I understand that in principle, a half-wave rectifier-based meter would read 45% of the RMS value. But that assumes ideal diodes. With a circuit using real diodes (1N34A), I get very poor linearity, which I assume is due to diode voltage drop. Is that correct? Since the scales of commercial analog meters seem pretty linear, how do they avoid the results I'm getting?

 

If they are powered, they likely use an op amp in an ideal rectifier configuration.

 

I have a very old multi-meter. Its low voltage AC scale is nonlinear near zero.

 

I understand that in principle, a half-wave rectifier-based meter would read 45% of the RMS value. But that assumes ideal diodes. With a circuit using real diodes (1N34A), I get very poor linearity, which I assume is due to diode voltage drop. Is that correct? Since the scales of commercial analog meters seem pretty linear, how do they avoid the results I'm getting?

A 1N34A is a point-contact germanium diode and the forward voltage drop is very temperature sensitive so you will have instability.

 

I have a circuit that I inherited. It uses a 1000V diode. The voltage is 0.75 volts less that it should be. I found a way to add a matching diode that adds 0.75V to the output. The diodes are next to each other. The temperature stability problem has gone away. The problem has not totally gone away but it is much better. Friday, I shipped board 1 and it has an offset of 0.03V. Board 2 has 11mV offset.