Does voltmeter's input offset voltage "leaks out"?

[dusan]

I need to measure and simulate circuit more accurately. So far I was simulating voltmeter as single resistor (variant A). However I was wondering if input offset voltage also "leaks out" somehow (variant B) or is it only internal? So I measured 2 voltmeters in parallel and I measured both to be zero (below 0.1mV I can measure). So does the voltage "leaks out"? How much?

[mk_]

So I measured 2 voltmeters in parallel and I measured both to be zero (below 0.1mV I can measure). So does the voltage "leaks out"? How much?

looks like the right MM is measuring AMPs and not Volts. And if not: measuring 100µV with such instruments is... hopeless.

[dusan]

Both are set to 200mV range. I don't know the offset voltage, just that minimal value is 0.1mV so if there is any it has to be below 0.1mV.

[dusan]

I came up with different measurement. I took 9V battery, connected unknown voltmeter in series with 10k resistor. I measured voltage on the 10k resistor with known 10Mohm voltmeter. I measured 117mV. Then I flipped unknown voltmeter leads and measured 135mV. I monte carlo'd it in simulator and found RUNKNOWN=704k and VUNKNOWN=642mV. This gives me exactly the same voltages across 10k resistor (117mV and 135mV) in simulator and in real life. But why can't I measure this offset directly? I should be able to measure 642mV with second multimeter when I connect them in parallel.

[WattsThat]

I think you’re confusing meter measurement uncertainty with what you appear to be calling “input offset voltage” which meters don’t have.

You assume both Chinesium meters are one megohm input impedance. That’s a bad assumption.

If you want help with the original problem, start there. As it is, you’re asking for help with your solution to solve your original problem. These are two very different things.

[dusan]

The test from reply #3 clearly shows that the voltage is there. If I flip the voltmeter value on second voltmeter (that measures voltage on 10k resistor) change, so the "offset voltage" is real.

I didn't assume both are 1Mohm input impedance. I assuming one is 1MOhm and other is 10MOhm.

[mqsaharan]

The thing you are looking for is usually termed as "input bias current".
Usually for desktop DMMs, manufacturers provide this information in their manual. But I haven't seen it in handheld DMMs' manuals except for Fluke 8060A. Attached here is the relevant page from 8060A's manual that contains some information about it.
Regarding how you can measure it, I don't know. But I am sure there are many members here who can help you in this regard.

[dusan]

I measured all ranges in both voltmeters if it helps someone figure out what is going on. Only the 200mV ranges are affected. Input impedance is smaller, they are asymmetrical with some extra voltage.

Voltmeter	Range	Vnormal	Vreversed	Rcalculated	Vcalculated	Note
PeakTech 2005A (blue)	200mV	332mV	61mV	448k	6.2V	Assymetrical with non-zero voltage
PeakTech 2005A (blue)	2V, 20V, 200V, 1000V	8.7mV	8.7mV	10.3M	0	Nice and symmetrical (zero voltage), this is what I expected
UNI-T M830BUZ (black)	200mV	117mV	135mV	704k	642mV	Assymetrical with non-zero voltage
UNI-T M830BUZ (black)	2V, 20V, 200V, 1000V	86mV	86mV	1.036M	0	Nice and symmetrical (zero voltage), this is what I expected

[Kleinstein]

In general a voltmeter will have both an offset voltage (that does not "leak out" or effect the input circuit) and some input bias current, that does also effect the external circuit.
With a 10 M input impedance meter the way to tell the 2 apart is by looking at the offset for an open input and with shorted input (take care of thermal EMF).
The test with the short will measure the offset voltage. The test with open and thus only the 10 M internal resistance will all the input biasing current times 10 M.

This would also effect the errors one would see with reversing a voltage. However the DMM may have additional (e.g. linearity) errors. In addition the input bias current may depend nonlinear on the voltage - so the model with a fixed bias current is only approximate.

With  0.1 mV resolution one may not really see the offset / bias current. There effect may be less then 1 LSB and thus hardly visible.


The difference shown look pretty large. Even for a cheap meter I would not expect that much error. This may be more an effect of superimposed hum or similar, that may lead to partial clipping - this can cause nasty errors and many meters (also good ones) don't detect it.

[dusan]

When I measure open and shorted leads of voltmeter they are both 0.0mV so it's less than 1LSB.

I am measuring above copper plate connected to - of the 9V battery. No mains connection. Away from anything. And the difference is still 300mV one way, 60mV other way.

[dusan]

Btw if I connect voltmeter parallel to ammeter the ammeter measures 0uA. Both ways.

[Kleinstein]

The expected range for the input bias is in the 10 pA range. So most normal ampmeters will not show anything. With 10 Mohm input resistance this may just get 100 µV, but possibly also less.
The input offst is often less than 1 LSB, maybe in the 10 µV range. Additional offset can come from thermal effects at the protection and the connectors.


For the test with the batteries, there is a chance that the battery volage may not be stable. With only 6.2 voltage left, chances are it is pretty dead and can change quite a bit.
The asymmetry shown looks more like a gross error, like hum, a loose connectio or dead battery.

[David Hess]

The input bias current is what leaks out, and if it is excessive, it will produce a reading on the multimeter across its 10 megohm input resistance when the inputs are not connected.

The input bias current is typically below 50 picoamps, but could be as low as 0.5 picoamps and lower for electrometer type instruments.