IBM floppy controller failure modes

[RichCini]

While working on the project referenced in my other thread (about PC serial numbers), I noticed the floppy controller (IBM standard-issue 6181682) was inconsistently working...until it didn't. Then I tried two more cards, all with the same issue -- worked and then wouldn't survive a reboot -- the PC would land in BASIC rather than reading the floppy. I finally found one that did work, so that's what I'm using now. But that leaves me with three non-working cards that I'd really like to get working as a spare.

Is there any particular thing I should be looking at if I try to repair these? Is there a common point of failure? Naturally nothing is socketed, so I need to be surgical about this.

Thanks!
Rich

 

[modem7]

I finally found one that did work, so that's what I'm using now. But that leaves me with three non-working cards that I'd really like to get working as a spare.

Comparatively speaking, these cards are very reliable. To me, three is a suspiciously high figure for one person to have, and I wonder if something else is going on (e.g. drive, cable, 'dirty' contact/s in slot, etc.)

Have you swapped these 'apparently faulty' cards (card and nothing else) into a another (working) computer, to prove that the card is at fault ?

 

[RichCini]

Comparatively speaking, these cards are very reliable. To me, three is a suspiciously high figure for one person to have, and I wonder if something else is going on (e.g. drive, cable, 'dirty' contact/s in slot, etc.)

Have you swapped these 'apparently faulty' cards (card and nothing else) into a another (working) computer, to prove that the card is at fault ?

I agree its very odd to have that many. Two other cards seem to work just fine with this system, across reboots and power cycles. To test further, I'd have to pull together another test system -- which I can do -- but once this other project is done. Just don't have the bench space. I just thought there might be some well-known failure point for me to start with.

Rich

 

[modem7]

I just thought there might be some well-known failure point for me to start with.

Not that I am aware of.

I agree its very odd to have that many. Two other cards seem to work just fine with this system, across reboots and power cycles. To test further, I'd have to pull together another test system -- which I can do -- but once this other project is done. Just don't have the bench space.

You could use a statistical method to prove (within reason) that the cards at fault:

For the following test sequence, only the card in use changes. E.g. No other variables (such as slot #) are to change.

Try the following test sequence, say, 10 times in a row. If the result is always the same, cards #2, #4, and #5, must be faulty.

Step 1: Use 6181682 card #1. Floppy operation works.
Step 2: Use 6181682 card #2. Floppy operation fails.
Step 3: Use 6181682 card #3. Floppy operation works.
Step 4: Use 6181682 card #4. Floppy operation fails.
Step 5: Use 6181682 card #5. Floppy operation fails.

 

[modem7]

(IBM standard-issue 6181682)

Per [here], there are two variations of the card. Your use of "6181682" indicates to me that all of your cards are of the later variation.