"Randy Yates" wrote in message


Arny,

This information is probably in this thread somewhere, but it has
gotten so long and convoluted that it's much easier just to ask: Are
you asking whether FM (Doppler) modulation at the high frequency is
the ONLY effect that results when that high frequency in addition to a
low frequency (purposely left undefined since the actual values depend
on a number of factors in the physical setup) are reproduced in the
same transducer, or is there some amount of AM modulation as well?

I'm not asking that question, because I know the answer, and I knew it
walking in the door last week.

The results of playing multiple tones through something as dirty as a
speaker produces copious amonts of both AM and FM. As a rule, the AM
dominates.

"The Ghost" gave me an idea for determining this without requiring any
measurement of the instantaneous cone displacement. Perform an FM
discrimination of the received (microphone) signal at the high
frequency "carrier." Call the discriminated signal m(t). Regenerate a
perfect FM signal using the modulating signal m(t) and subtract that
from the original signal. The result is the residual modulation on the
signal, which could then be AM-detected to determine if AM is present.

I've tried that, and a lot of other things. It has the usual problems with
nulling in the real world. You can get roughly a 2:1 to 10:1 reduction of
the unwanted distortion by that means.

Three practical issues which must be dealt with come to mind:

1) How to synchronize the regenerated FM carrier amplitude to the
original FM amplitude?

Pretty easy to do an fair job of in the digital domain.

Easy answer: emit a signal consisting of the
high frequency tone alone for a length of time adequate to measure the
amplitude.

If you've looked at the raw data page posted at
http://www.pcavtech.com/techtalk/doppler/ you'd know that finding that out
with pretty fair precision is a matter of reading numbers off a screen.

2) What modulation index, or depth of modulation, should be used in
the regenerated FM signal? Said another way, what gain (if any) should
be applied to m(t) when regenerating the FM signal?

At this point I should point out that since the AM dominates, it might make
sense to apply an AM signal to null the AM part out, leaving the FM.

3) How do you synchronize the regenerated signal in time with the
original signal? There are actually two synchronization tasks to be
done: phase synchronization of the carriers, and delay in the
modulating signal, i.e., tau in A*m(t-tau). (A is the parameter in
question 2).

Well, we know quite a bit about the signal that we are trying to clean up.

Does this make any sense?

Been there, done that. Seriously, I come back to this problem of separating
AM and FM from a real world signal every once and while, and learn a bit
more about solving it.

This time I realized that ideally, AM distortion related sidebands are
indepenendent of of the carrier frequency, but increase in amplitude with
carrier frequency for FM. Trouble is, this practical example is so heavily
dominated by the AM distortion. I hope to go back to studying jitter, and
play this card there.

I suspect that lots of people have been misidentifying AM distortion
products as jitter.