Early IBM "Small" Hard disk drives

[TV_Madness]

Hi All

I have a couple of early, IBM hard disk drives that I would like some help with. These particular drives are of the 8 inch form factor variety and would have been installed in systems such as the System/34 and System/36 minicomputers, however they are gathering dust and I'd like to do something with them.

Both of which are model 62PC.

I should imagine that using these drives to actually store useful data is probably out of the question as I don't actually own the disk controller attachment that would have been used with these, however how feasible would it be to lets say, get the drive to perform track seek operations and such like?

The spindle motor runs at 110VAC, however I am not sure what the power requirements are for the logic circuits on the drive.

If anybody has any knowledge of these devices, I'd be interested to hear from them.

Kind Regards

Matt

[woodchips]

Try the vintage computer forum? At vintage-computer.com.

These might be the older SMD interface drives, two 50 pin connectors? Can't really remember now. Also being 110V AC might mean they are 60Hz.

[TV_Madness]

The AC motor is rated 47 - 62 Hz, so this shouldn't be too hard to drive. These drives use a bus and tag data / control interface. They're not SMD. I think this was proprietary to IBM.

[Dual Standard]

Matt, It's been many years since I last laid hands on an IBM 62PC (Piccolo) drive so my memory is rather patchy. As well as System 34 & 36 the 62PC (and its successor the double density 62SW) were extensively used in the 8100 series, that is 8130, 8140, 8150 processors and 8101 external disk drive. The 8101 could hold two 62PCs and was connected to the processor via bus & tag. There was also the 3310 which looked just like a 8101 but was for attaching to System 370 machines (usually 4331). The disks themselves were quite reliable but didn't like being powered off/on, we usually switched the 8101s to 'local' when working on a system so that they could be left powered on while the rest of the system was shut down. I am pretty certain that the 62PC interface was proprietary so it is likely that publicly available information will be limited to what was contained in the Maintenance Information manuals if you can find them. The 8100 series manuals were quite detailed but the 3310 manual may be better. Sorry I have no longer have any manuals or other documentation for these machines. If anything more exciting comes to mind I'll let you know. Mike

[TV_Madness]

Hi Mike.

Many thanks for the insight and I didn't know that these drives were used in storage facilities designed for the S/370. You learn something new every day!! One of these units is dated 1979 whilst the other is much later, being dated 1987 (am I right in saying that these may have been manufactured right up to the early 1990s?)

I have actually managed to find a datasheet on the web for these devices (over 80 pages) however there is very little in the way of logic on the drive itself. It looks as though the minimal electronics that are actually present on the drive chassis are really only focussed on servo control, voice coil actuator drive and read / write pre-amps / drivers. There seems to be some head / track address logic on the drive, however there is no mention of how this is addressed directly at the drive interface. There is also no mention of the power requirements for the logic circuits.

The way these drives are physically connected looks a lot like the old ESDI format, with a dedicated data cable to each drive and then a tag / control bus that is chained between drives (or with a terminator card installed on the end of the cable if only using a single drive).

The majority of the drive control looks to be carried out by the disk controller attachment itself (which in reality, I was assuming all along given their age). There is plenty of documentation in the manual on the various data flows to and from the drive controller, however as I don't have this, I feel that this project could be very difficult.

Would be lovely to see these things in operation, however I could be biting off more than I can chew here.

[arjoll]

Originally Posted by TV_Madness

One of these units is dated 1979 whilst the other is much later, being dated 1987 (am I right in saying that these may have been manufactured right up to the early 1990s?)

I don't know anything specific, but if these were used on the 36 then I'd guess that's probably the case - when I started at Ernst & Young in 1990 as a trainee the time and cost and debtors systems were still run on the "36" somewhere up-country, with a couple of terminals in our office for access.

It would have been when I was away studying in 1992 or 1993 that we migrated to ISIS-CA, a client/server system based on CA-Ingres running Digital Ultrix workstations in each branch, syncing up to a central database in Auckland. The client ran under Windows for Workgroups 3.11 with the Win32 extensions - I was one of the rebel regional network admins who started running it on Windows 95 well before National IM gave formal permission

[TV_Madness]

Attached are a couple of images of one of these drives for anybody who hasn't seen one of these in the flesh.

Image 1 - gives you an idea of the scale of these things! Very heavy devices also.

Image 2 - The interior of the disk enclosure. Look at the size of that voice coil actuator, bearing in mind these are 8 inch platters! There is a separate external actuator transit lock also on these that swings a little paddle round up against the lower part of the rotary portion of the actuator. The very lower surface of the disk pack is the servo surface, whilst a second set of servo data is embedded between user accessible sectors.

Image 3 - Card cage interior. Working from the top down (and with a little help from the manual) the cards appear to be as follows;

1. Servo track processing and actuator control.

2. Servo fine control / Write clock generator.

3. Servo coarse control, track sector identity, byte counter and write clock oscillator control.

4. Track address logic and servo safety control.

5. Read / write encoders, clocks, drive safety interlock circuits and essentially all of the core data I/O processing of data flowing to and from the drive.

The clear Perspex section on the top houses 4 TO3 package transistors, These form the voice coil actuator drive final output.

Image 5 - The backside of the card cage after unlocking it from the chassis. The top row of interconnects on the board look as though they are all power related as they terminate crudely at a cut in the cable loom just to the left of the picture.

The centre connector in the top row is for the drive spindle motor brake whilst the lower row all go off into the enclosure that contains the power output stage for the voice coil actuator. The white colour leads go off into the disk enclosure and provide drive to the voice coil. I'm not sure what the second set of white leads are. These terminate at a little device, bolted to the side of the actuator magnet assembly and looks to resemble a Piezo electric transducer.

The orange Mylar flex behind the cage carries all of the I/O to and from the R/W heads in the disk enclosure whilst the two white flex cables are the control and data bus cables respectively. This drive has obviously spent its life at the end of a chain of two drives in whatever it came out of, as there is only one control cable installed. Where the chained control bus cable would have left the drive on its way to a second unit, there is a terminator card installed.

Cheers

[Nicklyons2]

The old Cox video framestores used a sort of 'cake glass' cassette of discs which was vaguely similar to this - but on its side. If you desired another range of images you had to swap the 'cake glass' over. Not as quick as you may imagine because a substantial time had to elapse for the drive to slow and stop before you could swap 'cakes' and then the replacement had to run up to speed. I can't remember who made the devices (in any case they were probably badged Cox), but they did look similar.

[Dual Standard]

Hi Matt,

Your photos bring back the memories.

The top row of interconnects are power, there are multiple ground & +5 volt feeds, a -5 and a +24v, there may also be a +12. The power supply was of course linear, no switch mode back then, with a ferro resonant mains transformer and a separate bucking transformer to adjust the 5 volt feed which was quite criitical.

The arrangemnent of the ground, +5 and -5 connections were pretty standard across all IBM machines at the time but I cannot remember now what that was!

As well as the manual (transit) head lock there was also a manual lock for the motor pivot using toothed segment, on the 8100s these both had a linkage to a bowden cable that terminated in a large handle so the customer could unlock the disks on installation without needing any tools (customer install).
I don't think the System 3x had these (engineer install) so if the cranks and toothed segments are missing from your drives I suspect that they are off a System 36, 38 or Series 1.

There should be a cover over the card cage that holds the cards in place and ensures they get cooled by the fan.

A quick Google search suggests that the 62PC was manufactured until Feb 1990 which surprised me as 8100 production ended arround 1986 when the 9370 was introduced though we maintained them for several years after that. I think the System 36 etc carried on longer until the AS400 arrived

I hope this helps fill in the gaps.

Mike

[Nickthedentist]

Incredible pictures.

For someone like me who's never had anything to do with professional computers from this era, the unit looks like something out of a washing machine at first glance.

Thanks for posting them.

Nick.

[Colin Ames]

I agree!

Colin

[G6Tanuki]

Originally Posted by Nickthedentist

For someone like me who's never had anything to do with professional computers from this era, the unit looks like something out of a washing machine at first glance.

Yes, back in the last millennium DASD [Direct Access Storage Devices, a.k.a. hard-drives] were indeed mechanically-huge, with spectacular power demand and distinctly-temperamental environmental requirements. And achieved a storage-density which is truly pitiful by modern standards.

In a previous life I curated an IBM 4381-12 'minicomputer' that had an entire floor of the datacentre occupied with '3380' DASD.

http://www-03.ibm.com/ibm/history/ex...age_3380b.html

We're talking a row of 7-foot-tall freezer-sized cabinets to store a mere 10Gbytes.

These days you can fit 5x that amount of data on a USB memory-stick you can entirely forget is in the pocket of your jeans when you put them in the wash.

[jon_the_id]

These days you can fit 5x that amount of data on a USB memory-stick you can entirely forget is in the pocket of your jeans when you put them in the wash.

Make that 100 x as much data, you can now get 1 Terabyte USB sticks.

[flyingtech55]

Presumably in those days you repaired these drives to component level and didn't simply bin them when they failed? If this is the case, did the field techs diagnose and exchange the failed PCB for a service exchange board and send the failed board back to a centralised service depot?

TimR

[G6Tanuki]

On the 33x0-series drives at least, the whole "Head/Disk Assembly" [disk-platter stack/bearing/spindle, motor, brake, heads and head-positioner] was sealed, and serviced as a single Field Replaceable Unit [FRU] - they were returned to the factory and sometimes rebuilt, sometimes not depending on the nature of the failure.

Controller and interface boards were FRUs. Likewise the power-supplies. There was a special IBM 'hand-operated fork-lift' trolley thing to remove/lower a failed HDA and lift the new one into place in the cabinet (they were heavy!).

[Dual Standard]

Tim,
The basic construction of these drives was very similar, but smaller and much higher capacity that their predecessor the 62GV (Gulliver), so the repair methodology was much the same. The drives were repaired to the FRU (Field Replaceable Unit) level, so for example, the disk enclosure (everything inside the smoked plastic box in photo 1) would be replaced as a unit, the motor as another, each card in the cage another etc.
The logic rarely failed, I don't think I ever replaced a motor, but did change many disk enclosures and quite a few noisy fans. These disk enclosures were quite light and could easily be changed single handed so there was no need for the lift/trolly device that was needed for the 3370/75/80 series. The replaced FRUs would be returned to manufacturing for failure analysis and/or possible repair.

I also remember now that there was another variant of this (62PC) disk that had fixed heads on one platter, it was easily identifiable by an extra ribbon cable to the logic cage, the idea was to give speedier access to a limited amount of data. I only ever saw them used as the 1st disk in a system. Care was needed when changing these as in the wee small hours it was quite easy to loose concentration and fit the non fixed head variety by mistake. This did not go down too well with the customer!

Mike

[TV_Madness]

Apologies for the belated response.

I haven't been feeling to great recently and haven't really had an opportunity to get close to a computer.

Yes indeed some of those early DASD facilities were colossal!

Certainly the IBM 3330 Storage facility was a behemoth. I actually have a maintenance manual for one of these (codenamed "Merlin" during development) which goes into incredible detail. No nut, bolt or scope trace is left unturned!

The 3330 I do believe was the first example of a hard disk drive to utilise a voice coil actuator, operating under track follower servo control.

Certainly the IBM 1130 processor did use a voice coil in its cartridge hard disk drive, however this used some form of mechanical detent system rather than a track follower.

The 3330 had a capacity of 100 MB per spindle but the model II doubled this, presumably through tighter tolerances and improved electronics. Incredible really when you think that this was introduced back in 1970.

The 2314 DASD facility looked quite similar to the 3330 but used Hydraulic actuators which probably explains its 29 MB per spindle capacity. Still incredibly impressive considering it was announced in 1965!

Very interesting that there was a variant of the 62PC that used fixed heads on one platter. Presumably this was to aid in retrieval of "hot" blocks of data or information that was frequently required.

I really do have an immense amount of respect for this stuff. Some of the innovations IBM were making back in the 50's/60's were staggering and really did contribute to where we are today. The 8 bit byte, instruction pipelining, the hard disk and even hardware virtualisation, all of which were pioneered by IBM.

Incredible... Bit of an IBM nut really

[G6Tanuki]

Originally Posted by TV_Madness

Very interesting that there was a variant of the 62PC that used fixed heads on one platter. Presumably this was to aid in retrieval of "hot" blocks of data or information that was frequently required.

That was certainly the case with the 3380 DASD controllers; they would autonomously look at the seek requests they were dealing with and unobtrusively shuffle the data about on their downstream discs so as to make it 'first seek' available for future requests.

There was an obscure IBM RPQ[1] that you could specify which gave you special microcoded instructions in CICS[2] to better handle such stuff in transactional applications.

[1] Request for Price Quotation - IBM's way of billing you for special features. http://en.wikipedia.org/wiki/Request_price_quotation

[2] http://en.wikipedia.org/wiki/CICS

[TV_Madness]

That really is something else. The specs of the 3380 for the time were once again incredible.

[yestertech]

Hard to believe all the equipment mentioned above, on which I worked many a happy hour is now reduced to the status of museum pieces.
All that knowledge , completely useless today - that's progress I guess.
This phenomenon has now happened to me in two separate careers .... Maybe it's time I retired !!


Andy