Salvage: Conner CFS210A, Maxtor 7080AT, Seagate ST52520A, Quantum ProDrive 2100S Hard Drives

As part of a large bundle of stuff which was generously salvaged on my behalf, these are the hard drives that I managed to get. Definitely a trip down memory lane for some of these drives, and others, completely fascinating.

Conner CFS210A

Conner Peripherals drives were a common sight if you worked around Compaq branded computers. Very often, they would be included as Compaq were a significant backer of Conner. Conner Peripherals was started up by Finis Conner, the co-founder of Seagate Technology and merged with CoData which was a Miniscribe-based startup. I did have contact with a 40Mb Miniscribe Winchester hard disk, but lets just say that their company’s accounting operations became famous when it collapsed. Interestingly, Conner was reabsorbed into Seagate in 1996. Miniscribe itself was purchased by Maxtor in 1990.

Aside from that, Conners were not popular choices amongst system builders and were more often the cheap, slow and sometimes unreliable choice. The main philosophy in Conner drives was to reduce cost by innovation – their “flat” base and tub-shaped lid were ways to reduce the cost as it didn’t require milling out or casting a complex base, and their use of microcontroller firmware based servo, spindle and self-test operation was also a distinguishing feature.

DSC_8896I was expecting this one to be dead … but no. It functions perfectly even today.

Manufactured in Singapore on Week 25 of 1994, this drive is 20 years old. Its geometry is 685/16/38 giving a size of 210Mb (hence the model number CFS210A). The power requirements are a very modest 410mA on the 5v rail and 215mA on the 12v rail, no doubt because its spindle speed (3600rpm) is probably quite slow by modern standards. It even has a 32kB cache.

Amazingly, specifications can still be found on Computerhope and Red Hill Technology’s websites.

DSC_8897

The underside is a plethora of ICs, with a pair of Atmel flash chips and Motorola microcontrollers in control. There is a Hyundai memory chip, likely to be used for cache. The Cirrus Logic chip is probably for the IDE interface, whereas there are dedicated chips for spindle motor, voice coil motor control and possibly even head driver/amplifier.

DSC_8898

On the IDE interface side, there’s not much to note – the molex is the “regular” bevelled edge-up orientation, although there appears to be a three pin white connector on the far left which might be an alternative power connector with just 5v, 12v and a single ground.

DSC_8899

DSC_8901

Viewed from the side, the flat-bottom-plate construction with a tub-shape lid and the use of a minimal amount of screws (four torx) to secure it with a gasket is apparent.

DSC_8900

The front of the drive has the serial number barcode label and a host of jumpers used to configure the drive. The majority are reserved, and likely, used for diagnostic purposes in the factory.

Maxtor 7080AT

Those with a memory of the recent Maxtor drives, prior to their acquisition by Seagate, would probably remember some of their drives as being tragically unreliable, but early Maxtors were like clockwork. This Maxtor 7080AT is a good example, amongst others, of drives of this era still persisting today.

DSC_8902Maxtor, formerly known as Maxtor Colorado Corporation, acquired Miniscribe in 1990 after they went bankrupt and then, itself was close to bankruptcy in 1992 (not long after this series of drives were made). They went on to acquire Quantum Corporation (which was Matsushita’s hard disk arm), and were then acquired by Seagate. It’s fascinating to think just how we came to have so few hard disk manufacturers today.

This drive is an 80Mb drive, and in fact, I’ve had two 120Mb drives in the past from the same series which I had disposed of eventually. It has the classic Maxtor/Miniscribe HDA, PCBA, UNIQUE, and UPLEVEL coding on the label. It was manufactured 28th May 1991.

The geometry of the drive is 981/10/17 according to Computerhope, and its power consumption was a mode hefty 500mA on the 5v rail, and 1A on the 12v rail. It seems to be one of the older RLL encoding drives with a 3703RPM spindle speed (stason.org). Odd rotation speeds were common in early drives, which made for annoying “beat” frequency hums from boxes with dissimilar drives installed. It also has quite a few jumper configurable geometry options for computers which needed them.

DSC_8903

A similar design approach can be seen, with a flash firmware chip and a Motorola microcontroller in plain sight. A similar Cirrus Logic chip also handles the interface, and there is a Hitachi cache RAM chip. A few configuration jumpers sit near the interface, and an LED header is on the front of the drive. It looks a little bit more crowded than the Conner.

DSC_8904

This one shows just how careful one has to be with their molex connectors. Many earlier drives had the molex connector oriented the other way around so that the bevels face downward. This one also has the three pin power connector which isn’t seen in systems today. Note that the drive was manufactured in Singapore, similarly to the Conner, but very few drives I have seen today are manufactured there anymore.

DSC_8907

DSC_8905

The paper labels serve as warranty seals, but looking at the drive from the side makes it clear that the drive follows the traditional method of producing the complex base and closing it up with a flat (or near flat) lid.

DSC_8906

A barcoded front serial label is visible with a colour coding, and the front facing LED header pins.

Seagate ST52520A

Seagate drives are pretty common, as are Western Digital drives, but this one is a bit of an oddball. Part of the Seagate Medalist Pro series, this drive was a refurb (as you can tell by an additional label) and unfortunately sounds like an angle grinder. It isn’t functioning anymore, but it seems to be an attempt to reduce the volume of the drive (maybe to save on shipping and logistics costs). It’s obvious when placed aside a regular 3.5″ size drive.

DSC_8908

It’s a 2564Mb drive, and it seems like it’s dated to 1999 after refurbishing, although I can’t be sure. It was also made in Singapore – how amazing. Not only is it shorter than a regular drive, and exposes the rear connectors, it’s also …

DSC_8909

… thinner! It has the silver tape around it, as was common for medalist series drives, but it also had screws holding the top lid in. Later medalist drives had a can which would “fall off” if you undid the tape (they are recognizable with a black round plastic covering over the air inlet chambers).

DSC_8913

The underside seems to make a few different assertions as to date and time. The etched code implies Week 33, 1996, whereas one of the painted markings implies Week 16, 1997 and the other, Week 43, 1999. It looks like this drive may have been manufactured initially around 1997, and been refurbished two years later.

The underside of the drive follows the Western Digital philosophy of later drives, which is to place all the components on the other side. This stops careless system builders from accidentally knocking off a few components. This was before they introduced the later Seashield which was a metal plate or rubber “harness” over the drive to protect the electronics from ESD and accidental impact. That proved to be impractical later on (maybe due to cost), and present drives almost always have the components hidden away on the “other side”.

DSC_8912

DSC_8910

From the edge, its slender thin profile is most evident.

DSC_8911

The rear layout is a much more familiar layout with the molex the right way up, and configuration jumpers pointing to the rear. The jumper table is provided on the label, as usual.

Quantum Prodrive 2100S

Quantum is a brand which was highly renowned for high performance and fairly reliable drives, and I used to ask for them by name. Later, upon disassembling several in my youth, I managed to understand that it was Matsushita (the guys behind Panasonic) who really manufactured the drives, which explains why they were so good. They weren’t quiet, but they were fast and often low power consuming.

DSC_8890This drive, however, is a behemoth I have never seen.

Dated 26th July 1995, this one identifies itself as a Quantum Prodrive 2100S, part of the Empire Series. It claims to be Made in the USA, where I have never ever heard of a drive being manufactured in.

From a quick look online, there is ample information about the drive. It is a 2Gb SCSI drive, and this one has active termination (type 2 PCB).

 

 

 

DSC_8891

The underside is a mass of Quantum ASICs, some with Texas Instrument logos. There are Analog devices chips as well as Hitachi and Hyundai RAM. There is a long row of configuration jumpers along the left side, but it seems clear that these drives were all microcontroller based intelligent drives, as most SCSI drives were, which made them more expensive.

So why did I call it a behemoth? Because it was a full height drive. I’ve never laid my hands on a full height Quantum drive before – probably because these were the things used in expensive servers which I never had the opportunity to play with.

DSC_8892

The drive is a tub with a flat top, as per usual.

DSC_8893

DSC_8895

Not much to say about the sides really.

DSC_8894

A big long flex traverses the front side, into the platter chamber. Unfortunately, while the drive spins and sounds alright, the drive is completely confused about its geometry and thinks its size is either 0 or 1 sectors. Maybe this is a sign of a loss of firmware, or a failure of the drive to calibrate to its existing servo marks on the media. Whatever it is, this one didn’t survive.

Conclusion

It’s surprising to think that some of the older hard drives are still perfectly healthy – they were the lucky ones indeed. But even then, this salvage provided some unique drives I hadn’t seen before, which was a treat in itself. It’s a shame the latter two were not functional – but I don’t think I’ll take them apart because they’re pretty special in themselves. Hope you enjoyed seeing some old tech!

About lui_gough

I'm a bit of a nut for electronics, computing, photography, radio, satellite and other technical hobbies. Click for more about me!
This entry was posted in Computing, Salvage, Tech Flashback and tagged , , . Bookmark the permalink.

11 Responses to Salvage: Conner CFS210A, Maxtor 7080AT, Seagate ST52520A, Quantum ProDrive 2100S Hard Drives

  1. sparcie says:

    I’ve seen that extra connector on the Conner driver before. That is for power as you suspect, used when the drive is installed in a laptop. My Amstrad ALT386 has a drive very much like that one installed. They may also turn up in other early laptops, but not having any others I couldn’t say for sure.

    That Seagate medalist drive looks like a higher capacity version of a drive I have! I thought its size was interesting to, which is part of the reason I have it.

    I tested all the old hard disks in my collection, with the exception of the MFM drives, and found that most of the older drives are still in working condition. There were a few that had failed, interestingly they were all quantum drives of various vintages.

    Perhaps it is just the larger tolerances and bigger more robust mechansisms that have helped these drives last. I’ve certainly had a much higher failure rate on more modern drives.

    Cheers
    Sparcie

    • lui_gough says:

      I agree, I have had comparatively lower levels of “luck” with modern drives, however, the failure rates are in no way alarming for the majority of models I’ve owned. It also boils down to early drives not having SMART diagnostics, which meant that it was a case of “listening” out for strange noises or seeing issues (ID CRC, Sector not found) in Scandisk/Chkdsk. Good old surface scan often was the way when you were “uncertain” of a drive’s health, but marginal drives sometimes managed to pass only to grow bad sectors later. At least now, drives have a chance of telling you when they “feel” sick, but its pre-failure predictive powers are still fairly weak when it comes to electrical “sudden death” failures (e.g. head coil or preamp failure).

      That being said, I did have some older Quantum Prodrive SCSI drives from Macs but they failed due to rubber stops turning into goo “sticking” the head assembly in place. There was a mass of Quantum Fireballs which had issues with the spindle motor controller which resulted in some heating up to the point of burning through their plastic packaging. Other than that, the spindle bearings often started to get noisy and eventually get too much friction to have a steady spin-rate and platter vibration level to actually work, especially in computers which were powered on 24/7 as in computer labs. I’ve seen many of those myself, but when they were being used, mine always seemed solid.

      Of course, it’s hard to gauge the reliability of the salvaged drives since many of these came from a University where it is likely they saw at least 8/5 to 24/7 operation for many years depending on what system they were installed in. Often drives you would own in your own machines would see less duty cycle, and that might help the spindle motor bearings.

      As for a theory why the newer drives are less reliable – they are extremely sensitive to tolerances. We’ve reached a point where thermal expansion of the platters, vibration from the spindle motor with a smooth fluid-film bearing, density of the air (which controls the head fly height), and vibration conducted through the chassis are critical parameters which can cause seek, track following, read and write errors. Apparently, the latest Seagate drives even have tiny head heaters to control the fly height by warming up the air a little to reduce the density and get closer to the platter. That being said, we have advanced dramatically in terms of encoding (PRML, over RLL and by adding large amounts of error correction code), servo technology (embedded, rather than external or whole-surface), and even in producing the media and components themselves in terms of surface smoothness, magnetic particle size, coercivity, and head gap size. Even the controllers which power the drives are now much more powerful and developed in the arena of Digital Signal Processing.

      All in all, I personally think that the drives were designed to be the “best” that could be achieved at the time in terms of data density for a given reliability (design life, hard errors per x bits read). It’s all a case of trying to bring bigger capacity and lower prices to the market, so inevitably this entails some level of risk. It just so happens that they might have been more conservative with their reliability estimates for components, that it manages to still survive. If memory serves me right, the hard error rate for hard disks has always hovered about 1 in 10^14 to 10^16 bits read, so if this was the design target, a smaller drive would have to be read many-many-many more times to see an error. But I think it’s just a “blanket” figure for use in failure estimation where drives are incorporated into systems, and in reality, we don’t see it happen as often.

      1 in 10^14 bits hard error would mean that you would see one read error every 11.368Tb. On a 4Tb drive, this means if you read the drive three times over, you can expect one error – doesn’t sound great, but RAID array guys sweat this every time they have to rebuild an array. But on a measily 80Mb drive, you would need to read it 149,012 times over to reach the same figure (and nobody has time for that!)

      Food for thought, I suppose.

      – Gough

      • sparcie says:

        The drives I have come from a different source, so possibly have had easier lives. I got the older SCSI ones all together for $5 at a computer fair many years ago, long after they were obsolete. Being SCSI and used in unix workstations (they appear to be from DEC and sun hardware) they may be have made to higher standards. The IDE drives were collected over a number of years working in IT support, after being wiped of course. They were used in home machines for the most part and had easier lives for sure.

        SMART whilst not perfect sure has saved me a few times. It has alerted me to marginal sectors that could frequently be fixed by removing the data for backup and writing zeros to the drive as a form of low level format. When this has failed you know there is some thing really wrong!

        I remember using scandisk/chkdsk as well before SMART became common. There was also the norton disk doctor which was reputed to be quite good.

        It seems amongst older drives, the likelyhood it’s still working depends on manufacturer. Probably because the techniques used in them were still quite different. I don’t know why the older quantum drives I have all failed, but it’s likely something didn’t age well compared to the others. Perhaps the same rubber problem or a different type of media?

        Most modern drives are fairly similar, but I have noticed that server grade hardware is noticably more reliable and feels different when handled. I’d say they are simply designed with reliability in mind over capacity, they are usually smaller capacity wise. Interestingly the 2.5 inch formfactor is being used by HP in their current SAS server drives. Although they are thicker than laptop drives.

        Temperature stability does seem to affect modern hardware quite a bit compared to older equipment. We used to see A LOT more repairs over the start of summer in all areas of failure. As you say the tolerances are smaller, so it’s no real surprise.

        Cheers
        Sparcie

  2. Found the old 210mb Conner CFS210A in the garage (my first hard drive!)

    Got some old data from 2003 or so that I never deleted (or recovered!) but unfortunately it isn’t detected by my bios, so I can’d do anything with it in linux. Worst case gotta get it professionally recovered/repaired in a full clean room – maybe after buying another CFS210A for spare parts. Well maybe in 5-10 years when after if I’ve saved a spare $5000 -_-

    Do you know what the gray-coloured port on the underside near the jumpers?

    • lui_gough says:

      The grey coloured port connects the head and voice coil motor inside the sealed clean-room grade platter chamber to the rear PCB – later drives pioneered pressure bed-of-nails type connectors or used flexible flat cable instead. You can see the pins from the drive chamber poking into the grey port from the underside.

      Does the drive spin? Does it make clicking noises? Does it keep clicking, or does it click a few times (calibration) and then stop clicking as you would expect from normal operation? Is it visible in the BIOS? If not, triple check power and data cables. I’m afraid there’s not much that can likely be done for these old drives – spare parts aren’t easy to come by, and knowledge of the drive’s inner secrets is probably quite scant even at professional facilities. Good luck with your quest!

      – Gough

  3. Just tried it again, and it detects. Unfortunately looks like we must have formatted the drive since it only has some random windows installation files. Still made backups using dd and ddrescue and will try linux “TestDisk” program to attempt to ‘undelete’ other partitions

  4. THIRD POST. It turns out it wasn’t formatted, it was Windows 98 DriveSpace compression. I couldn’t boot off the drive in VirtualBox fully, so I booted another win98 and mounted the drive with Start Menu->Run->DRVSPC and hooray my stuff is back!

    Sorry for triple post but I’m over joyed!

    • lui_gough says:

      Ah … good old Drivespace, successor to Doublespace (which was patent encumbered) full drive compression. At least the disk should have mounted as a FAT drive with a massive file which contains the compressed drive and possibly a text file to remind you that you’re dealing with the “host drive”. Nice to see it all worked out in the end.

      – Gough

  5. Zim says:

    I beg to differ on the maxtor reliability issue. They were the most un-reliable drives I ever touched. I had a good amount of maxtor drives and all failed miserably. One drive just ripped the head off the arm, it was rattling inside. After disassembling the drive, the head was lone in a corner.
    The reliability actually went up on later 8GB-10GB drives until they left the market.
    A curiosity about maxtor is that the HDD board will have a drawing silkscreened and sometimes a name too. You were unlucky to find one without both.
    I had a few rare drives on my hands in the past with brands like Kalok, IDC, some Conners, ALPS and the latest one I found was “Areal” branded, model A130 (130MB), made in japan. It’s a curious drive because I found it on a pocket 486 computer and it is half the height of a laptop drive and like 1.5CM shorter on the long side.
    It is interesting to see so many players that left the HDD market. And also the weird IC manufacturers, NCR? I saw them on Maxtor drives too, that one with an é on the quantum drive.
    Manufacturers, especially Seagate, will often ask for specially labeled parts for their drives, they coded each part differently, even if they were cheap components from the bin.

    • lui_gough says:

      Yeah, I had lots of problems with Maxtor with later drives, but earlier drives never seemed to faulter. Part of it is probably because some of them were mostly used in a “home-like” environment and probably low-usage and good temperature/humidity. That being said, I’m not unfamiliar with Maxtors that grow bad sectors like mad, and throughout history, every manufacturer has had bad series of drives – it comes down to the design, hardware and firmware.

      I’d be lucky to come across some of those brands today – Kalok and IDC I have heard of but never had a chance to handle. Areal is very interesting, Made in Japan would imply precision quality :). There were some experimental form factors – I know WD did make an oddball sized drive for laptops, with a matching odd interface with CD-ROM daisy chaining or something. Redhill Technology is always a good place to admire some of those. http://redhill.net.au/ig.html

      NCR I know mostly from POS systems, with NCR standing for National Cash Register. Surprisingly NCR was one of the earliest SCSI ASIC makers (http://smithsonianchips.si.edu/ncr/scsi-2.htm), which makes their appearance on storage devices not entirely without reason. I know there have been many ASICs which are adapted from existing technologies on drives – you will find Texas Instruments and STMicroelectronics (on Seagates), Lucent/Agere (on Quantum and Seagates drives), Marvell (WD, Samsung) amongst other well known semiconductor fabs. Many of them will be custom design, as hard drives sell in large volumes and it makes sense to do so, however, some of the Marvell ones are quite “generic” in that other vendors are known to use them with their own firmware (similar to how Marvell SSDs are implemented today).

      – Gough

  6. cheapie says:

    If you still have the Quantum, have you tried low-level formatting it? “sg_format –format –count=-1 /dev/whatever” should do the trick. Add “–wait” if it complains (this goes for all SCSI drives, actually – some don’t like IMMED).

Error: Comment is Missing!