PDA

View Full Version : Digital Systems Disk drive



Dwight Elvey
May 26th, 2018, 06:26 AM
Hi
It is not my sale but I saw a Digital Systems disk drive on ebay. The price is a little high but it is a dual 8 inch. This was one of the first, if not the first disk drives available for the IMSAI/ALTAIR. It is missing the buffer board that would go on the S100 ( not a complicated board as it has bus buffers and simple address decoding as I recall ). The buffer is something that would be easy to wire wrap.
The controller is built into the drive box. Rather than having code that runs on the CPU bus, it DMA's the sectors onto the S100 bus, so that it takes no RAM space with EPROMs. It DMA's a boot strap sector into RAM on reset. These were made by the same fellow that made the first disk controllers for G. Killdal. They were close friends.
The drives and controller were originally made for non-S100 systems but easily adapted to the S100 with the buffer card.
Dwight

Terry Kennedy
May 26th, 2018, 06:00 PM
Hi
It is not my sale but I saw a Digital Systems disk drive on ebay. The price is a little high but it is a dual 8 inch. This was one of the first, if not the first disk drives available for the IMSAI/ALTAIR.
If it says Digital Systems and not Digital Microsystems, it is one of the earliest units and probably single density. I believe the later units were double density, but those bits of my brain cells don't have parity.


It is missing the buffer board that would go on the S100 ( not a complicated board as it has bus buffers and simple address decoding as I recall ). The buffer is something that would be easy to wire wrap.
There were some persnickety bits to that hardware. I had 2 of them and the S-100 interface boards didn't interchange between them. That may have been due to the systems the boards were in, though.


The controller is built into the drive box. Rather than having code that runs on the CPU bus, it DMA's the sectors onto the S100 bus, so that it takes no RAM space with EPROMs. It DMA's a boot strap sector into RAM on reset. These were made by the same fellow that made the first disk controllers for G. Killdal. They were close friends.
I went through a number of different memory boards before I found some that would work reliably with that interface card.

Dwight Elvey
May 26th, 2018, 06:49 PM
If it says Digital Systems and not Digital Microsystems, it is one of the earliest units and probably single density. I believe the later units were double density, but those bits of my brain cells don't have parity.


There were some persnickety bits to that hardware. I had 2 of them and the S-100 interface boards didn't interchange between them. That may have been due to the systems the boards were in, though.


I went through a number of different memory boards before I found some that would work reliably with that interface card.

That is interesting about being fussy about what RAM. I guess I've been lucky. The low level software I wrote always puts the buffer in a fixed location and then the processor moves the data to the desired location. I think I use the same board as the boot sector loads to. I'm using 3 different RAM boards. I believe your right about them being single sided, at least mine is single sided. I recall reading about the difference in the buffer cards. I think the main difference was that they split the cable into two cables for one of the versions.
Another thing was that not all the controllers had the ROM to do the formatting. I believe it was assumed that one would buy preformatted disk. I also recall, there was a slight timing difference between those formatted on these compared to what was later a standard. I don't recall which way these were a problem. I've never taken one of my disk to another machine. My setup still works with CPM2.2, I put on it myself.
Dwight

Terry Kennedy
May 27th, 2018, 05:20 PM
Another thing was that not all the controllers had the ROM to do the formatting. I believe it was assumed that one would buy preformatted disk. I also recall, there was a slight timing difference between those formatted on these compared to what was later a standard. I don't recall which way these were a problem.
The canonical standard was the IBM 33FD. Earlier IBM media was either read-only (factory written once) or more primitive (no protective jacket). In addition to the various discrete logic formatter implementations, there was the Western Digital 1771 and later derivitaives, and the NEC uPD765 and later derivatives. The WD chips "formatted" the diskette by simply recording a track-length buffer provided by the user (so you had to create all of the bits yourself, but had full control over what was written to disk). The NEC part took a description of what you wanted and built the track image itself, which was a lot more convenient but meant that you didn't have the level of control that you got with the WD part. Early double density (MFM) implementations were even more bizarre - both the DEC RX02 and the drive in the Datapoint 15xx systems wrote the sector headers in FM, but the data field in MFM. You could create Datapoint 1550 media with a WD chip (as long as it was an MFM-capable one). I never tried to do that for RX02 media.

The IBM drive was slow and noisy - an IMPL on a 370 was a symphony of noise, as the heads unloaded after each operation, so you got a random series of "brrrrzzzzt wunka wunka wunka" sounds out of it as it loaded the microcode.

The 5.25" floppy made the complicated 8" world look normal. There were 96TPI drives and 100TPI drives in every possible combination of single/double density, single/double-sided, soft/hard sectors, etc. And then you had systems like the Durango F85 which used a pair of Micropolis 100TPI drives with a unique controller that did GCR, getting 1MB on a single 5.25" diskette. [The maximum "legal" capacity of a 8" DSDD drive was around 1.2MB (77 tracks of 8 1KB sectors * 2 sides). Some people managed to squeeze 9 sectors in there, but it was at the very edge of what was possible.