How to reduce noise when measuring between voltage references?

[Victorman222]

I'm trying to measure between two zener vrefs using my newly acquired HP3478A to characterize one diode by varying its current to observe change to tempco in my temperature range.
 
Im a bit worried by the last 2 digits (1uV and 100nV) of the 5.5 digit reading on the 30mV range are completely random. My setup is coax cable going into banana plugs into dmm and soldered to opamp output on the other side. Im trying to figure out where this noise is coming from but having no luck. I tried replacing 2 cables by one coax, soldering directly to op amps, twisting the long(in order to put into thermal chamber) second diode cable.

Your advice would be greatly appreciated!

[MasterT]

Try noise reduction, simple RC 10k + 1000 uF (Low Leak !). 

[Mickle T.]

The OP37 (KR140UD26A) is not unity-gain stable and can oscillate in this configuration.

[Andreas]

Hello,

can be too many reasons in your case:

- OP37 is not unity gain stable so the cirquit may oscillate (try a OP27 instead?)
- Same would apply for too much capacitive loading of the OP-Amp (coax cable length?)
  you may want to have a cirquit with capacitive isolation

- integration time of DMM too short?
  but even with 100 NPLC on my 34401A I have 0.3-0.5 uVpp noise in 100 mV range
  and with the K2000 and 10 NPLC it is around 0.7-1 uVpp.

- missing isolation from air drafts around the zeners.

- EMI-Sources like mobile phone, WIFI, USB cables near the setup?

- I would put all the setup at least on a metal plate against EMI (connected short to the housing of the DMM).
  (better would be to put the whole setup in a tin can)

- and simply the noise of the zener could be also high.
  with a typical zener in a 10 Hz bandwidth you already have 4-6 uVpp on the 10V output.
  Of course the integration time (reducing the bandwidth) of the DMM would filter some of this noise out during measurement.

with best regards

Andreas

[jonpaul]

No surprise on this very old meter, perhaps out of calibration?

Suggest Tests...

1. .place gold shorting plugs, see noise, digits for zero input

2. Use standard cell, or any fresh battery eg 1.5v AA, D cell, wire up direct to the gold plated banana plugs, with copper wire, see digits and stability

Check meter specs for ENOB, accuracy, résolution, noise.

Jon

[Victorman222]

Thank you very much for the sound advice. I measured opamp output and yes it was oscillating  . Will save up and get something like OP07.