Null Meter...build one?

[redrabbit]

The meat:

With recent advancements in opamp accuracy, is it possible to construct a sensitive DC null meter?.....down to 1 microvolt ?


The details:

I have several commercial Null Meters (null detectors) from Fluke, HP, and JRL.
All of these but one (Fluke 887), use mechanical choppers.
Can a modern version be built, DIY ?

Now I will assume that to achieve the sensitivity of say, a Fluke 845,....that controlling noise, EMF, and signal guarding is half the battle...maybe more. It seems to be an art to attain stability at the microvolt level on the Fluke, or any of my other meters.

Am I making any sense?....or wasting my time?

=RR=

[Conrad Hoffman]

This is almost trivially easy today. Use battery power and one of the chopper stabilized opamps from LT. Way back I had a design for such published in Electronics Now as one part of a metrology series- voltage reference, null detector and KVD. What you get when you buy a Fluke 845 is mostly the wide range divider, 300V to 1uV or something like that. The LF noise performance isn't all that great. The HP419s are also nice, though the neon choppers are getting a bit long in tooth. Both units have trouble with their rechargeable batteries at this point, and the batteries are actually used as voltage regulators, so they have to be right for the units to perform well.

[redrabbit]

Did you use an analog meter (needle)?
So what is the basic circuit look like ?
(pardon my naivete)

For example, I have two voltages I want to match.
One is near 10volts DC (the unknown), the other is a 10v standard (the reference).
Is it something like ....one signal is passed through an inverted opamp, and the other through a non-inverted.......and the "unknown" is adjusted until they achieve a "null" at the two opamp's outputs?

If the answer is yes, is gain used in the opamps ?


I have some (highest grade) chopper opamps that are waiting for your answer (ha).
LTC1049 cn
LTC1052 cn
TLC2652 ai
TLC2654 ai

=RR=

[redrabbit]

....or did I just describe an AC null ..?

I 'dun confused myself.

[Conrad Hoffman]

Take the simplest case, comparing two ten volt sources. Ground 'em together, then hook your DVM between the two hot terminals. No ground connection to the meter. Read the difference. Now, let's assume you can adjust one of them, or that they're very close together. You need more gain so you can see 1uV or so. You'd use your battery powered stabilized opamp as a front end for your DVM. You could also just include a meter, but DVMs are common as dirt, so that's what I used. An interesting thing happens when the voltages matched. Whatever the impedance of the circuit is, high or low, the voltage across it is near zero. Thus, no current flows. It's because of this you can also do exact scale factors and measurements using a KVD or other divider.

I have to see if I can find an electronic copy of the circuit I used. With my luck its probably on a 3.5" disk that I have no way to read anymore. I do have a paper one I can scan, but obviously that's less desirable. I used an LT1050, but several others are fine. It was a non-inverting amp, with a divider on the input ganged with a divider on the feedback. A DVM was used for the meter, set to 0.1VDC full scale. The most sensitive range was 10uV full scale input, giving insane resolution with any reasonable DVM. Watch out for the limited supply voltages on the CMOS opamps- I used a 9V battery and split it with a TLE2426. I also have a PCB layout for it. Gimme a couple days to find this.

[Conrad Hoffman]

Here's something that should work.

[redrabbit]

That's great.
Thanks for finding it.

The 100k trim is for....cal?

Would implementing a "guard" be worthwhile?
If so, it would be driven by....?? and start from...???

Actually something like this could be built in a small box with pins mounted (protruding), so it slides right onto the DMM's input jacks.....like a parasite.

=RR=

[Conrad Hoffman]

The trim for for residual DC offset. Even though the opamp is chopper stabilized, you're bound to have some slight offsets at the lowest scales. If I remember right, these opamps have enough trouble driving their own feedback, much less any load, so if you add an analog meter, add a second stage to drive it. No problem with a DVM. I made mine in a small metal box- the box is the shield and guard as I haven't seen any need for double enclosing it. It connects to the DVM with a few inches of cable and banana plugs, but you could certainly make it plug in directly.

The other parts of the original article were for a stable ten volt reference, but LT stopped making the metal package reference it was based on. Plastic packages don't have the long term stability of metal. There was also a scheme for doing a nice cheap KVD without expensive rotary switches. It simply soldered the resistors to the bottom of large pin headers, then you connected to the next bank with a plug on a pigtail. It was easily good to 10 PPM with careful matching of ordinary 1/4W MF resistors. With the voltage reference, KVD and null meter, one had the basics for a "Mini-Metrology Lab", the title of the series.

[AndrewT]

http://www.edn.com/article/CA74477.html

is this the one?
But, I can't open the diagrams nor load Part 1.

[AndrewT]

http://www.edn.com/article/CA74453.html
Part 1, but still can't open the pdfs.