[volt-nuts] 3458A Questions

[Bill Gold]

Joe:

    Even more than the basic questions you ask should be "Is the 3458A working to specs?".

    This was my experience.  I decided I want one of these to even better augment my test equipment collection.  The long scale 8.5 digits would help to eliminate the need to use my Fluke 720A for some applications.  I watched Fleabay for months and months but was always afraid that I might buy a meter that was not working to specifications, even though it passed the start up "self test" which most sellers seem to state, along with "I don't have the expertise or equipment to test this unit further".  As you have seen, these units can range from a starting bid of $100 (these usually have some sort of problem with self test) to "Buy it Now" of $5000 and up.  Some of the really expensive units ( > $5000 ) will include a ROR at your expense.  Some will even include a "recent calibration" but no information as to who did it.  Some do have a picture of the back panel with the S/N, a few will state the S/N, and some will answer if you ask them what the S/N is.

    Then I saw 2 units for sale from a seller that was only a few miles from where I lived, for $3500 each.  The seller even listed their phone number so I called and asked if I could inspect the units and bring a voltage reference (Fluke 731B which I know is well with in +/- 5 ppm) and a nominal 10 K resistor with me.  The answer was "yes" and I went over to their warehouse.  One failed at turn on with an error that indicated that it had more problems than a failed NVRAM checksum.  The other seemed to have no problems and I ran the extensive complete "self test" sequence with no failures.  I then connected by 731B and noted that the readout was somewhat low, on the order of -100 ppm.  But I had no experience with the 3458A, and it had just been turned on, after probably years of storage, and I figured that it would take hours to reach an accurate reading.  I did an "ACAL", and when done it indicated a better reading but not what I would have expected.  I hooked up the 10 k resistor but again got low readings.  I figured that the unit simply needed a good calibration to my local standards.

    I bought the unit and took it home, let it warm up for several hours and then did only a "CAL 10" using a Fluke 732A.  It then indicated the correct voltage and connecting a 10k standard resistor it showed that the reading was about -5 ppm from what the known value was for the 10k resistor.  After a week I began to notice that the 10 volt readings were dropping very slowly, but an "ACAL DC" corrected the problem.  But this downward drift continued for the next month and in the meantime I had downloaded Agilent/HP Service Note 18A and this indicated that the problem could be "long term drift" of the A9 internal reference, but that the problem would go away after 6 weeks or so of continuous operation.  After 6 weeks I reasoned that the problem was in the internal reference (03458-66509) I ordered a new one but since Agilent didn't have the standard 66509 in stock I ordered the 66519 (option 002) for a few hundred dollars more.  I installed it and did a "CAL 10" and everything was ok.  But a few days later the drift was still present and at the same rate.  I simply couldn't believe that the 03458-69503, A3, "PCA A/D and Inguard Logic" board was the problem, but it was. The service note S/N range didn't apply. This unit S/N 2328A-12449 had been manufactured at least 15 or more years ago.  I finally diagnosed A3 as being the problem by swapping A3 boards between my 3458A and another 3458A I was able to borrow, and the problem followed the A3 board.

    I then purchased a new A3 from Agilent and everything has been working correctly ever since.  $900 for the A3 and $700 for the A9 boards.  So my $3500 bargain was finally fixed for a total of $5100 and months and months of effort.

    The new 002 A9 option reference has not proven to be as good as expected.  The original A9 has been far more stable.  But I have been "burning in" the 002 A9 for many many months now, on the bench, and I will try it out again some day when I get "a round toit".  The well seasoned standard A9 seems to be rock solid against my 732Bs.  The 001 option is useless to me but came with this 3458A.

    Bottom line here is that the 3458A can have subtle problems with which even a 30 day ROR would not have helped me.

Bill