Are you putting them on a cable on their own? Some older drives may have trouble negotiating with much newer drives as slaves.

Also I doubt a 120MB drive supports anything above PIO Mode 4, so DMA turned on may confuse it.

Thirdly, where you have to use 2 drives together, sometimes the master needs SP set also.... even if it says you don't need to, it's on there for backwards compatibility.

Then also worth trying the CS if DS doesn't work, which is cable select setting, that should work for a single drive without a CS cable (has a hole punched in it)

I also have seen issues with drives formatted on machines with various size barrier limitations, not reading, or reading as corrupt on later machines without them, due to changes in the block addressing and vice versa. One troublesome one was supposed to be the 32GB barrier on Award BIOSes prior to June 99, which had a 65536 cylinder limit, or some versions tell it, bodged up code for 8GB plus support, that was incompatible with other implementations.

This is very good info! Thank you!

I'm testing drives on their own cable, since its more time consuming troubleshooting master\slave settings for drives from different eras. As you said, some just won't work together.

I tried turning UDMA off in the BIOS, but this board is quite limited due to being a Dell. I'll probably go through at the end and test all of the oldest drives on an older system.

Before SMART existed, was there a good reliable way to determine if a drive was in decent condition without doing a massively time consuming surface scan? I have lots of very very old IDE drives and a bunch of MFM drives around. Some of them are tested and "working" but due to the lack of SMART support I can't say if they are reliable or how many bad sectors they have.

Drives with capacities below ~500 MB often do not support HDD auto detection. Modern BIOSes and all USB controllers rely on the fact that the attached drive supports auto detect.
So the drives probably work fine, you just need an older BIOS where you can enter the CHS geometry of the drive.

Everything IDE should autodetect, maybe autodetect when introduced in 486 BIOSes used to speak louder and slower for the drives like the IDE ST-225, WHAT.... ARE.... YOU ???

That's a point, slower! Older drives took longer to be ready and report from when the reset signal was triggered, modern drives go "blip" and have reset and are recognised, older drives do "Whirrrrrrr durrrrrrr durrrrrr blip" and there's a timeout on later controllers... there used to be a setting to allow longer timeouts for older drives, probably in the PII Super7 era.

Edit: Have a vague memory of needing to turn off fast POST to allow extra time for older drive detection on some machines.

I was actually recently thinking about some sort of "sound emulator" of old drives. Just record various sounds into one of those cheap Chinese sound player modules, add an Arduino to trigger proper sound at proper time...
I never got past the idea, and didn't really do any checks of IDE timing. I'm somehow hoping that without any additional hardware, Arduino itself could handle at least basic PIO modes.

BitWrangler wrote:

Everything IDE should autodetect, maybe autodetect when introduced in 486 BIOSes used to speak louder and slower for the drives like the IDE ST-225, WHAT.... ARE.... YOU ???

You may think so, but it is wrong. Also auto detect is primarily a feature of the hard drive. Auto Detect as a feature in BIOS was introduced in 386 BIOSes.

That is interesting. I have seen Conner and Quantum drives from the early nineties being detected and reported with their full name and model numbers when being plugged in to a ~96 pentium computer. Does this mean that the BIOS on this system has got a database including most of the common drives released before HDD auto detection? Perhaps my drives aren't as old as I think they are?

The same PC also detects and reports drives released long after its date of manufacture. It failed however to detect the ST-506 with an IDE converter that comes with my 1986 Compaq.

Yeah, it's the IDE part of IDE which means it has a microprocessor controller and firmware that knows it's own block mapping then reports a logical version of that to the host, you think your single platter drive has 16 heads? Interesting. So yah, all IDE drives know what they are and can say so. It's not some buzzword for a hard drive it means something, RLL or MFM drives without controllers onboard are obviously not smart enough to do this.

DeafPK wrote:

That is interesting. I have seen Conner and Quantum drives from the early nineties being detected and reported with their full name and model numbers when being plugged in to a ~96 pentium computer. Does this mean that the BIOS on this system has got a database including most of the common drives released before HDD auto detection? Perhaps my drives aren't as old as I think they are?

The same PC also detects and reports drives released long after its date of manufacture. It failed however to detect the ST-506 with an IDE converter that comes with my 1986 Compaq.

My AMD K6-2/300 detected my Maxtor 6L200P0 HDD with the name and everything, as well as the new Maxtor (Quantum) D740X-6L (6L060J3). The BIOS string always detects the drive, type, etc. My Packard Bell Pack-Mate 28 Plus only shows the size of the drive, as well as the appropriate parameters, but not the drive name, model, and so on. So, in other words, it varies by the motherboard manufacturer and processor (Pentium and higher).

Kubik wrote:

I was actually recently thinking about some sort of "sound emulator" of old drives. Just record various sounds into one of those cheap Chinese sound player modules, add an Arduino to trigger proper sound at proper time...
I never got past the idea, and didn't really do any checks of IDE timing. I'm somehow hoping that without any additional hardware, Arduino itself could handle at least basic PIO modes.

I think we had a thread about this in the past year or two. The easiest way to do it would be to attach some kind of circuit to pick up "blinks" from the HD-LED header and then play a sound on a small speaker. Attaching a speaker directly to the pins was also talked about but I can't remember if anyone had actually tried this or not (probably won't work because speakers are AC), and it wouldn't be too authentic. An arduino might be able to decipher the kind of access going on though.

I think its easier to just use a hard drive. 😁

Anyone know what it is in an old hard drive that makes a little tiny short buzz sound periodically? Like, the drive is spinning normally, but once in while there's a tiny additional buzz that's a different tone and lasts a fraction of a second. I've had lots of drives do this. Especially older WD Caviar IDE drives from the mid to late 90s. The drives test out okay in smart and even pass full write-read surface scans. It seems like the sounds eventually stop, so its probably just something they do after sitting for many years, but it always struck me as odd that there would be some kind of change in the drive that caused that noise and yet it doesn't seem to have any impact on the drive's operation.

Not a lot of love for old drives here eh? 🤣

I have another question, probably a bit easier to answer.

I'm running SMART tests on every drive I have that supports SMART, using Hard Disk Sentinel. I've had some drives with 100% health and performance in SMART, went through an entire Write-Read surface test and passed with ZERO bad or weak sectors... then at the end of the test the SMART status updated to 0% and showed one single Raw Read Error. I had one drive that did this, then when I ran the full write-read test again, it wiped out the raw read error which brought the health back to 100%!?

100% to 0% to 100% just by running some surface scans... I have to say it makes me a little less concerned about SMART attributes and status. What do you guys think? I have dozens of old drives from the 90s, the vast majority of which have 100% health according to SMART. If one has an error that makes the computer say it is basically DEAD, and yet it has no other problems, and then the error goes away completely after the next test, what should I do with the drive? I doubt these will generally be used to store integral information ever again, as the ones I've tested with SMART are within the 400MB to 13GB range. At best they'll be a nice drive for an authentic period-correct retro rig or a DOS test bench. I'll keep some and probably sell a bunch of them.

I know people that toss drive after a mere reallocated sector or two, or a few read errors. However in my experience it is not that concerning. Anything important should be on a good backup or two.

Seagate drives in my experience ALWAYS have read errors after a couple years of use. Heck, I retired a 2005 SATA Hitachi 160gb drive with 14k reallocated sectors this year. Didn't trigger smart and had no issues under windows at all, If you are just using a drive to play games or old programs or even just as an OS drive there really is no reason to worry. Drive dies, swap it out.

ODwilly wrote:

I know people that toss drive after a mere reallocated sector or two, or a few read errors. However in my experience it is not that concerning. Anything important should be on a good backup or two.

Seagate drives in my experience ALWAYS have read errors after a couple years of use. Heck, I retired a 2005 SATA Hitachi 160gb drive with 14k reallocated sectors this year. Didn't trigger smart and had no issues under windows at all, If you are just using a drive to play games or old programs or even just as an OS drive there really is no reason to worry. Drive dies, swap it out.

This is sound advice for older computers, for sure. As long as the data isn't integral (or at least has frequent backups) it really shouldn't matter.

I think what's causing me problems with testing these old drives is that many of them don't seem to keep track of bad sectors in the SMART record. I actually just realized this when testing a 100% working drive, only to have a full disk reinitialization find one bad sector. Not bad I thought... but SMART still says the drive was 100%? When I checked the details of SMART on this drive though, it had no entry for sectors at all, only for read errors and some other things.

I have another drive that seems to work fine despite SMART flagging the system BIOS on startup on my test system. At the end of the drive are several bad sectors (lots of them). I know in the old days people would just partition the drive so that this area would be unused. I guess that could work in this case too.

Hard drives a certainly a complicated part of retro computing...

Ozzuneoj wrote:

Anyone know what it is in an old hard drive that makes a little tiny short buzz sound periodically? Like, the drive is spinning normally, but once in while there's a tiny additional buzz that's a different tone and lasts a fraction of a second. I've had lots of drives do this. Especially older WD Caviar IDE drives from the mid to late 90s. The drives test out okay in smart and even pass full write-read surface scans. It seems like the sounds eventually stop, so its probably just something they do after sitting for many years, but it always struck me as odd that there would be some kind of change in the drive that caused that noise and yet it doesn't seem to have any impact on the drive's operation.

It's probably a thermal recallibration.

The bad sectors thing. They can be "soft" bad sectors, frequently windows used to cause a bad sector when it crashed while writing. Also several common virii used to create bad sectors to hide in. I'm forgetting the name of the util you could use to "low level" format it, in quotes because it wasn't a true low level as used on MFM drives but cleared bad sectors that weren't really bad.

Ozzuneoj wrote:

Anyone know what it is in an old hard drive that makes a little tiny short buzz sound periodically? Like, the drive is spinning normally, but once in while there's a tiny additional buzz that's a different tone and lasts a fraction of a second. I've had lots of drives do this. Especially older WD Caviar IDE drives from the mid to late 90s. The drives test out okay in smart and even pass full write-read surface scans. It seems like the sounds eventually stop, so its probably just something they do after sitting for many years, but it always struck me as odd that there would be some kind of change in the drive that caused that noise and yet it doesn't seem to have any impact on the drive's operation.

If I'm understanding right, it's a thing I've always attributed to aging ball bearings. A Fujitsu MPE3136AH with ~29,000 power-on hours I use in my 440BX box produces this sound very often. It's not something I've heard out of any newer HDDs with FDB bearings.

Not just old drives, I remember a whole room full of systems prepared for something - in the Win3.x days before rampant background junk, and they all chattered, a tick tick tick sound from the drives, in glorious octophonic as the machines did it in random order - pretty sure it was thermal calibration as they were warming up.

Can't remember exact drive models now, but there was quite a difference between a Seagate and Quantum I had, think they were around 120 to 150MB models, maybe a medallist and a prodrive... The Seagate would do a fairly quiet and low brrrrrp which you could easily miss if you weren't listening for it... the Quantum on the other hand did a really alarming Snick Snack Snacker Snickwhich no matter how often you heard it, would still have you lunging in panic for the off switch from time to time before you remembered.