• Please review our updated Terms and Rules here

Preserving older hard drives

falter

falter

Veteran Member
Joined
Jan 22, 2011
Messages
6,573
Location
Vancouver, BC
I just bought a lot of random IDE hard drives - 120-500mb for cheap. Sizes I vaguely remember during the run up from 100mb to 1GB, for machines in my collection that came with these drives or had them but were failing. I had sort of forgotten these capacities - I remember the 'important' ones as a child - 10, 20, 40 and 100mb. 100mb was Mt. Everest for me once. And then I remember 1GB as that was kind of a big milestone. But I forgot the in betweens and some of my machines used these.

I'm a bit of a purist - I find I don't get the same nostalgia rush from a CF IDE card or emulation. It's interesting how just sounds and even smells of old equipment can sometimes trigger memories long forgotten. I realize that no more of these lower capacity drives are being made and they're all destined for doom, but I was hoping to acquire enough of a spare stock to avoid having to face modern replacements while I'm still a collector. I have in some cases managed to get older machines to recognize larger drives (like north of 4-6gb) but even these are getting super old and scarce these days and they don't have quite the same persona those low capacity drives do.

Does anyone have any recommendations on how to sort of keep the drives alive longer? Is it a matter of using them less, or more? Storing them a particular way? Or is my best bet to just pile up a collection of them and hope it lasts? Are there certain brands/models that seem to do better than others (obviously I'm avoiding Miniscribe.. baha)?
 
When time is involved, your biggest enemy is chemistry. Polymers will break bonds, oils will decompose and binders will degrade (this is why plated media last longer than plain oxide coatings). Rubber will get gummy or brittle; plastic parts will lose their plasticizers--and then there's the problem of corrosion. My best advice is to store the drives in a cool, dark, dry place. As to whether or not to fire the things up periodically, recall that most electronic devices fail with the initial power surge, so your guess is as good as mine.

The big MIniscribes were pretty well made--so well-made that Maxtor put their name on them. Now, if you're talking about JTS or Octagon drives, I'm not so confident...
 
I just store mine in the basement stacked up and open to the air. Over the years I have snagged anything 20-500MB for my old DOS rigs thinking the ones in my machines will die sooner or later.
 
I use static bags , and a packet of silica and sheet of cardboard under the pcb in each. 10-12 to a plastic shoebox dollar store container. Once every year or 2 ill pull them out and exercise them for 30 mins then store them back away. Have had pretty good luck since I have been doing it this way.

EDIT: Also lay them power connector side down. Dunno if that does anything but I like to think it help stiction.
 
Last edited:
Don't forget that these drives (anything after MFM/RLL not using a stepper) have servo tracking data on the magnetic media. Unlike user data, you can not restore that tracking data by doing a format or something. It will fade just like data on a forgotten floppy disk and once faded, the drive becomes a brick. There's nothing to avoid this, it will just happen.
 
Agreed, if the grey code field fails the drive is essentially a brick. Might still be useful in a zombie apocalypse.
 
Servo tracks (any magnetic media) lose half their magnetic properties in something like 70 years so not an issue for most of us.
 
Chuck(G) said:
When time is involved, your biggest enemy is chemistry. Polymers will break bonds, oils will decompose and binders will degrade (this is why plated media last longer than plain oxide coatings). Rubber will get gummy or brittle; plastic parts will lose their plasticizers--and then there's the problem of corrosion. My best advice is to store the drives in a cool, dark, dry place. As to whether or not to fire the things up periodically, recall that most electronic devices fail with the initial power surge, so your guess is as good as mine.

The big MIniscribes were pretty well made--so well-made that Maxtor put their name on them. Now, if you're talking about JTS or Octagon drives, I'm not so confident...
Click to expand...

Is there an actual document anywhere describing what actually fails in 80's disk drives, and in what drives?

I've never come across anything beyond anecdotes like head sticking to media (conner) and rubber bumpers (micropolis)

Troubleshooting anything with a servo track is difficult because there are so few schematics around. It's taken me years to come
up with the small schematics collection on bitsavers
 
I'm just extrapolating and adding my own experience. I've certainly got old drives where the bad spots have "crept" seriously and new ones developed. Spindle losing lubrication is a thing--I've got a Priam 100MB drive where that happened--started up, then started spinning slower and slower...
On the other hand, I have drives that have put in 35 years without issues.
 
More of a repair question than preservation per-se, but as those two will be ever more closely linked as time goes on:
I know on some older pack-drives, the servo had it's own dedicated platter, but I think on newer drives this isn't the case?
I'm faced with a data-recovery project where a head failed and left debris all over the drive, but didn't spin for long after so I suspect many surfaces are still fine. In newer drives they just pull out the platters for cleaning and inspection, then place them in a new drive. I'm wondering if this is feasible for older MFM drives (specifically a Miniscribe 1335/RD53) or if I'm going to loose sector timing and such?
If I can't move platters with comparative impunity, I'll be forced to try to inspect and decontaminate them on the spindle.

If it was just one surface though, it would mean with a high-precision positioner it wouldn't be that hard to make a thing to write servo data on a single-surface. ISTR Larry at Crisis Computers had one for larger pack drives BITD.
 
The old surface-for-a-servo drives have been superseded by the "embedded servo" technology. Ref.

On small drives, writing the servo (whether embedded or separate) involves very high precision. The one I saw in the Tandon lab was laser-interferometer positioned. You need that kind of accuracy with embedded servo. With the embedded variety, you really can't do a true low-level format; it would destroy the servo data--which is why IDE drives don't allow it.

A tidbit--the Drivetec high-density 5.25" floppies used embedded servo. If you stuck one in a regular floppy drive and formatted it, the servo data would be destroyed. There goes your $15 investment. I think that the UHD 3.5" floppies used the same technology, as there are warnings on the packaging not to format them.
 
Chuck(G) said:
The old surface-for-a-servo drives have been superseded by the "embedded servo" technology. Ref.

On small drives, writing the servo (whether embedded or separate) involves very high precision. The one I saw in the Tandon lab was laser-interferometer positioned. You need that kind of accuracy with embedded servo. With the embedded variety, you really can't do a true low-level format; it would destroy the servo data--which is why IDE drives don't allow it.

A tidbit--the Drivetec high-density 5.25" floppies used embedded servo. If you stuck one in a regular floppy drive and formatted it, the servo data would be destroyed. There goes your $15 investment. I think that the UHD 3.5" floppies used the same technology, as there are warnings on the packaging not to format them.
Click to expand...
Ah, but when was that? I imagine the old ST-412 drives with a stepper-motor just counted steps (?) so the question is were there surface-for-servo drives in the 5.25" or 8", or was it all embedded servo by then?
Still, it suggests that my target drive it likely embedded-servo. I guess I'll un-stack a test drive and see how it goes.

I've got a spare interferometer or two, but I'm unconvinced that you'd need that level for something like a ST-225. A lot of older drives topped out at about a thousand tracks on a surface. If a surface has about 1" of usable travel, that's only .001" and even if you wanted 10X that precision, you can easily get .0001" with a COTS linear scale these days. Not worth it yet, but perhaps someday.
I do live in dread of the day my RL02 packs start failing.
 
One way to get a pretty good indication of a drive using separate servo surface is to look at the specs. Martin Bodo's "Hard Drive Bible" is a pretty good source for early drives.
If the number of heads in the spec is even, then no separate servo surface. Odd; otherwise.

My visit to the Tandon lab on DeLaCruz (close to the Coleman Still) was around 1984 or so. The servo writer was assembled on a granite surface plate and the actuator was powered by a cylinder of compressed nitrogen. Again, this was a lab, not the production line. Mostly the guys there arrived late, took long lunches and left early. I suspected a big tax deduction for Tandon.

Of course, for larger IDE drives, the geometry doesn't mean a whole heckuvalot.
 
Last edited:
Chuck(G) said:
One way to get a pretty good indication of a drive using separate servo surface is to look at the specs. Martin Bodo's "Hard Drive Bible" is a pretty good source for early drives.
If the number of heads in the spec is even, then no separate servo surface. Odd; otherwise.

My visit to the Tandon lab on DeLaCruz (close to the Coleman Still) was around 1984 or so. The servo writer was assembled on a granite surface plate and the actuator was powered by a cylinder of compressed nitrogen. Again, this was a lab, not the production line. Mostly the guys there arrived late, took long lunches and left early. I suspected a big tax deduction for Tandon.
Of course, for larger IDE drives, the geometry doesn't mean a whole heckuvalot.
Click to expand...
Of course you are right, I feel foolish for not thinking of that. The drive in question is listed as having 8-heads, but opening the drive shows there is very much a ninth. So, my expectation is it's separate-servo and while you might retain a useable drive if you did a platter-move, you are very unlikely to keep your data intact.
 
You must log in or register to reply here.
Back
Top