Using UNIX to control experiments with a 386 box

[Timothy S Sullivan]

	First I want to thank all of you who took time to respond to
my inquiry about controlling experiments with UNIX and a 386. I will
attempt to summarize the information received.
	The first concern is choosing the operating system. UNIX
itself has the problem of not guarenteeing that it will get around to
any particular task in any particular time, so it is necessary to have
real-time extensions. I knew about VenturCom's VENIX/386 version 3.2.
(VenturCom, Incorporated; 215 First Street; Cambridge, MA 02142;
(617)661-1230). This product incorporates real-time capability with
ISC 386/ix 2.0.2. This would appear to do the job at a cost of
(operating system + development system, 2 user) $1643.  Unfortunately,
VenturCom was unsure of whether the system would work with my
hardware. The concern centered around my Data Technology DTC 6280 ESDI
controller with its CDC Wren III hard drive, and my ATI VGAWonder
card. That they coudn't guarentee it would work is not surprising
considering the large number of boards out there. But they also
wouldn't let me have a no-risk trial. I can't afford anything like
$1600 without assurances that it will even run. Can anyone explain why
this is a sensible policy?  Seems to me that they would want to know
of additional systems that the software would work with?  (ISC said no
problem for 386/ix, but I didn't get the impression that the person I
talked to thought about it very much and the hardware is not on their
386/ix compatible devices list I got from VenturCom. Has anyone used
either of these boards with 386/ix?)
	As a result of my posting, Vik Sohal (voder!lynx!vik at apple.com)
 of Lynx Real-Time Systems contacted me about LynxOS. (Lynx Real-Time
Systems; 550 Division Street; Campbell, CA; (408)370-2233; FAX
(408)370-2456). LynxOS is a "clean room" re-write of UNIX and has
real-time capability. Mr. Sohal didn't seem to think that my hardware
was a problem. The price quoted for the LynxOS development system was
$1495 (which appears to include TCP/IP and no mention of the number of
users). Mr. Sohal was very helpful and sent along a packet of
information including a book of example applications code called
"Developing Real-Time Applications with LynxOS". I'm exploring this
further now.
	The next problem is drivers for IEEE488 boards and a video
capture card. The IEEE488 card is easy and I rate a RTFM at least when
it comes to VENIX. I didn't know that you should try the board
vendors.  Anyway, National Instruments ( (800) IEEE488 (!) or (512)
794-0100 thanks to fritzz at lamont.ldgo.columbia.edu for the numbers)
has a UNIX System V 3.2 driver available for one of their boards.
(National also has Xenix drivers according to Jan Parcel
(jan at orc.olivetti.com) as does IOTech (216) 439-4091). There also
exist converters from RS232 to IEEE488 available from Black Box (?)
according to Scot E. Wilcoxen (sewilco at DataPg.MN.ORG) Bill Hatch
(uunet!bts!bill) has actually used such a device from IOTech.
	The video capture card is more difficult. If you have a lot of
money there is the Parallax board (reportly about $8000) which is
supported on many different platforms including ISC 386/ix.  (Thanks
to Rick Chimera carm at umd5.umd.edu for this information. I've heard
about this board for Sun's. It can put live video into your X-windows
system.) Vik Sohal tells me that Lynx is working on a driver for a
Data Translation 2853 frame grabber, about $1700.
	 Tom Herbert (herbert at kodak.com) told me about a German board
called ines that works with 386/ix, but didn't tell me if it was an
IEEE488 board or a video board.  (ines - Innovative
Elektronik-Systeme; Neuenhofer Allee 45; D - 5000 Koln 41; West
Germany). Chris Barr (frog!barr at harvard.harvard.edu) saw ISC display a
video demo at UNIX Expo in NYC almost a year ago, but doesn't remember
the name of the video board vendor.
	In case there is anyone out there looking for other drivers,
VenturCom knows of drivers for Analog/Digital I/O boards from Data
Translation and Stepper Motor Controllers from Scientific Solutions.
	Everyone was in agreement that I needed at least 4 Megs to run
X Windows and preferably 6. I also asked a question about using using
DOS device drivers under VP/ix to get around the device driver
availability problem and then use real time extensions to make the
VP/ix task a high priority. The answers were mixed. Some thought it
was possible but very slow, others thought it was possible, but had
some limitations on what could be done, others thought it was
impossible.
	Thanks again for your help.
					Tim Sullivan