when does a MM interfere with what you want to measure?

[gaussmarkov]

when does a MM interfere with what you want to measure?  well, the answer is probably "always."    so i guess that i should ask how to learn how much a MM reading could be off from the corresponding value (volts or amps or resistance) in a circuit without the mm included as an active component.

cheers, gm

[Paul Perry (Frostwave)]

The problem is on the voltage ranges: in an ideal world, the input impedance of the multimeter would be infinite.
But, it's not...... it's pretty high, but it's not infinite.
So when you measure a voltage, you are also putting a high value resistor across what you are measuring. Usually it doesn't matter... but if you put it across a 5Meg resistor, then yes, it will probably interfere with the circuit.
Weirdest shit I've seen: a faulty circuit that only worked when a scope probe (1 Meg) was hanging across it (fixed the bias, until the probe was taken away)

Simple explanation: http://mrtmag.com/mag/radio_technically_speaking_test/index.html

Too much info: http://zone.ni.com/devzone/conceptd.nsf/webmain/64CA4B715425FA7D86256E0B0071FF1C

[R.G.]

In general, what is significant is guided by the rule of thumb that you can ignore things that are only 1/10 of the effect you're measuring.

That is not true in precision measurements, of course. Maybe it should be stated that the effect of the measurement device should be less than 1/10 of the precision you're trying to measure.

Multimeters once were not active. They were ... a meter movement ... and some divider resistors. Meters were denominated in terms of ohms per volt - so a 10K ohms per volt meter had a 100K input resistance on the 10V scale and a 1M input resistance on the 100V scale. 20K ohms per volt was a modest standard for well-thought-of meters. The meter movements tended to be either 0.1V or 1.0V full scale, so the resistance of the meter movement itself was on the order of 2K or 20K. The rest of the resistance in higher ranges was added by the divider resistor chains.

As you can see from this, measuring a low voltage range was NOT a high-impedance kind of thing, so everyone who knew what they were doing worried about meter impedance a lot.

If the meter was more than 10 times the source impedance of the voltage you were trying to measure, then you could to a first order ignore it. So on the 1V range, you knew that you could not measure accurately in circuits with internal impedances of under 2K ohms. You could calculate the right answer but that was done in your head because there were no computers (for ordinary people) nor pocket calculators.

Precise high impedance measurements were done with the VTVM (Vacuum Tube Volt Meter) or its successor the FVM (FET volt meter). This used a diffamp pair of triodes or FETS for the high input impedance buffering. These things always drifted, so you zeroed them before every measurement. Really high impedance measurements were done by means of measuring current between the meter and the voltage being tested, and adjusting an internal voltage source to force the current between the two to measure zero.

But I'm wandering.

You have to know the input resistance of your meter.

You can figure this out by setting up a voltage source like a battery, in series with a really high resistor, perhaps 10M or 22M, or multiples of those in series. Measure the battery voltage right at the battery, then at the end of the resistor string. If you happened to have the exact resistance in the string as the meter's input, the voltage will be half. But even if it's not half, you can calculate it by the voltage divider law.

Once you know your meter's input impedance, you know what impedance sources you can measure with some accuracy, and what the inaccuracy is, approximately.

[Dolly Parton]

When I have this problem with voltage measurements I wire up an op-amp as a follower, connect its + input to whatever is being measured (using a FET input amp) and take a reading from the amp's output; this way you avoid loading and get the true measurement.   Not always possible to do (or advisable), but just thought I'd mention it.

[gaussmarkov]

thank you, thank you.

great story Paul!    and helpful advice.  i am going across at least a 2.2M resistor in my experiment, so i will be more careful.

When I have this problem with voltage measurements I wire up an op-amp as a follower, connect its + input to whatever is being measured (using a FET input amp) and take a reading from the amp's output; this way you avoid loading and get the true measurement.   Not always possible to do (or advisable), but just thought I'd mention it.

cool.  it seems like a function on a mm should be designed this way. 

R.G., thanks a ton.  just what i needed.  a little background, a little theory, and a practical bottom line.  what more could one ask for?

[A.S.P.]

...So on the 1V range, you knew that you could not measure accurately in circuits with internal impedances of under 2K ohms...

under, or rather over?

[gez]

When I have this problem with voltage measurements I wire up an op-amp as a follower, connect its + input to whatever is being measured (using a FET input amp) and take a reading from the amp's output; this way you avoid loading and get the true measurement.   Not always possible to do (or advisable), but just thought I'd mention it.

cool.  it seems like a function on a mm should be designed this way. 

In an ideal world, but my mm goes up to 11 600V and there aren't many op-amps whose inputs would appreciate being connected up to that!  For our purposes though, it would be a nice feature...easy to roll your own though.