Re: Homebrew STG card

["Steve Carlisle" <quatsein-at-island.net>]

I am not sure if you know  about a IC manufacturer called
LSI, they produce the LS7166 and the LS7266(dual) and
a lot of other motion related chips. My pricing is not current,
so there is no sense on commenting on the price,

Steve

----------
From: Matt Shaver <mshaver-at-erols.com>
To: Multiple recipients of list <emc-at-nist.gov>
Subject: Re: Homebrew STG card
Date: Sunday, February 27, 2000 11:24 AM


> From: Paul Corner <Paul.Corner-at-tesco.net>
> Matt asked :
> > I'd like to add two 8254 ICs to provide high frequency stepper outputs
as
> > well.
> 
> Would you want the timer outputs brought out to a connector. Or would an
> interrupt need to be generated from at least one timer ? - Subject to
space, a
> small scratch pad area with access to the IRQ and timer outputs could be
> included.

Yes, the timer outputs are to be the step signals. There isn't any
requirement for an interrupt. The clock inputs (3 for each 8254) need to be
driven from one of those metal oscillator modules (one module drives all
three 8254s together), probably at 1MHz. The gate signals (3 for each 8254)
will probably need to be pulled up (or down, I forget which) to enable
continuous counting. A small scratch area would be nice, but I'll try to
pin
down the exact requirements by experimenting with a prototype card I have.
The main issue with the gate signals is whether we can set the output
frequency to zero by programming the 8254, or whether we'll need some bits
on
a port connected to the gate inputs. The direction bits will have to be
assigned to one of the byte wide i/o ports somewhere, but this is just a
software issue (although it's likely that the direction bits will need an
entire byte wide port all to themselves).

I'm all for starting with the ISA bus as PCI seems tough to do unless you
go
with a ready made solution like the PLX chips, and I agree wholeheartedly
about using generic, multiple source parts wherever possible (the 7166s and
MAX d/a being somewhat of an exception here). It's true that ISA is
disappearing on new motherboards, but I think most will continue to have at
least one for a while yet and older machines will be with us for years.
When
progress finally catches up with us we can either re-layout the board with
the interface du jour, or build an ISA to whatever adapter. The ISA bus is
rather generic and as long as we stay away from interrupts, DMA, etc. we
could probably make a parallel port, serial, or USB to ISA adapter in a box
that the old card could plug into for folks who want/need an alternative
interface (this might be something to do sooner, rather than later).

I had suggested the 100 pin high density SCSI connector as a way to get all
the signals out of the box on the edge of the card (I wish IBM had built
the
PC like VME and Multibus, with the long edge of the card facing out of the
rack...). This is kind of expensive, so maybe we should just use 2x25 pin
connectors like STG and we can use standard (inexpensive) SCSI cables to
connect to the external (screw terminal) interface. I like the idea of DB9
connectors for the encoders, but this will need to be on the external
interface board due to space considerations. The digital i/o should
probably
be on a 50 pin (2x25) DIL connector that conforms to the OPTO22 standard so
that folks who care to can use that interface.

> I have a target length of 160mm for this card, 200mm is the most I really
want
> to go.
What's the reason/

When you get a chance, I'd like to see your preliminary sketches. In the
interval I'll try to get some other information together.

Matt

Reference:

Scroll down near the bottom (past the cows):
http://jhd486.mast.queensu.ca/~jon/