17

I've always been fascinated by the rarity and elitism of such devices.

However, from a practical standpoint, I don't understand why such low transfer rate device ended up in a bus commonly used for its performance.

Best candidates I can think of were Akai audio samplers that implemented the bus and also had a floppy drive.

For what usage in computers were these devices for?

Note: I am talking about 1440 KiB floppies, not Iomega Zip drives; for those, the benefit was obviously evident.

user3840170
  • 23,072
  • 4
  • 91
  • 150
Eric Cartman
  • 6,760
  • 5
  • 35
  • 59
  • 1
    I could be wrong, but external floppy drives for the Amiga are generally SCSI. That's because that same port was intended for anything else to be connected like CD ROMs or hard drives. – Omar and Lorraine Jul 30 '17 at 09:05
  • 3
    External Amiga floppy drives use a (modified) Shugart bus, not very special except for the 23(!) pin connector. SCSI was generally optional except for the 2500, 3000/T and 4000T. – Zac67 Jul 30 '17 at 10:30
  • 2
    An advantage of SCSI over IDE is that SCSI requires less CPU. Maybe the same applies to floppy drives. Back in the early Windows days, while something was accessing the floppy drive, the whole system froze until it was done. I think Amigas didn't have this problem. – snips-n-snails Aug 01 '17 at 18:14
  • I used to have an Adaptec SCSI controller that had a rattlesnake of hard disks, floppy disks, and a tape drive hanging off it that were scavenged from my further employer's surplus. But still, the installation and operation were much more straightforward than operating all these devices off their dedicated controller cards. – tofro Jan 20 '18 at 11:04
  • It was more about compatability than performance. HP used SCSI a lot in the 80's, even tape drives, I seem to remember. – Steve Jones Jan 27 '18 at 20:47
  • 1
    Why add a floppy disk controller when you already have a SCSI controller? – fuz Apr 29 '20 at 12:09
  • Elitist? I suppose so - I paid $200 for an Adaptec bus-mastering SCSI card (the disks were, ahem, "DEC surplus") on my first PC because I looked down my nose at the idea of programmed disk I/O. Though I still had the ordinary floppy. – dave Dec 20 '20 at 16:42

3 Answers3

21

When you're designing a system you might want to avoid messing with floppy controllers. Just implement a SCSI interface as a one-for-all and use SCSI drives, no matter what medium. Clean approach.

Zac67
  • 3,914
  • 1
  • 17
  • 23
  • 1
    That makes sense ! – Eric Cartman Jul 30 '17 at 22:43
  • 8
    SCSI was really the USB of its day. Drivers could get messy, and a missing terminator could make things act wonky (or not at all), but it provided a fairly elegant abstract interface that worked with a variety of devices. Usually. – Robert Columbia Aug 03 '17 at 20:05
  • SCSI was itself a bus that you could attach several devices to - didn't have limits as low as a floppy controller or IDE - so you could just get a ribbon cable with a bunch of connectors and stick your floppy, your disk, your tape, your plotter, etc on it without needed additional controller chips (or boards). – davidbak Dec 20 '20 at 16:11
13

For non-PC/home computer systems (UNIX workstations, small implementations of mainframe architectures) it could have been cumbersome to use some of the existing floppy controller chips, since they were usually optimized for certain microprocessor and/or bus systems different from what these machines used - or knowledge how to set these chips up was not available in house and/or nobody wanted to need to keep this kind of knowledge in house.

Using SCSI meant making floppy drive controlling hardware somebody else's problem, and since it is a comparatively complicated (quite some semi-analog decoding stuff etc) but per definition not high-performance or performance-critical subsystem, there was little value in NOT making it somebody else's problem.

rackandboneman
  • 5,710
  • 18
  • 23
  • 6
    +1 By reading your answer I came up to the conclusion that SCSI was more or less the one size fits all USB of the time, while it had drawbacks it definitely had many advantages as well. – Eric Cartman Aug 01 '17 at 15:22
  • 7
    USB mass storage actually uses a SCSI-based application protocol :) – rackandboneman Aug 01 '17 at 17:38
  • USB-attached-SCSI (UAS) is only used by high-performance devices. Simpler devices use the standard USB mass-storage protocol. – Zac67 Aug 03 '17 at 20:14
  • 3
    The standard USB mass-storage protocol is described to be based on RBC. "The purpose of this document is to provide a command set of reduced requirements and options from SCSI Block Commands for block devices." from a t10.org (these are some official dudes that had some to say on scsi...) document that describes RBC. SCSI enough... Oh, and thinking in "device classes" has always been a SCSI thing too... – rackandboneman Aug 03 '17 at 20:19
  • 1
    In the PC world, it was more like "one size doesn't fit your budget at all" .... – rackandboneman Aug 04 '17 at 21:00
  • 2
    On the other side of the spectrum were development boards and single board system (also bus based systems like VME) for industrial and embedded designs. Complex boards attached a SCSI controller sometimes, but the space for a FDC and glue logic on the high integrated boards weren't spent very often (if ever). – Johann Klasek Oct 01 '17 at 20:55
3

Because the machines using them had no other storage bus. At the end of the 90s I had an (already old) Sun Sparcbook. It only had a SCSI bus, so its built in floppy was SCSI.

chriscowley
  • 131
  • 2