Differential Voltmeters

[KK6IL]

What makes a differential voltmeter, such as the Fluke 887A sell for as much as $150 on ebay? One can get a more accurate, more versatile meter, such as a HP3456A for that price.  Do some people really LOVE twisting knobs?  Must be some virtue as most differential meters sell.

I was looking to see what a fair price might be for a 885A I saw on Craigslist, and was shocked at what seemed high prices to me.  Even saw a Wavetek 210 sold for $95, one I would have calibrated in 1969. No documentation existing anywhere in the world. Pilot run of 10 or 25.  It was 0.005% DC, 0.1% AC, 0.005% Hz, but the buyer likely wouldn't have known that.

[srb1954]

What makes a differential voltmeter, such as the Fluke 887A sell for as much as $150 on ebay? One can get a more accurate, more versatile meter, such as a HP3456A for that price.  Do some people really LOVE twisting knobs?  Must be some virtue as most differential meters sell.

I was looking to see what a fair price might be for a 885A I saw on Craigslist, and was shocked at what seemed high prices to me.  Even saw a Wavetek 210 sold for $95, one I would have calibrated in 1969. No documentation existing anywhere in the world. Pilot run of 10 or 25.  It was 0.005% DC, 0.1% AC, 0.005% Hz, but the buyer likely wouldn't have known that.

At balance a differential voltmeter has essentially infinite input resistance so it is still useful for accurately measuring voltages with really high source impedances. Modern bench DVMs generally have very high input resistance( >10G ) for low input voltages but revert to the standard 10M input resistance for higher voltages so the differential voltmeter loads a higher voltage circuit much less than a standard DVM.

A second advantage is that the differential voltmeter can be useful for adjusting circuits for very precise minimums or maximums sitting on top of a higher voltage. The higher voltage can be nulled out and the meter sensitivity increased to see the small variations. With a modern DVM you have to use a graphical trend display to achieve a similar effect although adjusting for a minimum or maximum null is made more difficult by the slower updates of the trend display.

[Conrad Hoffman]

IMO, they are cool collectors items, but have little justification in the modern world. Sure, you can invent some situations where they'd be useful, but mostly it would sit on a shelf. I've had a couple and sold them cheap or parted them out. I think there was a Fluke that had a built-in standard cell as a reference, but all those small cells have been junk due to age for a long time now.

[ddavidebor]

They're really practical to calibrate stuff!

[TimFox]

Differential voltmeters are the natural instrument to monitor the drift of a reasonably stable voltage with time or temperature.

[Kleinstein]

I would also consider them more a collector Item than a practical voltmeter. The voltage source may be usable as a calibration source or rater stable adjustable voltage reference.
They were once high grade instruments and can be thus of good build quality.

With an electronic amplifier for the null-meter part, there can still be some residual bias current at the input. So the often repeated claim of infinite input impedance is misleading. They were still high impedance for there time.

[MadTux]

Combined with a calibrator (Fluke 335D, 5440B....), you can basically use any good lab voltmeter (HP3456A, Keithley20xx ....) as a differential voltmeter, by measureing the difference between calibrator and device under test. The good thing is that you get the precision of the calibrator that way, a few ppms in case of 5440B. Normal 6.5digit labDVM usually is a magnitude worse in precision.

That setup is more or less what's inside a Fluke 887A, if I understand correctly. Calibrator/DVM setup also has near infinite input resistance, if DUT and calibrator voltage (+/-1100V) matches and >10Gohm if within +/-10V of calibrator output, as most DVMs have >10Gohm input resistance up to that range. A nanovoltmeter like Keithley 181/182 would be the most accurate instrument for that, but it's often overkill, in my experience. Connecting guard input (HP3456A has it, K20xx don't) of DVM to calibrator high output further decreases input current.

For calibration/comparing of voltage calibrators, that differential setup is the only setup that gives meaningful readings, except maybe with 3458A. Most LM399 based DVMs have far too much noise to do direct comparison between calibrators.

[guenthert]

https://sfbay.craigslist.org/nby/ele/d/cotati-fluke-801br-differential-dc/7214537147.html
(not affiliated, can't comment on its condition)
seems quite affordable.  I don't have the space, otherwise I'd take it (as collector's item).

[Kleinstein]

With the nV meters one may have to be careful, when used together with a calibrator that can provide a high voltage. The protection if often not good for higher voltage, so some 200 V may already cause damage.

Not many of us have a calibrator, so the voltage source part may the more interesting half of the differential VM, not just the use as intended.

[dietert1]

Unless the buyer is sentimental, maybe he wants well aged precision resistors or references. If you get five decades with ten or eleven resistors in each decade, a price of $100 is a good deal. To be honest i also bought a Fluke 845 last year, out of curiosity. I studied it a little bit but i can confirm even after some mods it wasn't a better null meter than a HP 3456A. Meanwhile i bought three more HP 3456As.

Regards, Dieter

[MadTux]

Nice, another one who is in love with 3456A ;-D

[kleiner Rainer]

I own one - a Fluke 893A built in 1982 for an air force contract. It came with user manual and schematics, the built-in recorder output can be seen as a galvanically isolated, amplified output with a gain of abt. 100 in the most sensitive range - nice for nulling bridges (my ESI 250DE comes to mind  ).

Do I need it? Heck - no, but it is fun to use and tbh who NEEDS ten scopes or counters or DMMS?

Greetings,

Rainer

[srb1954]

Nice, another one who is in love with 3456A ;-D

I am of the opinion that the 3456A has a better quality analog front end than the one in the 3457A successor. The calibration of my 3457A seems to have drifted considerably more over time compared to my 3456A.

Have other forum members experienced similar levels of performance with the 3456A vs the 3457A?

[MadTux]

3457A is cheap garbage IMO ;-)
Cheap plastic case, LCD that is hard to read, bad keyboard

Frontend is also weird, any normal DVM has 10GOhm input up to 10V range, like that:
100mV range, AMPx100=> 10V to ADC
1V range, AMPx10=> 10V to ADC
10V directly to ADC
100V div/10 to ADC
1000V div/100 to ADC

On 3457A, 10GOhm input is only up to 3V range and range amplifier/divider goes like that:
30mV AMPx333 => 10V ADC
300mV AMPx33.3 => 10V ADC
3V AMPx3.33 => 10V ADC
30V div/100, AMPx33.3 => 10V ADC
300V div/100, AMPx3.33 => 10V ADC

That's why 30V and 300V range and similar in Ohms mode, are so bad on that one
Not to mention the input goes more or less directly into the range switching hybrid, so if you toast one of the FETs or the range divider resistors in there, you're screwed and need a new hybrid, whereas in 3456A, you just grab your soldering iron and replace the JFET with a J113 or similar.

[KK6IL]

That setup is more or less what's inside a Fluke 887A, if I understand correctly. Calibrator/DVM setup also has near infinite input resistance, if DUT and calibrator voltage (+/-1100V) matches and >10Gohm if within +/-10V of calibrator output, as most DVMs have >10Gohm input resistance up to that range. A nanovoltmeter like Keithley 181/182 would be the most accurate instrument for that, but it's often overkill, in my experience. Connecting guard input (HP3456A has it, K20xx don't) of DVM to calibrator high output further decreases input current.

Interesting thread - Guess I can find an occasional use for a differential. Guess if I don't get the 885, I'll have to repair my Wavetek 204. The Transfermatic switch is all gummed up and might not be repairable.

The Fluke 903 and 825 vacuum tube meters have 500V on the KVD so infinite impedance at null all the way to 500V. The 883, 885, & 887 are infinite impedance at null only up to 11V and 10M above that. The 893 & 895 have 1100V on the KVD so infinite all the way up.

The Wavetek meters 201-207, 210 had 10V for the KVD.  The preproduction model 220 12V or 120V for a nulling voltage, BCD coded A/D converter rather than a KVD.