[volt-nuts] Agilent 34970A (34401A) self test error 606

Andrea Baldoni erm191ba3 at ermione.com
Sat Dec 13 13:35:46 EST 2014


Hello all.

Following the thread on time-nuts about the ebay seller yixunhk and afraid to
have been a victim of their way to do business, I decided to thoroughly test an
Agilent 34970A unit I bought from them some years ago in "almost-new" shape and
practically never used since then.

The 34970A is a data acquisition unit with builtin multimeter identical
to 34401A, with some math function absent in 34401A like the direct measurement
of RTDs. The ADC is a "Multislope III" and you may find it in other instruments
as well, like the E3631A power supply.

I took the opportunity to use it to watch the temperature of an experiment, so
I connected a 100R platinum RTD and begun.
After an hour I noticed a sudden 10x or more increase in noise, so much that
the measure could not be read easily while was still to the tenth of C just
a moment before.
After having excluded cabling pickup, sensor defects, and so on, I launched the
self-test and got Error 606.

I left the instrument on for a full day, to let it stabilize and maybe dry up
from the moisture, even if it has always been kept in the lab, and repeated
the test with the same error. The test imply a power cycle and the error
didn't wanished just with that.

Today I was ready to do further checks, but after a cold start, the noise is
normal (I think, being +-17PPM worst case in the two wire 100R range measuring
a 100R precision resistor at 6.5 digits resolution) and the self test pass.
(the 34401A noise with the same resistor is much better, +-2.5PPM worst case,
but I expected this because of relays in switch units and wiring issues in my
setup of the 34970A)

The Error 606 description is:

Rundown gain out of range - This test checks the nominal gain between the
integrating ADC and the A1U205 on-chip ADC. The nominal gain is check to +-10%
tolerance.

Someone ever had such kind of fault?
Do you think that the problem may come from the reference LM399 and/or
reference amplifiers, the integrators, or the on-chip ADC? (in that case
probably the instrument could not be repaired at component level)

Someone has the detailed description of the "Multislope III" patent?
It seems it's exactly what's described in the HP Journal - April 1989, page
10, figure 4, about Multislope Runup, using the on-chip ADC instead of a
comparator to extract some more bits of resolution.

I have also some questions about the schematic diagram of the 34970A: the
componente marked never_load (page 237 of the service manual, A/D converter
schematic, component R445 for instance) aren't populated on the board or
the meaning is another? Also: the precision of the runup setting current (by
100K resitor part of the array U102-E) isn't lost by dividing it through
non-matched R440+R443 ?

Best regards,
Andrea Baldoni


More information about the volt-nuts mailing list