34786Re: [midatlanticretro] S100 Floppy Controller
- Mar 18, 2014The points below are: a UART as a step "up" from TTL designs; and the
cost/benefits decision about using cheaper parts but more development time.
There were a lot of kitchen-table companies producing lots of kinds of
microcomputer products through the 1970's. S-100 and other "bus"
architectures made it even easier, to just produce a PC board and not a
whole desktop computer. At any time, any new company (person at a
kitchen table) could produce a "latest and greatest" product. But old
companies had old inventory, and a base of users with that old technology.
The common catchphrase in the era was "to decide when you are going to
waste your money", as what was new today would be old and cheaper
tomorrow. But you had to buy something, to participate at all.
So, while any one company would make one, or two generations of floppy
controller; the impression of the era was of rapid change, because LOTS
of companies followed EACH OF the trends. FDC chipsets became standard
(later) because the chipsets did all the work for any new design.
But old designs, already paid for, still producing revenue, did fine for
years, until those customers found it convenient (eventually necessary)
to buy a "new" computer. That often took years.
Put another way: a response to "you could have made a more advanced
design" is "but I made something rather than delay, and made money, and
helped my customers". It's not always true, of course, in individual
cases; but it's true in the aggregate, given the conditions I've mentioned.
On 3/18/2014 6:50 AM, Paul Birkel wrote:
> Which is why I wondered about the IBM brow-ness :->. Surely using a
> UART was a step "up" from discrete TTL (if that's what was the original
> IBM S/370-era implementation)!
> bob applegate wrote:
> And the UART was far more advanced than an Apple disk controller;
> all the serialization and track stepper control was done in software!
> Bear in mind how expensive an FDC was back then compared to other
> chips. It was probably hard for engineers to justify using such
> expensive devices. If the floppy controller board cost $200 to make
> but needed only 500 hours of software work, versus $40 in parts and
> 2000 hours of software, usually the latter won. Been there, done
> that with having LEDs rejected from a design because of the cost of
> the two resistors and two LEDs!
Herbert R. Johnson
http://www.retrotechnology.com OR .net
- << Previous post in topic