Re: [time-nuts] OT: Prologix GPIB and HP3478A...The Answer!
Interesting point there. I seem to recall quite a few pullup/pulldown
options in the Atmel port-configuration registers; this may just be a matter
of selecting a mode that looks more like the resistor configuration Chuck
mentioned.
Personally, I think it's fine if you support only 99% of the GPIB devices
out there and leave the remaining 1% to the competition. Your boards have
proven compatible with an incredibly-wide array of instruments at this
point. If the 3478A problem can't be fixed in software, I think the best
answer is, "Sorry, we don't support that one, here's a refund." It
certainly isn't worth adding more chips (and cost) to the board IMHO.
-- john, KE5FX
-----Original Message-----
From: time-nuts-bounces@febo.com [mailto:time-nuts-bounces@febo.com]On
Behalf Of Prologix
Sent: Wednesday, November 28, 2007 10:25 AM
To: 'Discussion of precise time and frequency measurement'
Subject: Re: [time-nuts] OT: Prologix GPIB and HP3478A...The Answer!
Hi Chuck,
Good work identifying the problem! Leaving out the bus driver chips was a
conscious decision based on available board space and was not
made lightly.
Extensive testing had shown that not to be an issue, so far. In
fact, yours
is the first reported case of a drive related problem. However,
we consider
even one incident to be one too many, and will be revisiting the decision
during a future release.
Thank you for brining this to our attention.
Regards,
Abdul
Hi John,
I agree in part.
If it is known that certain GPIB instruments are not supportable,
then it is essential to have that documented fact easily available to
the buyer. Thus far, the Prologix website, and documentation, make no
mention of any possible limitations, or deficiencies. [For instance, it
clearly cannot support the maximum defined GPIB system, or cable lengths.
That should have been mentioned.]
The GPIB standard is very well defined. There really is no reason
to expect that a commercially made controller wouldn't be able to operate
with any GPIB controlled instrument. The HP87B's controller will operate
any device, and I am sure that the controllers I designed as a graduate
student would as well.
I would be angry, except that I already knew that the design had a major
short cut in it before I bought my Prologix. If you recall, I addressed
that issue many months back on this forum.
I bought my Prologix controller based on the good reports that all of
you guys posted. And the need to control a single device, a Tektronix
7854. It will do that, so I am satisfied.
The "Go away kid, you bother me!" approach won't win over many
customers.
-Chuck Harris
John Miles wrote:
Interesting point there. I seem to recall quite a few pullup/pulldown
options in the Atmel port-configuration registers; this may just be a matter
of selecting a mode that looks more like the resistor configuration Chuck
mentioned.
Personally, I think it's fine if you support only 99% of the GPIB devices
out there and leave the remaining 1% to the competition. Your boards have
proven compatible with an incredibly-wide array of instruments at this
point. If the 3478A problem can't be fixed in software, I think the best
answer is, "Sorry, we don't support that one, here's a refund." It
certainly isn't worth adding more chips (and cost) to the board IMHO.
-- john, KE5FX
Hi John,
I agree in part.
If it is known that certain GPIB instruments are not supportable,
then it is essential to have that documented fact easily available to
the buyer. Thus far, the Prologix website, and documentation, make no
mention of any possible limitations, or deficiencies. [For instance, it
clearly cannot support the maximum defined GPIB system, or cable lengths.
That should have been mentioned.]
Agreed, documenting this sort of limitation is pretty important, now that
it's known. That said, his intent has always been one dongle per
instrument, plugged directly into the back without any cables at all. The
first few generations of the board used a DIP switch to set a hardwired
address, in fact.
I have run mine with at least a half-dozen daisy-chained instruments on
20-30 feet of cabling with no problem at all. Some of the 8650-series
signal generators cause loading problems when they're powered off, but
that's true with the NI boards as well.
I would be angry, except that I already knew that the design had a major
short cut in it before I bought my Prologix. If you recall, I addressed
that issue many months back on this forum.
If you want 'anger', try spending time supporting other boards, especially
the off-brands that claim to be NI488.2-compatible. Anger is only the
second or third stage of the grief that process brings.
And GPIB itself is not exactly a textbook example of how to design a
multiuser bus.
I bought my Prologix controller based on the good reports that all of
you guys posted. And the need to control a single device, a Tektronix
7854. It will do that, so I am satisfied.
The "Go away kid, you bother me!" approach won't win over many
customers.
I'm sorry; who said that?
I need to clarify that I've helped Abdul with some testing now and again,
but have absolutely no other business relationship with him, other than the
fact that I happen to live in the same area, and regularly let him (and
pretty much anyone else who asks) drop by to use my test gear.
-- john, KE5FX
In message PKEGJHPHLLBACEOICCBJIEPGCNAB.jmiles@pop.net, "John Miles" writes:
If the 3478A problem can't be fixed in software, I think the best
answer is, "Sorry, we don't support that one, here's a refund." It
certainly isn't worth adding more chips (and cost) to the board IMHO.
I fully agree.
--
Poul-Henning Kamp | UNIX since Zilog Zeus 3.20
phk@FreeBSD.ORG | TCP/IP since RFC 956
FreeBSD committer | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.
Hi John,
I have run mine with at least a half-dozen daisy-chained instruments on
20-30 feet of cabling with no problem at all. Some of the 8650-series
signal generators cause loading problems when they're powered off, but
that's true with the NI boards as well.
With the loading limits on most small microprocessors, this seems reasonable.
And I am glad to hear that it is possible. It extends the use of the
Prologix from one adapter/one box model, to being able to run something
like an 8505A network analyzer where there are 3 or 4 boxes all on the bus.
I waited for version 5.00, with its multi address capability, because I
had envisioned using it one day to add automation to my 8505A system.
I would be angry, except that I already knew that the design had a major
short cut in it before I bought my Prologix. If you recall, I addressed
that issue many months back on this forum.
If you want 'anger', try spending time supporting other boards, especially
the off-brands that claim to be NI488.2-compatible. Anger is only the
second or third stage of the grief that process brings.
Abdul has been great with his support. I have no complaints. I would have
some, if I didn't jump into this deal already knowing pretty well the potential
caveats of the design.
A given instrument's support of the IEEE-488 standard is a little different
from a competitor's support of NI's house brand standard. Kind of like
trying to be a windows compatible operating system when microsoft only
wants you to buy its operating system.
And GPIB itself is not exactly a textbook example of how to design a
multiuser bus.
Certainly not. But how many use it that way? It is generally used as
a single controller with multiple talker/listeners, working sequentially
when addressed by the controller.
I bought my Prologix controller based on the good reports that all of
you guys posted. And the need to control a single device, a Tektronix
7854. It will do that, so I am satisfied.
The "Go away kid, you bother me!" approach won't win over many
customers.
I'm sorry; who said that?
It was the tone of your suggestion that everyone that buys a unit and finds
out that it doesn't work on such-and-such an instrument should be given his
money back, and sent away.
I am all in favor of a product that has known design constraints that keep
its price low. I bought this adapter knowing that was true. It would be
nice, though, to have the option of having the same design available with
true 488 drivers. I think that capability would smoke out a few more
customers at very little extra cost.
I need to clarify that I've helped Abdul with some testing now and again,
but have absolutely no other business relationship with him, other than the
fact that I happen to live in the same area, and regularly let him (and
pretty much anyone else who asks) drop by to use my test gear.
That is what I believed to be true. You have said nothing to imply any
other relationship.
-Chuck Harris
In message 474DD168.3080903@erols.com, Chuck Harris writes:
Hi John,
I have run mine with at least a half-dozen daisy-chained instruments on
20-30 feet of cabling with no problem at all. Some of the 8650-series
signal generators cause loading problems when they're powered off, but
that's true with the NI boards as well.
Part of the IEEE spec is certain limitations on how many devices on the
bus can be powered off.
--
Poul-Henning Kamp | UNIX since Zilog Zeus 3.20
phk@FreeBSD.ORG | TCP/IP since RFC 956
FreeBSD committer | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.
In message PKEGJHPHLLBACEOICCBJCEPLCNAB.jmiles@pop.net, "John Miles" writes:
Agreed, documenting this sort of limitation is pretty important, now that
it's known. That said, his intent has always been one dongle per
instrument, plugged directly into the back without any cables at all. The
first few generations of the board used a DIP switch to set a hardwired
address, in fact.
I bought two Prologix (since the shipping is such a large fraction here
to .eu I might as well)
I have found that it works best if I use one Prologix for my Tek
'scope and the other one for all the HP-kit (7 instruments, most
of which are powered down at any one time)
--
Poul-Henning Kamp | UNIX since Zilog Zeus 3.20
phk@FreeBSD.ORG | TCP/IP since RFC 956
FreeBSD committer | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.
From: "Poul-Henning Kamp" phk@phk.freebsd.dk
Subject: Re: [time-nuts] OT: Prologix GPIB and HP3478A...The Answer!
Date: Wed, 28 Nov 2007 21:58:59 +0000
Message-ID: 1608.1196287139@critter.freebsd.dk
In message PKEGJHPHLLBACEOICCBJCEPLCNAB.jmiles@pop.net, "John Miles" writes:
Agreed, documenting this sort of limitation is pretty important, now that
it's known. That said, his intent has always been one dongle per
instrument, plugged directly into the back without any cables at all. The
first few generations of the board used a DIP switch to set a hardwired
address, in fact.
I bought two Prologix (since the shipping is such a large fraction here
to .eu I might as well)
I have found that it works best if I use one Prologix for my Tek
'scope and the other one for all the HP-kit (7 instruments, most
of which are powered down at any one time)
It is so nice that Tek and HP could agree on a standard for printouts. :P
That was irony for you. From whatever I have seen, they have never worked
together.
The Tek Spec we have at work has a nice feature. TWO GPIB connectors.
One for remote instrument control and the other for printouts only.
Cheers,
Magnus
The IOTech Micro488/P has the same basic design (microcontroller drives bus
directly) and therefore should be assumed to have the same potential
deficiency and costs about $500 today new from the factory (serial interface
only). The only limitation they show on their web site is that it can only
control up to 8 instruments, there is no mention of data rate or cable
length. I have two of them and so far they control everything I have thrown
at them. Being powered from the serial bus (they draw about 5mA), they don't
have much pull-up capability by themselves, probably very far from the
IEEE-488 spec requirements. They advertise it as being intended for laptops,
but many laptop serial ports won't provide the drive power/voltage it
requires...
http://www.iotech.com/catalog/ieee/mini_controller.html
I would like to make sure we find out where the problem is. I agree with
John that the Prologix has proven to be dependable and that the design
approach was a reasonable compromise. Measurement Computing sells a GPIB
driver chip that is in a small surface mount package, but it costs more than
the Atmel chip Abdul is using, and it's just a driver.
Didier
-----Original Message-----
From: time-nuts-bounces@febo.com
[mailto:time-nuts-bounces@febo.com] On Behalf Of Chuck Harris
Sent: Wednesday, November 28, 2007 1:12 PM
To: Discussion of precise time and frequency measurement
Subject: Re: [time-nuts] OT: Prologix GPIB and HP3478A...The Answer!
Hi John,
I agree in part.
If it is known that certain GPIB instruments are not
supportable, then it is essential to have that documented
fact easily available to the buyer. Thus far, the Prologix
website, and documentation, make no mention of any possible
limitations, or deficiencies. [For instance, it clearly
cannot support the maximum defined GPIB system, or cable lengths.
That should have been mentioned.]
The GPIB standard is very well defined. There really is no
reason to expect that a commercially made controller wouldn't
be able to operate with any GPIB controlled instrument. The
HP87B's controller will operate any device, and I am sure
that the controllers I designed as a graduate student would as well.
I would be angry, except that I already knew that the design
had a major short cut in it before I bought my Prologix. If
you recall, I addressed that issue many months back on this forum.
I bought my Prologix controller based on the good reports
that all of you guys posted. And the need to control a
single device, a Tektronix 7854. It will do that, so I am satisfied.
The "Go away kid, you bother me!" approach won't win over
many customers.
-Chuck Harris
John Miles wrote:
Interesting point there. I seem to recall quite a few
pullup/pulldown
options in the Atmel port-configuration registers; this may
just be a
matter of selecting a mode that looks more like the resistor
configuration Chuck mentioned.
Personally, I think it's fine if you support only 99% of the GPIB
devices out there and leave the remaining 1% to the
competition. Your
boards have proven compatible with an incredibly-wide array of
instruments at this point. If the 3478A problem can't be fixed in
software, I think the best answer is, "Sorry, we don't support that
one, here's a refund." It certainly isn't worth adding
more chips (and cost) to the board IMHO.
-- john, KE5FX
time-nuts mailing list -- time-nuts@febo.com To unsubscribe,
go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
I need to clarify that I've helped Abdul with some testing now
and again,
but have absolutely no other business relationship with him,
other than the
fact that I happen to live in the same area, and regularly let him (and
pretty much anyone else who asks) drop by to use my test gear.
That is what I believed to be true. You have said nothing to imply any
other relationship.
No problem; I just didn't want people (including lurkers) assuming that
Abdul shares my opinion or is unduly influenced by it.
The real issue with trying to be all things to everyone isn't just adding a
driver chip to the board. It's that you have to write firmware and software
to round out the compatibility picture. Abdul has thought about adding an
NI488.2 driver layer, which I agree would be nice, but then he'd be in the
business of supporting every NI488.2 application ever written, including
some flaky ones that only work because of undocumented behavior in NI's API.
/Not that I would ever write any code like that,,,
//ducks, whistles, shifts eyes, walks away slowly
-- john, KE5FX