Calibration standards: AC voltage

Calibrating an instrument that indicates AC voltage is a surprisingly big challenge for various reasons. If the instruments is geared towards RF frequencies (such as an RF Milliwatt meter), the standard of choice should be the RF level generator presented elsewhere on this website.

Should the instrument however be a regular AC voltmeter as it comes as part of every multimeter, it is necessary to first clearly understand the parameters involved in the measurement.

Effective and Peak Voltages, RMS and non-RMS voltmeters

The voltage displayed by any standard AC voltmeter is always "effective" voltage. What that means is that it will have exactly the same heating effect on a resistor than the same DC voltage.

In other words: if a voltage of 10V DC heats a resistor from freezing point to a hundred degrees, the exact same thing will happen if you apply 10V AC (eff).

Unfortunately though, it is not easy to measure this. There exist very expensive instruments that basically measure the heating effect that a voltage has and thereby provide an accurate AC voltage value. Those will certainly not be available in an amateur environment.

How regular multimeters work in comparison to this is by measuring the PEAK AC voltage of the applied waveform (the highest voltage that the waveform reaches) and from that they calculate the effective voltage. Unfortunately again, that is not trivial. For a completely clean sine wave, both voltages would be related by the factor 1.414 (root of 2).

For other wave forms this factor differs, for completely random waveforms, the factor is not predictable.

This is why cheap AC voltmeters can only provide accurate readings for sine-shaped waveforms. More expensive voltmeters can to a degree provide measurements for other waveforms too (called RMS voltmeters) but even there limitations exist as to for what frequencies this can be accomplished (usually 50Hz only).

How to calibrate?

The method I found to work for me is to use a sine function generator to generate the cleanest 50Hz sine shaped signal I can provide, then measure it's peak voltage with a peak detector circuit in parallel to the voltmeter and derive the effective voltage by dividing the peak value by 1.414

The simplest peak detector circuit consists of a diode and a capacitor. The only problem here is that the diode creates a voltage drop (0.2 to 0.6v, depending on diode type) which needs to be factored into the calculation.
Peak1.jpg

A better peak detector

A more precise way of detecting the peak voltage is by compensating the diode forward voltage drop with an operational amplifier.

Operational amplifiers however come with an internal error called the "offset error" which must be compensated prior to measurement. Just short the instrument input leads and adjust the trimmer to zero voltage on the output.
Peak2.jpg

Further reading

I just ran into this interesting article from 1976 HAM RADIO magazine which very nicely describes the first method mentioned above.

W7KQ_AC_Calibration.pdf

PDF Article from HAM RADIO magazine 1976
pdf, 420,4K, 07/11/10, 349 downloads