[time-nuts] Off the wall: anyone with experienceprogrammingHP3456A?

Magnus Danielson cfmd at bredband.net
Sat Sep 10 17:13:05 EDT 2005

From: "John Miles" <jmiles at pop.net>
Subject: RE: [time-nuts] Off the wall: anyone with experienceprogrammingHP3456A? 
Date: Sat, 10 Sep 2005 12:20:11 -0700

> The unfortunate truth is that GPIB is actually a pretty flaky thing.  It
> seems there is always a certain amount of swing-a-dead-cat empiricism
> involved with getting any complex GPIB hookup to work reliably.

I think you need to be fair to it. GPIB traces its roots back to the 60thies
and some of the oddities in the GPIB programming environment is to handle older
instruments that is pre-IEEE-488 era. Then again, IEEE has been revisioned and
today we should speak about IEEE-488.1 and IEEE 488.2. There is more revisions
comming, including the NI acceleration. Then look at the command improvements
from original IEEE-488, to IEEE-488.2 and SCPI. It has been an evolving
technology. Now we run it over IEEE-488.1, RS-232, RS-422, IP/RPC (VXI-11) and
what have you. No wonder that combining various generations can cause trouble.
When did it not cause potential trouble?

> I can't talk to my Tek 492AP SA unless I turn on the 8566B or the 8657A that
> shares its GPIB bus, for instance, and I am fairly certain that all of my
> hardware and cabling is in good shape.  I've run into many little hangups
> like this over the past few years.  None of them have been showstoppers, but
> none of them have made me think too highly of the GPIB spec as a robust
> communications mechanism, either.

Having looked the monster into the throat somewhat, I tend to agree. Some parts
are rigid and nicely done, but then as you come into other aspects it gets a
bit more diversified and maybe not so well worked though originally. As the
spec reads today, it is a bit strange, but fits together in the end.

> It is very possible that one 5370B may have problems in a particular GPIB
> configuration that will not show up in another configuration.  Someone has
> already noted that there were a couple of different firmware releases for
> the 5370 counters that addressed GPIB support issues.

Just look in the HP5370B service manual and you see that the GPIB interface is
"passive" (i.e. TTL chips rather than LSI chip). It may not be the marvel of

> It seems wise to keep GPIB programs as simple as possible, using as few
> features as you can.  Whenever I've tried to get fancy, I run into trouble.

Depends how much of the modern features you have with you. Also, devil is in
the details. Hmm, maybe I should study those little evil details. 


More information about the time-nuts mailing list