Back in the day, there were all kinds of weird compatibility issues like this.

1. What are the jumpers on the drive set on? If it is set to Cable Select, set it to master or slave. If it is set to master, try slave. If it is on slave, set it to master. Basically, try all the different settings.

2. Do you have anything else hooked up to the same cable? If so, try it by itself. Also try setting it to "single drive" when on a cable by itself.

3. Try having it installed with absolutely nothing else connected to the IDE controller. I've seen issues where drives did not like to work together on the same controller.

4. Does your motherboard have the latest BIOS available?

The drive is jumpered as master, will try other settings (even CS which I've never used). I'm always keeping the Secondary IDE free for testing devices, so no other stuff on that cable. Latest BIOS (F5) is installed.

Currently I'm torturing the drive via the IDE-USB bridge, nothing seems out of order. What software can I use besides HDTune? (preferably something that repeatedly fills sectors with data then reads them to verify).

Thanks!

f3write/f3read
https://manpages.ubuntu.com/manpages/eoan/man1/f3read.1.html

It's actually for testing for fake USB sticks, but it serves here too. Fills the device with 1GB files, reads them, reports any errors & bad sectors. I run short SMART before and after.

You have weird symptoms. I would pause for sanity check. See if another IDE drive tests okay, See if the RAM makes two passess of memtest86+. That sort of thing. Also, I'd look at the caps in my PSU, and check what the voltages are reporting. Something smells off somewhere.

EDIT
Incidentally, when I really intend to test a drive, I do badblocks. It does four bit-flipping patterns for a nice massage. But f3 is virtually as good, and faster.
10101010
01010101
11111111
00000000

When I test a quantum fireball drive, I put it in the computer and turn it on.

Then I hear the bearing from the old quantum drive and turn the computer back off again. Because that is a noise that deserves to stay in the past!

Aw c'mon, not helpful. The "hard drives aren't worth the trouble" thing has been done to death here. Leave people to their preferences.

Back to topic: Not sure if you're Linux-savvy, but I also vote for badblocks. I like to have a Linux CLI open with two sessions. On one, run "smartctl /dev/sd? -a" to get the SMART stats before you start block-checking the drive. On the second session, run "badblocks -svw /dev/sd?". If it starts going slow or you hear unhappy drive noises, you can jump back over to the first session and re-run SMART to see what's incrementing. Run at the end to compare again. It's normal for seek errors and read error to increase a little, but the normalized values (the ones that typically start at 100 and dip down to some threshold value when things are going south) should not have moved much. If they're getting close to the threshold values, you've got imminent failure to consider.

I don't know if there's a DOS or bootable image that does as thorough job as these tools combined. If so, I'm happy to learn of it.

SirNickity wrote on 2020-02-05, 02:02:

Aw c'mon, not helpful. The "hard drives aren't worth the trouble" thing has been done to death here. Leave people to their preferences.

Yeah, I apologise 😀 Wasn't meant to be anything more than a friendly jibe.

Normally, I would run the drive though a single pass zero wipe, this will give the drive a chance to reallocate any pending bad blocks if it can.

ddrescue --force /dev/zero /dev/sdn (where n is the drive number, check system log to find which one)

then

ddrescue --force /dev/sdn /dev/null (where n is the drive number)

Any drive that doesn't pass with flying colours (any unresolved read errors or excessive slowdown) would get tossed.

I would then use review the SMART attributes of the drive, and also toss any drive that also had excessive error counts.

Doing multiple passes of badblocks or other tools is largely unnecessary. A drive in poor condition reveals itself pretty easily.

I don't do much more than that normally, few drives are worth any more effort than that to fix or get working properly. Of course, it makes sense to visually inspect the PCB for burnt traces, broken connectors (partiularly of the power and IDE connectors) and fix those if the drive is a keeper.

canthearu wrote on 2020-02-05, 02:19:

A drive in poor condition reveals itself pretty easily.

That's been my experience ..although I can say that because I tend to do several passes.

Of interest, a handful of the bad-sector drives stabilized after a couple of passes of badblocks. Ran several more passes, then stuck them on casual duties to see what happens. One has 696 bad sectors.

Yep, that can happen too. A bit of a head crash, but not enough actual damage to cause it to cascade to complete failure. When you are lucky like this, the released particles are scavenged by the internal air filter so they stop doing further damage, and the heads manage to escape in working condition.

If the drive is rare or difficult to replace, using such a drive may be required. It is of course generally more likely to totally fail, but we are working with 20-30 year old hardware, so it can be a calculated risk to keep using an old drive and keep things authentic.

Back when I was young, this was the thing I did because I was poor and hardware was expensive. Now I'm just swimming in the damn stuff, and can easily buy more, so I tend to be more fussy when I can.

I'm totally not a stranger to getting use out of drives on their last legs. In one particular case, I had an old 80meg IDE hard drive that would return a drive not ready error during formatting over a small portion of tracks. So you could never fully format or use the drive. To overcome this, I actually wrote a TSR that captured the int 13h format commands and changed the error code from not ready to CRC error, so the format utility would mark those sectors as bad rather then failing the whole format process. I was able to continue using that drive for a couple more years before it completely failed. It was a success for me at the time for sure.

canthearu wrote on 2020-02-05, 02:53:

Yep, that can happen too. A bit of a head crash, but not enough actual damage to cause it to cascade to complete failure. When you are lucky like this, the released particles are scavenged by the internal air filter so they stop doing further damage, and the heads manage to escape in working condition.

I'm going to go out on a limb here, because I'm not totally sure I know what I'm talking about, but....

I always do the full badblocks 4-pass suite as well, for the same reason. It will often run through the first write pass successfully, then hit errors on the read pass, then rip through the 2nd write pass, then the 2nd read pass succeeds without issue. I would assume this is because the drive is reallocating sectors, and is going to wait until the next write pass to do so.

It may even have a conservative algorithm that wants to see the sector fail more than once (once could be a fluke), so the additional passes could give it a chance to do that. Do any firmwares do that? I have no idea.

Also, I would hazard an ill-informed guess that not all sector issues are physical damage. IIRC, hard drives use a guide track that is written at the factory. Over time, that track can lose its magnetism like any other media, right? I'm not sure where that track is, physically, so it's also possible (if it exists on a writable platter) that the write head could have overstepped and damaged it. Loss of magnetism could be a sign of pending failure (except maybe a material defect in the platter that causes loss of signal fidelity earlier than its average), while mishaps during write could be rare or indicative of poor mechanical stability.

In short: Who knows what's going on in there. So I run all four passes, give the drive a chance to correct any issues it might have, and then if it had any issues, often run another four passes as a confidence test to see whether things have stabilized. Drives that can't survive a single four pass test without repeated errors are immediately discarded. Drives that get their #%$^ together and pass a second run through the gauntlet are, IMO, good to go. As good as any other drive, anyway. 😀

Thanks, I've learned a lot from your replies. I'm not a Linux user, but seems that I will have to delve into this as well. Is there a distro that can run from a LiveCD and help me with the above mentioned commands?
Unfortunately today I had no time at all to further test the drive (or the computer), but will do that soon and will update this thread accordingly.

system rescue cd is probably the easiest option for this.

http://www.system-rescue-cd.org/

Just burn to CD/DVD, boot and run. Or a USB drive if you run it on a modern computer.

Well, some time has passed.
I've conected the drive to the "workhorse" PC via the IDE-USB adapter and the PC froze when accessing the drive. At this point, I'm baffled. Will further test the drive on another PC that I'll build from scratch in the near future (I have a sweet EpoX P55-TX2 board that sits unused in some dark corner). If the drive will behave (and I expect it to), well, I'll be more baffled 😀

Good plan. Never hurts to get a second opinion from a different motherboard. Could just be that board does not like that drive.