"Steven Swift" wrote in message
...

Hey, don't worry about the rrs gripers. We've discussed detectors there
before, and we will again.



Take a look at RDH4 page 1073.

Even with extreme design precautions you can't get much better than about
2%
(see page 1080-1081).

This guy calculates 0.4% distortion at 50% modulation and 1.6% distortion at
100% modulation.

http://www.amwindow.org/tech/htm/diodedistortion.htm

In a typical diode detector test circuit he measured 0.32% distortion at 50%
modulation and 2% distortion at 100% modulation.



To work, a diode detector has to "cut-off" for 1/2 the carrier cycle. This
requires that the diode always go through the "knee". A tube diode has a
voltage to the three-halves characteristic. A semiconductor diode has an
exponential characteries. Expand into a Taylor series, and look at the
first couple of terms. Distortion! Diodes as "multipliers" can be made
better.


How would a precision rectifier do?


I'll look for the analysis.

BTW, if you trickle current in a tube diode and keep the signal small, you
have built a "square-law" detector.

It's been my impression that the "knee" area of the curve is the
"square-law" area of the curve. Again, it's my impression that the crystal
set guys use foreward bias to get their detectors out of the "square-law"
area of the diode's curve in order to maximize sensitivity and minimize
distortion.


Not a diode detector. You can also see
this in the Taylor series.

Steve

Why isn't a forward biased diode not a diode detector?

Frank Dresser