LM199 10V reference

[ivo3510]

Hello, I'm looking for advise for a simple reference I made recently.
I've made LM199 + OP177 10V voltage reference, and I'm not sure if it can be improved in anyway, any advise, please?
Below a simple histogram, taken with Agilent 34401A, calibrated in 2021.
I am feeding the reference board from linear PSU, with 15V from LM317.
The board is notning fancy, just the reference chip, the opamp, and some filtering, as per LM399's datasheet.
The output is set to 10V with a multiturn pot, as per opamp's datasheet - the one used to set the offset voltage.
I've waited 24h with the reference board to settle termally, and the multimeter was sitched for about 2h prior measurement.
This reference is to be used to check from time to time a multimeter - the same HP34401A, and one Fluke 45 I use for my hobby.
These multimeters are used when an analog audio units are to be repaired, or modded - amplifiers, FM tuners, cassette decks, CD's, and other similar units, nothing fancy.

Thank you!

[xvr]

LM199 noise is 7-20 uV (by datasheet). This is fits in your measurements.

[Kleinstein]

The histrogram looks reasonable, about what one expects.

Using the offset trim pot for a fine adjust of the ref . voltage is however not a good idea. The OP-amps offset is temperature dependent (about proportional to absolute temperature). So adding a significant shift from there is effecting the TC. The zero adjustment at the OP-amps should only be used for the OP-amps offset.

[ivo3510]

Thanks,, Kleinstein,
I suppose I should try and regulate the feedback then.
Do I need to swat the opamp with something better?

[Kleinstein]

ADjusting the feedback is the right way to trim the voltage.

The OP177 is well good enough.

[SteveThackery]

Just a quick question: surely this is already a couple of orders of magnitude better than a multimeter can resolve, isn't it?

[tggzzz]

Just a quick question: surely this is already a couple of orders of magnitude better than a multimeter can resolve, isn't it?

I have an 8 digit multimeter. That will be sufficient to spot changes. It hasn't been officially calibrated in decades, so I wouldn't want to rely on absolute measurements.

I also have several 6.5 digit multimeters from competent manufacturers who will have carefully selected the internal voltage references. They will probably be sufficient to spot changes.

If the OP has a known stable voltage reference and a decent KVD, then any ordinary multimeter would be able to spot changes. That's the basic concept of "null voltmeters".

[ivo3510]

Thanks, xvr,
At first I have put the reference chip in a DIL-socket, and the result was a bit worse than shown.
I have taken the idea of doing so from the multimeter construction itself, as I wanted to swat the refernces - I have couple of them in hand.
I assumed the contact resistance was changing with the temp, and I got rid of the socket.

[ivo3510]

Thanks, SteveThackery,
I believe the plot shows the uncertainty of both the reference, and the multimeter used to measure it.
As they both use the same chip, maybe I have good-enough reference now.
I will use this reference board to check multimeters with resolution below 6.5 digits, nothing critical.

[David Hess]

Do I need to swat the opamp with something better?

The OP177 is more than good enough, so there is not much to be gained with anything better.  Drift and noise are dominated by the reference.

The performance of the operational amplifier can be improved by buffering its output to remove load from its output.  This will improve load regulation and open loop gain.  The buffer could be a transistor in class-a, class-ab diamond buffer, or another faster and higher current operational amplifier.