Classic busses (was Re: Whats a reasonable collection?)

From: Roger Ivie <IVIE_at_cc.usu.edu>
Date: Thu Jun 28 16:22:48 2001

> >Turbochannel is the bus I've seen most often with DEC RISC workstations,
> >although I think they were not the only vendor to use it. The MIPS boxes
> >and the early Alpha boxes were Turbochannel. Mostly, it was used for
> >frame buffers, but I think there were one or two non-graphic Turbochannel
> >cards.
>
> I know that there have been at least TC VME expansion chassis, an FDDI
> adaptor, SCSI and Ethernet.

A company I used to work for produced a VME expansion chassis, serial ports,
parallel ports, T1, MIL-STD-1553, a real-time clock, and did some initial
work on IEEE-488. I was the only device driver guy, and I supported most
of these (exceptions being VME (supported by DEC) and IEEE-488 (DEC did
a device driver, but I was too busy to document and package it; had I not
been too busy, _I_ would have written the driver to begin with) under
Ultrix, OSF/1, Alpha/VMS, and VAX/VMS. I did a device driver for one of the
modules under MIPS/OSF/1, but that never went anywhere because DEC pulled
the plug on that OS (I do have tapes somewhere, though).

I did some work with another company's TC QBus expansion chassis and I
designed the TC adapter for the VAXstation 4000/60 and /90. I'm aware of
other TC VMEbus expansion chassis (DEC had one, as did Bit3), other serial
and parallel ports (Magma), and several other options.

TURBOchannel was very easy to interface to; the bus interface on most of
our options consisted of a rank of registered bus transceivers and a
PAL22V10. I once built a bus torture device from a rank of registered
bus transceivers and an AM29CPL154; with only those five parts, I could
exercise the entire bus protocol, including determining the maximum DMA
burst length. PCI, on the other hand, requires some pretty complex state
machines.

Roger Ivie
ivie_at_cc.usu.edu
Received on Thu Jun 28 2001 - 16:22:48 BST