Continue to Site

What is the minimum voltage change I can measure with a 6.5-digit multimeter?

strahd_von_zarovich

Full Member level 2
Full Member level 2
Joined
Sep 25, 2015
Messages
124
Helped
1
Reputation
2
Reaction score
3
Trophy points
1,298
Activity points
2,512
Hi everyone,

I am testing a very sensitive analog circuit, and I am unsure if I am measuring the values correctly.

The resolution of the multimeter is crucial for me. I need to measure changes of 100nV.

Can I do this with a 6.5-digit multimeter? My multimeter model is the Agilent 34411A. According to the datasheet, the resolution is given as 0.03 ppm in the 10V range and 100 nplc. So, I concluded that I can measure a voltage change of 30nV in 1V range. Is this correct?

Here is the relevant part of the datasheet:

R1.PNG
 
Thank you very much for the answers. Yes, I am asking about resolution. I collect data via PC, and the multimeter provides readings with much higher digits down to the 1 nV range. I'm not sure if these readings are accurate.
 
@strahd_von_zarovich, is your concern absolute accuracy or just resolution ? I see
prior post ultimate concern is accuracy ?




Regards, Dana.
 
@strahd_von_zarovich, is your concern absolute accuracy or just resolution ? I see
prior post ultimate concern is accuracy ?




Regards, Dana.
Dear Dana,

Just resolution. My main goal is to measure changes rather than absolute accuracy.

For example, if I measure 0.100,000,000 V and then increase the source voltage by 50 nV, I want to measure 0.100,000,050 V. -

As long as I can measure the change correctly, I don't really care about the actual values.
 
O.k., PC interface provides higher resolution. Presumed specified 0.03 ppm of FSK noise at PLC = 100 applies also to 1V DC range, then useable resolution of measurement depends on expected significance. According to user manual a resolution of 2*Vnoise,rms is obtained with 4.5 % prohability of failure, 3*Vnoise,rms is obtained with 0.3 %.
 
Hi,
Thank you very much for the answers. Yes, I am asking about resolution. I collect data via PC, and the multimeter provides readings with much higher digits down to the 1 nV range. I'm not sure if these readings are accurate.
from my experience: everything below 10uV becomes difficult.

Depending on range, frequency response, measurement timing ... and so on one could go quite below 1uV .. But I see no change to get below 100nV with reasonable effort.

For example, if I measure 0.100,000,000 V and then increase the source voltage by 50 nV, I want to measure 0.100,000,050 V. -
This is 0.5ppm? it´s difficult to detect from 0nV to 50nV ...(without 100mV offset) but in the rane you say .. even the REFERENCE and it´s drift plays a major role.
Noise, thermcouple effects, mechanical stress on a PCB ... all will cause more error than 50nV.

Again: it all depends on the complete specifications.

Klaus
 
Thanks for the answers.

I will be measuring in the range of 100mV to 130mV.

It seems I need to buy at least a 7.5-digit multimeter to verify the accuracy of my measurements.

What do you think about the 7.5-digit Keithley 2001 and 7510? It seems like the Keithley 7510 is better than the 2001 but much cheaper. Why is that?

Should I consider the 8.5-digit Keithley 2002 or the Keysight 3458A instead?
 
For noise work we had multiple Faraday-cage "boxes" around probe chuck or socket, and these had high-zoot LNAs inside the cage before any instruments. A well designed gain block would help your DMM "effective range" though fractional resolution "is what it is". Then a lot of oversampling to drive down ambient and instrument internal noise floor. Triax cables are the done thing for DC leakage and could "double shield" AC incoming EMI. Line sync can help consistency though not accuracy@sync-point necessarily (crossing vs crest). If not synced then be sure to accumulate across many line cycles at many points per, for "unbiased" averaging.
 
I will be measuring in the range of 100mV to 130mV.
There's a strategy of making an 'expanded scale meter'. Place an opposing voltage (or voltage drop) inline with your source. Say 90 mV. Then your readouts span a range 10mV to 40mV.

Or say your source is 1V. That's great enough that you might take a plain diode and install a resistive divider across it. Tap wherever it yields a usable range. A diode is its own voltage regulator. Or add another diode in series.
 
Hi,

I don´t see the "meter" as the biggest problem.
The wiring, the thermal drifts in a room, the manual handling, mechanical stress .. all will cause errors above the said expected accuracy.

Just for own verification you could do some tests:
* disconnect wires, reconnect them. Write down the measurement result. Repaet it 5 times. Check the variations
* connect wires, leave them connected. don´t touch anything. don´t move yourself. Just write down the meter readings of 5 consecutive minutes.. Check the variations.

If you put the results in an excel chart ... you can see whether the variations are "random" .. or "all follow one direction". Usually its a mixture.

Measurements in these low PPM regions need a lto of experience and equippment.
Like:
* a grounded (earthed) metal plate with all measurement devices connected to them.
* a antistatic plastics between the metal plate and your DUT
* a thermally controlled box (maybe even electrically and magnetically shielded to reduce external noise)
* time for the temperatures (and temperature related drifts) to settle.

Every voltmeter will drift, every power supply will drift, all devices with power dissipation (resistors, diodes, ICs, batteries, ...) will drift on self heating,
Even if you move . may have an impact (thermally and noise wise). So better do the measurements from a distance.

Klaus
 


Write your reply...

LaTeX Commands Quick-Menu:

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top