I had the same problem as you when I got my used 720A years ago. Both
the "A" and "B" decade had a few resistors that had changed over the years.
I don't think as bad as you experienced but still they would not balance
within the range of the adjustment pots. So I "adjusted" those that needed
correcting with either parallel or series resistors as necessary. I used
the best tolerance resistors I could get at that time. I figured that the
"adjustment" resistors were such a small amount of the resistors I was
trying to adjust, that the TC wouldn't be much of a problem. However I did
have to go through these "adjustments" several times as I noticed that
correcting one would then affect others in the string. This is probably
what you are experiencing with needing to "readjust" others again. A few
passes in each string finally got the adjustments into a good range and they
have stayed there for years now. As far as I am concerned this is all "good
enough" for hobbyists or VoltNuts who don't want to spend $5,000 or more to
get FLUKE to refurbish your instrument, if FLUKE is even interested in doing
the job and has the parts. Notice that some resistors in the "A" and "B"
decades are selected at the factory anyhow. If you have the money, then be
a "purest". It might be cheaper just to buy a new 720A.
As you realize now the 720A input resistance is 100k. So you will get a
small voltage drop from the voltage source to the input of the 720A,
depending upon the size of the wire you are using. Of course the proper way
to insure that there is no drop is to use Kelvin connections from the
voltage source to the input of the 720A and then set the voltage source to
external Kelvin sensing. What I usually do is to set the ratio dials to
.999999X and then set the voltage supply to get exactly 10.0000000 on my
3458A. Then you have the best chance to get an idea of the accuracy of the
720A ratio at .8, .7, .6 and so on. Don't let the 3458A autorange when down
to the 0.1 ratio! Then to minimize the effects of the "B" decade, use
.899999X, .799999X down to .099999X. The only problem with this method is
that you don't get to check the .9 ratio. So what you can do is connect the
voltage source to the "1.1" input, set the ratio dials to 1.0000000 and then
set the voltage source to give you 10.0000000 volts on the DVM. Now you can
check the .9000000, .8000000 and so on ratios.
I am sure that I did the R203 adjustment as described in paragraph 4-25
of the manual. It has been years now.
All IMHO of course. You are now down into measurement "mud" so to
speak. This is probably why the 3458A development team used the JJA at
Loveland to check the overall linearity of the 3458A A/D. The 720A (or
equivalent) wasn't good enough at 0.1 ppm!
----- Original Message -----
From: "David C. Partridge" <david.partri...@perdrix.co.uk>
To: "'Discussion of precise voltage measurement'" <email@example.com>
Sent: Tuesday, August 08, 2017 7:39 AM
Subject: Re: [volt-nuts] Update on 720A
> I totally get your point that ideally you should use any KVD in
null-balance mode. I do note however that you can load it up to 11mA
without damage though I've not once gone anywhere near that (worst case
insult has been about 0.1mA). However it does also say that to avoid
loading errors you should load with an impedance >= 1TeraOhm (10**12).
> Clearly a 3458A on the 10V and lower ranges is merely > 10GOhm and will
show loading effects.
> So, yes, I do recognise I was "doing it wrong"!
> I don't believe I've inflicted any damage, and have managed to restore it
to a useable state from a fairly sorry condition: When I got it, two oil
bath resistors were *way* off value (202 and 100 ohms low respectively), and
the S2 shunt resistors were about 1.3 ohms high which was far enough off to
prevent S2 calibration. There were were also sundry other problems like
two open circuits in the final decade, a badly worn trim pot, and sundry
wires broken at solder joints.
> Ideally I'd like to replace the KVD "Bridge Balance" pot as that's a bit
noisy but shudder to think what that might cost if I could even find one.
> -----Original Message-----
> From: volt-nuts [mailto:volt-nuts-boun...@febo.com] On Behalf Of Chuck
> Sent: 08 August 2017 13:34
> To: Discussion of precise voltage measurement
> Subject: Re: [volt-nuts] Update on 720A
> Hi Dave,
> Fundamentally, any input impedance from your 3458 is going to throw your
720A out of wack. the 720A is not supposed to have even picoamps of
loading.... nothing, nada, zip.
> I would suggest that before you damage your 720A further, you read the
manual for the 720A and follow the instructions to the letter. An HP3458A
is not a substitute for a null detector and a voltage calibrator.
> The calibration of a 720A requires that no current, none, zip, be drawn
from the 720A on any port. It can only be done by using a voltage
calibrator, and a null detector, as a differential voltmeter.
> Remember, and heed my words, *no* current may be drawn from the 720A.
> -Chuck Harris
> David C. Partridge wrote:
> > Update: After changing the resistor I added in series with R1008 (A
> > decade position 1.0) from 204.8 to 202.4 Ohms (I wonder if this means
> > that R302 is slowly increasing in value back towards nominal) and
> > re-calibrating again, the A decade is now within 1ppm of linear (using
the 10V range and nominally 10V input).
> > I note that when I'd done this, with the A decade set to 0.9 (reading
> > 9.000000V on the 3458A), the actual input voltage needed was 10.000040
> > or so. Is that to be expected?
> > Two of the positions on the B decade wouldn't quite adjust for a null
> > calibrating it. In this situation I think I have two options:
> > 1) Add series resistors to compensate for slightly low value 9.898k
> > resistors in the relevant two positions of the decade and recalibrate.
> > 2) Adjust R203 to change the bridge balance slightly to the 10K ohm
> > decades and recalibrate.
> > Your thoughts on which approach to take is much appreciated.
> > Thanks Dave
> > _______________________________________________ volt-nuts mailing list
> > -- firstname.lastname@example.org To unsubscribe, go to
> > https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts and follow the
> > instructions there.
> volt-nuts mailing list -- email@example.com To unsubscribe, go to
> and follow the instructions there.
> volt-nuts mailing list -- firstname.lastname@example.org
> To unsubscribe, go to
> and follow the instructions there.
volt-nuts mailing list -- email@example.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts
and follow the instructions there.