On the 20V range it will be 20MΩ or thereabouts.
I just measured the impedance of one of my multimeters (a Tektronix
DMM916). It reads either 10MΩ or 9MΩ depending on the voltage range.
On the 4V range it reads 11MΩ, on the 40, 400, and 1000V range it reads 10MΩ. These were read on my
HP973A. On the mV range it is also 10MΩ
Reading it around the other way I get similar values.
Whilst the impedance does change a little as the ranges change, the impedance does not change the same way that a simple analog meter does. In general, the impedance stated on the can for a DMM is the impedance on all ranges.
I have an old high voltage probe that says it is meant to be used with DMM with 22 Megohm internal impedence.
The probe designed for a 22MΩ meter, if used on a normal DMM will read low. How low it reads depends on whether the meter is used to read the voltage across a resistor, or if the meter's impedance forms the lower leg of a voltage divider.
If the latter, it will read 50% low (i.e. 10kV will read as 5kV -- possibly 5V if it's a 1000:1 probe). All you can do is double the meter's reading.
This is VERY unlikely to be the case as this design makes the probe unsafe. An open meter will have the voltage on the tip of the probe present at the input terminal. You probably have a big caution on the probe saying that the ground lead MUST ALWAYS BE CONNECTED. This is to ensure that the voltage on the probe output does not rise to dangerous levels. It's also likely that your probe has 4 connections, a ground clip, the probe tip, and the 2 connections to the meter.
If the former (and you can tell because there will be a resistance less than 22MΩ between the probe's meter input and ground). Whilst you can't do much about this either, knowing that the probe is nominally accurate when that resistance is shunted by 22MΩ, you can determine the effect of shunting it with 10MΩ.
For example, if it reads 680KΩ, then the meter expects the resistance of the lower leg to be 659KΩ (and presumably the upper lg of the voltage divider will be about 658.3MΩ.
When used with a 10MΩ meter, the lower leg has a resistance of 637kΩ. With this lower resistance, the probe will read about 3.5% low (10kV will read as about 9.66kV).
In this case, I probably wouldn't worry a great deal, but I would assume any errors are it reading on the low side.
However, you can also run a test. Assuming you're competent to read high voltages, find some voltage source (mains will be fine) that are reasonably high. Measure that with your meter directly -- lets say it reads 254V. Now read it with your high voltage probe. if it reads 0.254V, you're good to go. Anything else will indicate some error. If it reads 0.245, then you know you will have to multiply your readings by 1.035 (that's the 3.5% we calculated above) or just ignore the error.
Note that you will have to do your own measurements and calculations unless the lower leg of the probe just happens to be 680kΩ!
