Wednesday 19 October 2016

ICL One Per Desk Microdrive Repair

This is an old post, preserved for reference.
The products and services mentioned within are no longer available.

The previous blog post on the repair of a BT Merlin Tonto (ICL One Per Desk) covered the cleanup and monitor / power supply.
Once it was up and running, it was clear that neither of the two microdrives were working. When attempting to access either drive there was no response, no motor noise, no activity lights.
The microdrive was a cassette tape loop based storage device, with 5m of 1.9mm tape spooled in a small cartridge, like an 8 track cartridge. I don't know why I think that will help, if you don't know what a microdrive is, you probably don't know what an 8 track cartridge is either.
There is no spool, the tape is pulled around in a big loop by a capstan from the side, the tape head sticks in from the top. Those foam pads tend to wear out, so need to be replaced to keep the tape pressed against the head.
These were introduced for the ZX Spectrum, as a replacement for standard audio cassette storage and a much cheaper alternative to disk based solutions. You could chain up to 8 of these drives together to provide massive storage capability for the Spectrum. (or why not just buy 8 cartridges and swap them over in the one drive?)
It may look like the microdrive just plugs into the side of the Spectrum, but there is in fact a larger interface unit the 'Interface 1' which the Spectrum is sitting on.This contains the drive circuits and ROM patches for the additional commands necessary to use Microdrive storage.
This provided the Spectrum with 'up to' 100K per cartridge, although in reality this was a lot less. As the cartridges aged, the tape would stretch, so you could format it again and the capacity would increase slightly (probably in proportion to a decrease in reliability).
The microdrives were again used in the Sinclair QL, with two of them intregrated into the QLs case and intended as the QLs main storage device. The BT Merlin Tonto was based on parts of the QL and included two microdrives as well.
One of the articles I have read say that ICL re-engineered these microdrives to be better than the Sinclair versions, I'm not sure about that. Here is a Spectrum microdrive.
Theses are the QL ones.
And these are the ICL ones.
The plastic chassis looks identical, same motors, same switches,  the PCB was laid out differently, but basically the same. Comparing one of those to the Spectrum one, I can't see much difference. Sinclair on the left, ICL on the right.
On the rear, again I'm struggling to see any mayor mechanical differences.
The main electronic part on these is, true to form for Sinclair, a Ferranti ULA, here in the Sincalir microdrive. You can see the two edge connectors on the rear PCB, one on each side.
These are chained in series. Most of the pins pass through from one microdrive to another. The exception is the data signal. This goes into the first device, is processed and then passed onto the next device. Presumably removing anything addressed to it, and passing on the rest.
The same part 2G007E is used on the ICL version, and like the Sinclair versions they are chained in series. When I checked, there was data going into the first chip, but it was doing nothing with it. It was also not passing on any data to the second drive, so it looks like a faulty ULA.
This could mean the second drive is actually fine, and just wasn't getting the data passed to it from the first drive. To test this, I rewired the connectors so the first drive was disconnected, and the second was connected to the main board.
Success, that drive was fine, One cartridge formatted with 204 x 512 byte blocks (102K). It should be noted that although they appear mechanically and electronically very similar, they use a different logical format, so you can't actually interchange date between the three system.
That would confirm data wasn't passing through the first drives ULA, which does look to be faulty. I was able to get a new old stock ULA and replace it. Not trivial as the 6 pads you can see below the socket are the back of the tape head which has to be desoldered before the board can be accessed to remove the ULA.
I fitted a socket and the new ULA and inserted the module back into the chain. I was going to place this as the second drive, but I found that although the modules looked identical, one PCB was slightly different, the input connector is positioned differently, which meant the linking cable wouldn't reach. (UPDATE: I've just noticed the left hand board is part number PWB 5400054, and the right is PWB 5400055, so yes they are different.)
With both modules reconnected and the new ULA in place, time to test again.
All looking good, both drives are formatting and reading back, and it has successfully copied from one drive to another.
All finished and ready to go back into service.