquid so that the chip can be lifted off. Quickchip is expensive for what you get, so I'd save that for a chip that doesn't have a replacement. For something like this, if you already have the spare IC's on hand, you can just cut the leads of the dead IC to remove the package, then use a soldering iron, wick, or desolder station to remove the left over leads.

feipoa wrote on 2020-11-15, 03:13:

Yes, my FDC works fine on this board after replacing the floppy IC and ensuring the leads are making electrical contact with the solder pads.

I reordered 3 super IO chips just in case.

Could you help me by listing the list of pins I should check for contact/signal that are related to the FDC ?

I doubt I have the pinouts for anything UMC. You just follow the trace to where it ends, or at a via, and check for continuity on the other end.

feipoa wrote on 2020-11-15, 06:29:

I doubt I have the pinouts for anything UMC. You just follow the trace to where it ends, or at a via, and check for continuity on the other end.

I verified continuity from each even pins of the FDC header to the UMC chip. All is good.
Every odd pin is grounded.

I am awaiting new chips, but this one is supposedly new. I swapped it 4 years ago when I began the project.

@Feipoa : Could it be a BIOS issue ? which one do you recommend to use for troubleshooting. Mine is a UUD-A VER:2

There are more pins on the UM8663BF than just the FDC pins. Ensure each of the UM8663 leads are making electrical contact with the pads.

I have a V2 and I damaged it years ago trying to desolder the rtc. Right in the same area and qfc chip.
The Floppy never worked and I never pieced it that its from the damage in that area. I never really looked far into it since all my floppy disks are now unreadable and the only thing the floppy is good for the the boot up sound.

However if it helps you can confirm similar damage and fdc error.

This is my exact error when enabling floppy in bios

CB18? It kind of looks like that component is an SMT ferrite bead. They look identical to SMT capacitors but aren't.
Of course I'm not saying that I am entirely sure that is what CB18 is.

Also, try checking the clock inputs for the UMC I/O chip. If both the COM and floppy ports are dead, that would cast some suspicion on the clock signal that drives both. I wish I could GET datasheets for UMC parts though, or else I'd tell you what pins to check!