Re: [volt-nuts] 3458A - To Modify or Not To Modify?
Poul-Henning:
Excellent point that I had overlooked. But I agree with you that HP should have used some sort of securing method for the NVRAM. One good drop from ye olde UPS brown suit gorilla and it would be all over. I have seen these sort of devices that are heavy, strapped in as you point out. No spec on how much G force the meter can be subjected to and still work correctly. I have had 3456A meters calibrated, and upon request Agilent will send you an empty shipping box designed for the instrument which is just like the original shipping box when the meter was first delievered. They certainly return the meter in this box, no matter what packaging you send it in, to the calibration facility. Seeing these boxes I would believe that the meter would survive most anything short of falling out of the plane at 40,000 feet.
I guess that then this would affect what Joe would want to do to his A5 board. Also I doubt that Loveland would take the time to change the NVRAM on Joe's board, just replace it with a "new" one. If they would put in an A5 board with the new surface mount "snap hat" NVRAM then he would not have anymore problems. Of course we don't know if Agilent is still making new A5 boards or if they are "refurbishing to factory specs" old boards. But then the 3458A is still a current product being sold. I don't know.
Since I have my local standards, shipment by UPS didn't enter my mind.
Bill
My conversation with the folks in Loveland led me to believe that they do
almost exclusively component level repair in an effort to preserve the
'character' of the meter. I also got the impression that they wanted to
minimize uncertainties about parts long term reliability by keeping as much
of the original meter as possible. When the NVRAM's are replaced, they
unsolder and resolder them using NVRAM's with date codes of less than a year
old. My understanding of the DALLAS chips is that the 'battery' is not
connected until the first time the chip is powered up. Therefore, the 'new'
chips should last 10 to 15 years although they consider them in need of
replacement when over 10 years old.
According to the parts list info on the website and Service Note 19A, it
would appear that only the 03458-80047 and 03458-80048 (Opt. 01) variants of
the A5 board are available. As best I can tell, the 03458-66505 was
replaced by the 03458-69505 which appears to have been a 'refurbished'
03458-66505. However, when searching for that, it is 'obsolete'. The only
thing I can find on their site is the 03458-80047 'Update Kit'.
From my perspective, if I want to have the meter long term, having the EPROM
and NVRAM's socketed would be desirable to permit archiving the data,
reprogramming new NVRAM's when the time comes, and ease of updating the
firmware if needed. Otherwise, it's unsoldering and resoldering. I hadn't
really considered the issue of them becoming dislodged from their sockets.
There was/is an issue with some NVRAM batteries not lasting as long as
originally thought, discussed in Service Note 20, and once the data is gone,
if you don't have it archived and can't reprogram the NVRAM, it's off to HP
for repair and/or recalibration at $550 to $2350.
However, perhaps there is a way to harvest the data by HPIB, archive it that
way, and use HPIB to 'reprogram' new NVRAM's without the need to do a 'cal'.
In any event, old NVRAM's, at some point, will have to be removed and new
one's installed.
I am trying to solve the HPIB conundrum now. It promises to be a fairly
steep learning curve. All suggestions appreciated.
Joe
-----Original Message-----
From: volt-nuts-bounces@febo.com [mailto:volt-nuts-bounces@febo.com] On
Behalf Of Bill Gold
Sent: Saturday, November 05, 2011 11:16 AM
To: volt-nuts@febo.com
Subject: Re: [volt-nuts] 3458A - To Modify or Not To Modify?
Poul-Henning:
Excellent point that I had overlooked. But I agree with you that HP
should have used some sort of securing method for the NVRAM. One good drop
from ye olde UPS brown suit gorilla and it would be all over. I have seen
these sort of devices that are heavy, strapped in as you point out. No spec
on how much G force the meter can be subjected to and still work correctly.
I have had 3456A meters calibrated, and upon request Agilent will send you
an empty shipping box designed for the instrument which is just like the
original shipping box when the meter was first delievered. They certainly
return the meter in this box, no matter what packaging you send it in, to
the calibration facility. Seeing these boxes I would believe that the meter
would survive most anything short of falling out of the plane at 40,000
feet.
I guess that then this would affect what Joe would want to do to his A5
board. Also I doubt that Loveland would take the time to change the NVRAM
on Joe's board, just replace it with a "new" one. If they would put in an
A5 board with the new surface mount "snap hat" NVRAM then he would not have
anymore problems. Of course we don't know if Agilent is still making new A5
boards or if they are "refurbishing to factory specs" old boards. But then
the 3458A is still a current product being sold. I don't know.
Since I have my local standards, shipment by UPS didn't enter my mind.
Bill
volt-nuts mailing list -- volt-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts
and follow the instructions there.
In message 000001cc9bd6$9bf05fb0$7ce45d47@home, "Bill Gold" writes:
Poul-Henning:
Seeing these boxes I would believe that the meter would survive
most anything short of falling out of the plane at 40,000 feet.
One of the employee newsletters (probably from hparchive) had an
article about the development of shipping boxes and how they tested
them by dropping boxes with CRTs etc.
Article claimed it saved them a fortune...
--
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 A429DA6E7F6843549CF0A35C3AF15508@S0028384766, "J. L. Trantham" wri
tes:
However, perhaps there is a way to harvest the data by HPIB, archive it that
way, and use HPIB to 'reprogram' new NVRAM's without the need to do a 'cal'.
There is.
You can read and write the entire address space using the MREAD and
MWRITE HPIB commands.
However, while you can just read the NVRAM with MREAD, you cannot
write it with MWRITE.
There are a number of safety circuits, which must be tickled just
the right way, to open the write signal to the NVRAM.
This is actually a really good design btw. It's made that way to
protect the NVRAM contents from all sorts of bogons, such as saggy
power-supplies, glitches, software-bugs and so on, and it works.
I have tried, mostly for fun, to overwrite the 40k/7V cal constants
via HPIB on my meter, right before I unsoldered the old NVRAM.
You can do it, but you have to download (to an array), and execute,
a 68000 relocatable procedure which does the dirty deed for you.
If you can't write a 68000 program in assembler, this is not something
you should ever even think about attempting.
The entry points you need to call depends on the software version
of your meter, so you will get nowhere without a disassembly of
your firmware version.
If anybody ever gets stuck with absolutely no other sensible way
to fix bad CAL constants (as in "Yeah, I'm on the South Pole and
can send it to Agilent for cal when the next plane comes by, in
six months time..."), I'll help them.
Until then: Don't even think about it, but use an eprom programmer
instead.
--
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.
A year ago, or so, I didn't understand EPROM programmers. I solved that.
Recently, I solved RS232C communication with a 7081. Now I face
communication via HPIB. That promises to be harder but potentially doable.
68000 Assembler would require a return to engineering school.
Thus, the desire for sockets.
Joe
-----Original Message-----
From: volt-nuts-bounces@febo.com [mailto:volt-nuts-bounces@febo.com] On
Behalf Of Poul-Henning Kamp
Sent: Saturday, November 05, 2011 12:13 PM
To: Discussion of precise voltage measurement
Subject: Re: [volt-nuts] 3458A - To Modify or Not To Modify?
In message A429DA6E7F6843549CF0A35C3AF15508@S0028384766, "J. L. Trantham"
wri
tes:
However, perhaps there is a way to harvest the data by HPIB, archive it
that
way, and use HPIB to 'reprogram' new NVRAM's without the need to do a
'cal'.
There is.
You can read and write the entire address space using the MREAD and
MWRITE HPIB commands.
However, while you can just read the NVRAM with MREAD, you cannot
write it with MWRITE.
There are a number of safety circuits, which must be tickled just
the right way, to open the write signal to the NVRAM.
This is actually a really good design btw. It's made that way to
protect the NVRAM contents from all sorts of bogons, such as saggy
power-supplies, glitches, software-bugs and so on, and it works.
I have tried, mostly for fun, to overwrite the 40k/7V cal constants
via HPIB on my meter, right before I unsoldered the old NVRAM.
You can do it, but you have to download (to an array), and execute,
a 68000 relocatable procedure which does the dirty deed for you.
If you can't write a 68000 program in assembler, this is not something
you should ever even think about attempting.
The entry points you need to call depends on the software version
of your meter, so you will get nowhere without a disassembly of
your firmware version.
If anybody ever gets stuck with absolutely no other sensible way
to fix bad CAL constants (as in "Yeah, I'm on the South Pole and
can send it to Agilent for cal when the next plane comes by, in
six months time..."), I'll help them.
Until then: Don't even think about it, but use an eprom programmer
instead.
--
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.
volt-nuts mailing list -- volt-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts
and follow the instructions there.