VOGONS


How to fix your dead on-board floppy controller

Topic actions

Reply 80 of 90, by amigasith

User metadata
Rank Newbie
Rank
Newbie

A little bit of thread necro here, but I hope you'll like it 😀

Put into a couple of sentences, I had to replace the Super I/O chip on my MB-8433UUD-A and it turned out to work great! Here are my personal 2 major conclusions from this adventure:

- Despite a couple of people reporting in this thread that replacing the Super I/O chip didn't solve their floppy disk drive issues, it definitely solved it for me.

- Although my board only had "UM8663AF" on its silkscreen, a UM8663BF works perfectly fine on my board.

Here's the long story in case you're interested:

Since a couple of weeks, I'm a proud owner of a Biostar MB-8433UUD-A mainboard. It's the original version aka Rev 1. Unfortunately, shortly after receiving this beauty, the floppy disk drive stopped working properly. While I still got the familiar grunting noise during POST when the floppy disk drive is being initialized, I only got I/O errors when the board tried to boot from floppy. On top of this, COM1 didn't work either, no matter which mouse I connected to it. COM2, on the other hand, still worked fine.

"Okay" I thought and started reading a lot about this problem, including this thread here, of course. Right from the beginning of this long journey, I suspected the Super I/O chip UM8663AF to be the culprit, but I wasn't really sure whether it really was the root cause of my floppy disk drive issues or not. Therefore, I made quite a number of experiments, in order to rule out any other possible issues:

- I checked all traces between the 34-pin floppy cable connector on the board and the UM8663AF, and they were all connected properly.

- Next, I double-checked whether all pins of the UM8663AF made good contact to other places on the board or not, aka "continuity testing". This step was the one that took me a lot of time to complete and I had to spread it over a couple of evenings. But it was worth it every minute 😉 Since some pins of the UM8663AF are connected to vias that are placed underneath of the IC, I had to look for some pictures with desoldered UM8663AF chips, in order to properly follow the traces. Our dear forum here turned out to be a great source for this 😀 See for example feipoa's post here.

- Since all pins of the UM8663AF turned out to be connected properly to other places on the board, I tried booting from floppy disk drive with a Multi I/O card and this worked like a charm.

- In addition, I checked all resistors, capacitors, and diodes for shorts / open circuits. I know that doing this in-circuit is not a 100% bullet-proof method, but still better than nothing. None of the parts were showing errors, however.

So in summary, all important connections / traces "along the floppy path" seemed to be intact: from the 34-pin floppy cable connector to the UM8663AF, from the UM8663AF to the ISA slot, and from the ISA slot to the southbridge. The proper connectivity from the ISA slot to the southbridge was proven by the Multi I/O card experiment. Realizing all this made me feel more comfortable that the root cause of my floppy disk drive issues really was the UM8663AF. And so I decided to give it a try and replace it.

Here is a picture of the area on the mainboard before the "surgery":

8433_01.jpg
Filename
8433_01.jpg
File size
196.4 KiB
Views
731 views
File license
Fair use/fair dealing exception

My first action was to simply reflow all pins on both the UM8663AF and the UM8667, which controls the serial ports. This did not help, however. I still got the mentioned I/O errors when trying to boot from floppy.

Next, I desoldered the UM8663AF. This is how the silkscreen underneath of the UM8663AF looks like on my board:

8433_02.jpg
Filename
8433_02.jpg
File size
167.74 KiB
Views
731 views
File license
Fair use/fair dealing exception

As you can see, it only says "UM8663AF", but not "BF" as it does on other pictures I have seen both on vogons.org and other places on the Internet.

Here is the desoldered UM8663AF - I used roses metal, a lot of no-clean flux, and hot air for desoldering it:

8433_03.jpg
Filename
8433_03.jpg
File size
75.06 KiB
Views
731 views
File license
Fair use/fair dealing exception

Last but not least, here is a picture of my board after the "surgery":

8433_04.jpg
Filename
8433_04.jpg
File size
196.57 KiB
Views
731 views
File license
Fair use/fair dealing exception

As you can see, I used a UM8663BF and not a UM8663AF that used to be on the board before. Why? Well, first of all, simply because I couldn't easily find a UM8663AF and UM8663BF's are still widely available, even on ebay, for example. And secondly, because somebody on vogons.org mentioned that the UM8663BF's are more reliable than the UM8663AF 's. And so I thought "why not give it a try".

Of course, I was quite nervous when I started the actual replacement of the UM8663AF, mostly because of 2 things: (1) Whether the ICs that I received were actually legit and not some relabeled dummies and (2) whether a BF would actually work as a replacement for an AF. But everything turned out to work just fine - my MB-8433UUD-A works again like a champ 👍 I can now finally boot from floppy disk drive again without any problems and also COM1 is now working again as it should. Mission accomplished!

If you've made it down here: Thanks for reading 👍 And sorry for the rather long post, but I'm so happy now that I had to tell the whole lengthy story 😊

Last edited by amigasith on 2021-05-20, 20:33. Edited 1 time in total.

Reply 81 of 90, by feipoa

User metadata
Rank l33t++
Rank
l33t++

It is nice to hear another users solved their issue by replacing the Super I/O chip. The issue of a defective COM port or diskette controller is becoming more and more common on these older motherboards, and is not specific to the MB-8433UUD. I have quite a few 4DPS boards in need of similar work, and they use a different chip.

Plan your life wisely, you'll be dead before you know it.

Reply 84 of 90, by pentiumspeed

User metadata
Rank l33t
Rank
l33t

I had to do same with my nearly new Asus FX chipset motherboard back in the day after ESD zapped through parallel port and replaced the multi-i/o chipset from junk board brought it back to living.

Cheers,

Great Northern aka Canada.

Reply 85 of 90, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Mine just died while in the case sitting in its usual spot after years of use, and one day I rebooted and tried to boot from the floppy and it was spitting out errors. Perhaps electro-migration killed it.

Plan your life wisely, you'll be dead before you know it.

Reply 86 of 90, by amigasith

User metadata
Rank Newbie
Rank
Newbie

Thanks for your comments, guys 👍

I forgot to mention that I also checked all passives for shorts and open circuits. I now added this to my original post. Not sure if this is a common fault on old 486 boards, but it definitely happens on Amiga boards, especially when they're flexed too much when they're put in and out of their cases. This can lead to cracked components, for example. Not that it happens very often, but it happens.

EDIT: The UM8663AF Super I/O chip by the way was the first IC that actually died on me in an old computer "just like that". It would be really interesting to know why this happened, but I guess I will never solve this one... I mean I had faulty ICs before, simply because I did something stupid like plugging in wrong devices but never out of the blue.

Cheers!

Reply 87 of 90, by CharlieFoxtrot

User metadata
Rank Member
Rank
Member

Gladly I found this thread as I have most likely a fried UM8663BF chip on my Gigabyt GA-5486AL that I working on.

Floppy doesn’t do anything even during boot, it just stays silent and no led lights up as if it is not connected at all. Also, the boot splash screen shows no LPT port available, although it is enabled in BIOS.

Everything else seems to work fine, on board IDE works as well as ISA and PCI slots, so I think SB is okay.

I also couldn’t find any abnormalities on the board. Few traces near and from BIOS towards the UM8663BF are darkened due to battery leakage, but shouldn’t be an issue, although I might solder some jumper wires at least temporarily to rule those connections out.

I already ordered replacement chip, so hopefully I can get floppy controller up and running again, shouldn’t be too hard of a job, although my experience with SMD chips is relatively limited.

If you guys any other ideas, I’m open to suggestions.

Reply 88 of 90, by Thermalwrong

User metadata
Rank Oldbie
Rank
Oldbie

Since you're seeing a non-functioning LPT port and the floppy isn't working, try booting off the hard drive and see if PCM sound works, if it doesn't then perhaps something is wrong with DMA which would break floppy access. That's a late enough board that that stuff should all be integrated though.

One thing I see rarely mentioned anywhere is that there's a BIOS setting on many 486 boards for "Burst Write" in the chipset options which I think should be for the Cyrix 5x86 CPUs.
Enabling that on an Advantech PCM-4825 / PCA-6145 (VIA chipset 486 boards) and a Freetech 486F55 (SiS 496 chipset) caused the same issue on all three, where the floppy drive could not be read at all while that option was enabled. I'm pretty sure that's not a hardware fault on all those boards since disabling that setting allows the floppy drive to be booted from and read without issue.

Reply 89 of 90, by CharlieFoxtrot

User metadata
Rank Member
Rank
Member
Thermalwrong wrote on 2023-05-06, 00:39:

Since you're seeing a non-functioning LPT port and the floppy isn't working, try booting off the hard drive and see if PCM sound works, if it doesn't then perhaps something is wrong with DMA which would break floppy access. That's a late enough board that that stuff should all be integrated though.

One thing I see rarely mentioned anywhere is that there's a BIOS setting on many 486 boards for "Burst Write" in the chipset options which I think should be for the Cyrix 5x86 CPUs.
Enabling that on an Advantech PCM-4825 / PCA-6145 (VIA chipset 486 boards) and a Freetech 486F55 (SiS 496 chipset) caused the same issue on all three, where the floppy drive could not be read at all while that option was enabled. I'm pretty sure that's not a hardware fault on all those boards since disabling that setting allows the floppy drive to be booted from and read without issue.

I had today time to tackle the issue. I got the replacement chip and soldered it in place, but it didn't fix the issue. I decided to check the traces that the battery damage had corroded and lo and behold, three traces between BIOS chip and IO controller did not have continuity. I soldered some jumper wires and now everything seems to work as should.

Although trace damage didn't LOOK that bad, I should've check those first. Well, at least I got bit of practice with SMD desoldering/soldering, but in hindsight, trace damage was the most likely cause as floppy controller was completely dead. I've repaired quite a lot of vintage stuff (mostly 8-bit stuff), but it seems that there are things to learn almost every time that makes the next problem solving perhaps a bit more straight forward. At least in the sense that never assume anything and think of the most likely cause with the information you have. But in the end, I'm one happy camper as another fine MB is in working order.

But thanks for the suggestion!

Reply 90 of 90, by tauro

User metadata
Rank Member
Rank
Member

I want to report a success story. After as much troubleshooting as I could do (recap, check traces, passive components), I replaced the UMC UM8670F Super I/O controller on a PCChips M550 v5.6. The original was from 9719-AXA / ZM3Q23.

First I replaced it with one from a dead board (not sure if it worked or not), 9645-AXA / ZM9U67, and... it didn't work.

Finally I got one from a semi-dead M550, I wasn't sure if it would work, it's from 9712-AXS / ZM2820 and to my surprise... this one worked! I couldn't believe it.

So whoever might be reading this, it's 100% feasible that your little, undocumented, stubborn Super I/O chip is the culprit.

I wouldn't underestimate the importance of checking every pin to make sure they are soldered correctly. Sometimes a little bent pin is the difference between failure and success.

It's not an easy process but it's worth it 😀