[Argus25]

It seems really odd that the designer wasted the opportunity to connect the unused 1/2 of the LM358 to the meter ground leg to improve the meter balance and stability, still in practice might not make a huge difference, but its just sitting there wasted.

from Mr Bungle post #5 :

"I played with that idea a few months ago and it seems impossible to measure things like this without having some kind of detrimental effect on the circuit".

Never a better word spoken.

This is really a macroscopic version of the uncertainty principle. Many don't realize that when you introduce test apparatus to a system, you modify the very system you are trying to measure by loading.

It is actually extremely difficult to get the loading low. I did some experiments a while back including things like loading a VFO with 6 cmos buffer gates in series, but with the output of the final one shorted, or not, it still pulled the VFO. In the end I had to resort to an optically coupled buffer for excellent isolation.

So you have to be careful with any test instrument-probe, that the loading is low, but the higher the input impedance, the more difficult stray capacitance is and frequency compensation issues. At least Tek were the masters of this dilemma with the input circuits they created for scopes, so there are some good lessons there to look at their input circuitry, if you are planning to make an RF milli-volt meter, or similar.