wtm@neoucom.UUCP (Bill Mayhew) (03/03/88)
Hi, A while back I mentioned that I was planning to look into the feasibility of using one of the low cost floppytape drives as a backup device for the Unix PC. Basically a floppytape drive attaches to an interface cable that would normally connect to a standard 5-1/4 inch disk drive. One could replace the original 5-1/4 Teac 55B drive of the 3b1 with a tape drive. What makes this intersting is that the cost of the tape drives is as low as $350. I talked to Walt Mazeur at Archive Corp. which makes drives. He is in their data systems division. He has been involved in writing software drives for various systems. One limitation he pointed out is that with current generation floppytape drives are not capable of operating in an interrupt driven mode. Basically, the CPU has to busy wait for the drive to finish a given operation. In 4 to 6 months, drives will be available that can generate interrupts when a tape operation has completed. Capacities up to 80 megs will be available that can still fit into a 1/2 height 5-1/4 cutout. Apparently Archive has been working with SCO on writing drivers for Xenix. He estimated that about 3 person-months of time would be needed write a fully debugged Unix driver. It looks like the current generation of floppytape drives might be workable, if you went to single-user run level so that you could be sure that nothing was going to clobber your timing. That sounds like a good idea anyway, as you wouldn't want some joker writing to a file that was being backed up. In short, the floppytape units are quite a bit different from a floppy drive to the extent that software that a floppytape appears invisible to software that is looking for a floppy disk -- so that an tape in the drive doesn't accidentally get munged by errant floppy writes. The electrical signals are similar enough that a standard floppy drive controller chip can be conned into generating them. Basically the trick to waking up the tape is to send it a sequnce of pulses on the head select and step leads. Exactly what the commands are, I don't have at hand yet. Any comments from netland are invited. --Bill
richard@islenet.UUCP (Richard Foulk) (03/07/88)
> software drives for various systems. One limitation he pointed out > is that with current generation floppytape drives are not capable > of operating in an interrupt driven mode. Basically, the CPU has > to busy wait for the drive to finish a given operation. In 4 to 6 > months, drives will be available that can generate interrupts when > a tape operation has completed. Capacities up to 80 megs will be > available that can still fit into a 1/2 height 5-1/4 cutout. Judging by the way that it works I'd guess that the floppy on the 3b1 is non-interruptible too. I have a vague recollection of reading some data several years ago on a tape drive similar to the one you mention that actually allowed sector addressing via the floppy interface. That is, it could radomly access sectors. From that I got the impression that such a drive might work on the 3b1 without writing a new device driver. The rest of the system could get some time in -- in between sectors, just like with the floppy. But then maybe I'm dreaming. > them. Basically the trick to waking up the tape is to send it a > sequnce of pulses on the head select and step leads. Exactly what > the commands are, I don't have at hand yet. That sounds a little contrary to my vision of things. At least requiring a new driver. Oh well. Maybe it's time to disassemble the floppy driver. -- Richard Foulk ...{vortex,ihnp4}!islenet!richard Honolulu, Hawaii