One fun aspect of 1970s-era hard disk drives is that they are big, clunky and are fairly easy to repair without the need for a clean room. A less fun aspect is that they are 1970s-era HDDs and thus old and often broken. While repairing a CDC 10 MB HDD for the upcoming VCF East event, the folks over at [Usagi Electric], this led to quite a few struggles, even after a replacement 14″ platter was found to replace the crashed platter with.
These CDC HDDs are referred to as Hawk drives, and they make the associated 8-bit Centurion TTL logic-based computers so much faster and easier to work with (for a 1970s system, of course). Despite the large size of the components involved and the simple, all through-hole nature of the PCBs, issues that cropped up ranged from corroded DIP switches, to head alignment sensors, a defective analog board and ultimately a reported bad read-write head.
Frustratingly, even after getting the platters to spin up and everything moving as intended, it would seem that the remaining problem is that of possibly bad read-write heads, as in plural. Whether it’s due to age, previous head crashes onto platters, or something else, assembling a working Hawk drive turned out to be somewhat more complicated than hoped.
We definitely hope that the bunnies can get a working Hawk together, as working 1970s HDDs like these are become pretty rare.
I and a friend did restore a couple of DGC Diablo disks a long time ago together with a Nova 4 plus a couple of boards with core-memory. We had extreme problems with the first disk, it failed to read/write properly and we worked on it for a week until we discovered that the signal for track-start was based on an optical signal where a lamp shined light on a notch in the spindle and an optical sensor detected the reflected light. It turned out that the optical sensor had been adjusted too close so it got several reflections due to the shape of the notch which confused the timing for track-start that then wobbled back and worth a couple of milliseconds for every rotation.
I remember in late 70s era computer and electronic magazine, there were ads for replacement r/w heads and spare platters. Back then an average Joe could probably swap the heads and disks easy. Today only the professional clean lab and trained pro can even consider opening modern hard drive.
Anyone can easily open a modern hard drive at home without specialised equipment.
Of course, getting it to work again afterwards is another story…
just get a glove box and some good filters.
About 15 years ago I opened and closed them and still had them work. Work quickly in a fairly clean room, place some plexiglass over it to stop dust getting in if you’ve not got your fingers in it.
Can’t remember now why we thought we needed to open them… maybe we didn’t need to… I do remember we did firmware updates on some bricked drives though.
I think the hardest part of getting them back together and working reliably, is that you need to give all the screws the exact right torque. And the torque values are often hard to find, in my (to be honest: very limited) experience.
I had a old 100 meg scsi drive where the heads would get stuck on the perished rubber stopper, so I took the lid off of it and nudged the heads during spinup. I needed that disk to get a system up and running as well as getting the software off of it… it ran this way off and on for over a week before it started getting “clacky”
Overstating it a bit, don’t you think? I’ve “considered” opening a modern hard drive a few times without a clean lab. Admittedly, out of desperation and cheapness, and only long enough to get a few files off. In fact, over a 30 year career, I’ve only had to go to a professional shop once — most data I work with is either backed up or generally not worth the cost of recovery.
Is it worth the effort
10 minutes with a security screwdriver? Sure, if you get lucky. Even if it all goes to heck, 10 minutes to get some nice magnets is probably worth it.
Not likely for the average Joe in the disc drives of the 70’s! These were washing machine sizes, and later in the 70’s were drawers in a 19″ rack.
The Field Engineering course Singer & ICL taught in the 70’s took a week just for the disc drive. Of course there was a bit more than just changing the heads.
I posted below, the cost of a special CE disk pack used to align the heads was rare, cost ~$25K, and you needed a Tektronics 465 CRO or similar and some electronics.
You needed to be careful by locking the head carriage to prevent head retract due to faults or power cut or dip after the heads loaded. Otherwise, a head retract was likely to result in lost fingers!!!
Not reall.
Nobody at these times would let you mess with his DATA or HDD costing thousands of 1970 US$.
For the read errors…
I presume the heads have been cleaned with >= 97% proof Isopropyl alcohol and cotton buds – yes we used cotton buds but just make sure there are no strands left (easily seen).
The platters are clean – but you cannot clean these like you do for the heads.
If the platters have been replaced, then you need to write them with a format routine. Normal writes will not necessarily work.
Over time, the heads migrate forwards slightly. Maintenance used to be done every 3 months to re-align the heads. This probably was not done in the latter stages of its’ useful life. Thus, read errors would likely be common. If you are trying to re-read the sectors, try this…
Read the disk, saving the data out to another medium, and noting the bad sectors.
Now back-off the head screws a half a turn, loosen the head screw clamps, tap the heads back (carefully!), and re-tighten the screw clamps.
Re-read again.
You may need to repeat this operation a few times.
The hope here is that you can read those sectors with errors correctly as you slowly adjust the heads back in. Of course, some good sectors may go out of alignment as you perform this process. Hopefully this will render a full set of good sectors between the number of attempts.
Now you need to reformat the drive.
Note I have assumed you do not have access to a CE disc pack – An expensive disc pack written with a special wavey track that is used to generate cats eyes on a CRO to do proper head alignments. The cost of these were well in excess of the whole drive. IIRC around $25K in the mid 70’s. And if you write to it it’s a write-off – no second chances. So a write disable link would be fitted (to the drives’ electronics) when using these discs.
“I presume the heads have been cleaned with >= 97% proof Isopropyl alcohol and cotton buds – yes we used cotton buds but just make sure there are no strands left (easily seen).”
Foam swabs. Used when working on VHS heads.
reformat the drive means reformat the disc pack on the drive.
Without proper head alignment, it is probably unlikely that discs can be swapped successfully between drives.
For years I’ve kept one of the platters from a head crash on one of these disks – it’s great to show kids how a hard disk works, and when it doesn’t work.. The head crash area is a pretty good story of what could go wrong..
It died when a large bushfire overwhelmed the air system in the data center I was working, smoke and particles got in the room, then in the disk drive, and that was it…
And if you forget to park.exe you hard drive before turn off your computer… HDD Kaput!
This type of drives has a huge capacitor that feeds through a control circuit directly into the linear motor the heads are mounted on and if the power goes out for any reason that isn’t a controlled shutdown that charge gets dumped into the motor and the heads retract *fast* with a very loud bang when they hit the stops, this can result in the heads getting misaligned but you avoid cratering the platters though.
Brings back memories. I worked at DG and those pizza oven drives were ubiquitous. Well done to get that system up and running. I knew the guy who designed the NOVA 4 CPU…Charlie Retter…a very bright PhD in Math who went on to a career in the Army. I worked on a board that re-used the bit-slice and microcode of the NOVA 4 as a serial data stream processor, so I had a few conversations with him about the microcode (I was changing some of the instructions). Great guy, he passed away a few years ago.
My first job in the industry was a summer job at DEC Westfield, where I was a production line test tech for the RK06 drive — 13+ Megabytes on two platters (3 surfaces plus a servo surface). It was the successor to the famous RK05 = Diablo disk
Super cool! I still have a tape from DG. It is microcode for the MV20000 I think.
We used quite a few DEC RP06 178 MB and RM03 drives (67 MB), both were multi-platter packs… the memories.
For a good overview of contemporary disk platter and head changing along with forensics, check out YouTube videos by Scott Moulton.
When I worked for Burroughs SOP was to wipe the platters over with a lint-free copiously moistened with isopropyl, then ditto with Freon which evaporated without leaving a residue.
At the top-left of http://pascal.hansotten.com/wp-content/uploads/2016/01/cropped-wirth-header1.jpg you can just about see what happens to a 14″ platter where the heads crash and fail to retract. The amount of swarf generated can be substantial…
I used to work with Data General MV6000 systems. I might have a disk pack or two available for parts if you are still looking for platters. I am not sure of the exact dimensions of the platters, but if you are interested I could measure them.
In high school, our Electronics Department received a donated mini-computer that had one of these beasts in it. The platter became our bathroom pass. I was given the wonderful Lambda power supply from it… the thing lasted almost forty years before only recently giving up the ghost.
OMG! I remember working on something similar to what is shown. The ones I worked on were 80meg disk pacs (5 platter). The card cage shown is the exact cage. I bet there’s an SGV card in it and using a volt meter on (I can’t remember the pin#) to ground and tuning each head to 5v would align the head.
I still have a bunch of platters from crashed pacs.
Regarding the issue of the read-write heads, many YT commenters made the suggestion of testing the read-write head wires separately, since they can fail due to constant flexing.
At NCR we had drives like these I believe made by CDC in grocery stores. They were called the NCR 6560 and 6566. Regular external filter changes were a necessity.The 5meg drive held the scanning file. In one configuration there were 2 NCR 605 Processors each with their own drives in a 6 foot tall cabinet. The main computer read and wrote simultaneously sales figures as well as items scanned to both drives (known as P2P). If the main processor went down a rotary switch was thrown, the secondary processor was booted up via a TEAC cassette into master mode and within 5 minutes the store was back to scanning.
When I was at NCR I had one grocery store that only a single processor/hard drive setup (not a dual fail over system called P2P). They had a head crash late one night because their manager never changed the air filter regularly. They left the drive on grinding and squealing all night while the cashiers hand keyed prices. I came in, took the drive apart and could see myself in the platter and I knew it was going to be a long day.The head cashier walked in and said “Can I still get my totals ?” I held up my oxide covered hand and said “Here’s your totals ! You should have turned off the drive when you heard the noise !” (Actually their basic sales totals were in ram memory). It took 8 hours of scrubbing the drive chamber clean, replacing the heads a new platter and reloading the scanning file from the TEAC digital cassette tapes before they could scan again. After a lesson on filter replacement regularity the drive ran flawlessly 4 more years until the store upgraded. The new system was the NCR 2126 with the scanning file in two of the lane registers and held in Bubble Memory.
About removable hard disk cartridges: IBM 2315 cartridge design was used not only for IBM devices, but the same mechanical design was adopted by quite a few other companies: DEC (RK05), Xerox Alto Diablo, HP, Wang…
I have some 2315 like cartridges, Scotch 3M and Nashua one, but i don’t where they come from, and there’s no model number on them. I’ve read that at least some of them (if not all?) are hard sectored, with slots on the hub’s rim. Beside the number of hard sectors, is there any other difference than can make a cartridge incompatible between different systems?
This is good to see, I may be able to help you out. I’m not sure how we can connect without giving my email to everyone. I’ve been cleaning out my storage lately. I just ran across a CDC disk exerciser. I know I also have an alignment pack and a service manual somewhere. I may have heads and spare boards buried somewhere. From the picture it looks like yours is the 50 pin ribbon cable interface. There were several types of interface.
I think you should try to contact [Usagi Electric] directly elsewhere, since it’s not him that wrote this article here on Hackaday. You can try to send him private messages through its hackaday.io account: https://hackaday.io/usagi.electric
It would be great if you could help him, and i’m looking forward to see new videos about it!
To bad i never hear of so cool and fascinating projects here in France, i would really enjoy to be a part of it…
Mr. Lewis, If you find your old CDC Hawk item please contact us because we are needing parts to keep the system running and help the other two Centurion systems recently found.
My company in the UK had a large number of customer systems which had the Pertek version of the Hawk drives with two platters. I used to be a field engineer looking after these. The customer was responsible for changing the filters and of course some forgot or even had builders in creating dust. They were often in a cupboard and head crashes we not noticed for a long time leading to drives full of aluminium swarf. The worst example was at Ford Motor Company in Dagenham. We took it back to the workshop and it took them many hours to clean it.
Unfortunately you can’t get good reliable performance without a CE pack and an oscilloscope as someone described. Also unless in a ‘clean room’ you need to change the filters which might be difficult now.
We would always change the heads if there was a crash (but the customer paid for it).
They were connected to DEC PDP11/73 systems with RSX operating system. In the DEC training school we had to as a test write a machine code program to perform an incremental seek on the drive. Tough to do but great fun to watch it work :-)