"Clifton T. Sharp Jr." wrote in message ...
John R. Strohm wrote:
"Clifton T. Sharp Jr." wrote...
(http://uweb.superlink.net/bhtongue/7diodeCv/7diodeCv.html).

In other words, the slope of the V/I curve
must change as a function of applied Voltage. The slope must be
steeper (or shallower) at higher voltages and shallower (or steeper)
at lower voltages than at the quiescent operating point.

I don't see how a changing slope
during forward conduction could do anything other than distort the
demodulated waveform, especially on tiny signals.

Think about the V-I curve on an ideal diode. Observe that, if the diode is
reverse-biased, no conduction takes place. In principle, you can crank the
reverse voltage up as high as you like, and STILL no conduction takes place.
(In practice, you run into avalanche and zener breakdown.) If the diode is
forward-biased, the diode conducts like crazy. There is a CORNER in the
curve, i.e. a change in the slope of the V/I curve, at NOMINALLY V=0V.
(Actually, V=0.7V for silicon, V=0.3V for germanium, V=1.7V for red LEDs,
about 2.2V for yellow, about 2.4V for green, and so on and so forth...)

I'm not sure I was clear on my question. The only nonlinearity I want in
my detector diode is the noncontinuous function

| Vin when Vin = 0.000000 micronanofemtoattovolts
Vout = |
| 0 when Vin 0.000000 micronanofemtoattovolts

...while the author seems to be saying that the nonlinearity which exists
just above the barrier voltage is essential to detection. He does say,

"The slope must be steeper (or shallower) at higher voltages and
shallower (or steeper) at lower voltages than at the quiescent
operating point."

Given my perfect diode and (let's say) a 0.25V peak signal, I contend
that my output envelope will be an exact reproduction of the input
envelope; but (see his graph #2) his required nonlinearity will seriously
distort the output envelope.

He seems to be saying that without that, you can't demodulate the signal.
I assert that without that, I demodulated it better than he did.

I think you and the author are saying the same thing. Consider that
in your ideal diode case, the quiescent operating point is zero volts.
Above that, the slope of your diode is infinte (zero ohms) and below
that it's zero (infinite ohms). The author is suggesting that you do
NOT need it to be so sharp, and in fact, I find that I can detect RF
down to below 100 microvolts or so with a Schottky diode designed for
zero-bias detector service, though the output voltage is very tiny
down there. If I could buy one of your perfect diodes, 100uV (RMS) of
RF would give me around 141uV of output, instead of the sub-microvolt
I do get. The low output is the result, of course, of the fact that
the diode dynamic resistance at -140uV is only very slightly different
from its dynamic resistance at +140uV.

In a real diode, the change in slope is gradual with no sharp corners,
but if you could find a diode which looked like, say, 1000 ohms in the
reverse direction for all voltages and 1000.1 ohms in the forward
direction, it would still work as a detector, albeit a poor one.

Cheers,
Tom