alt.sci.physics.acoustics

In article <13hlfig86n3t9b7@corp.supernews.com>,
Richard Crowley wrote:

>I fear that you're completely missing the point. The OP (deluded
>though he may be) is asking for a circuit to take any four arbitrary
>waveforms and "ADD" their frequencies. (As in the example of
>taking four 1KHz waveforms and ending up with 4KHz)
>
>The case of taking the harmonic and filtering for the desired
>frequency is trivial and not applicable to the question as
>posed by the OP. Note that it is also not arbitrary since
>you have to have pre-knowledge of the desired output
>frequency in order to design/tune the filter, etc.
>
>To explain the question better, lets take some different
>starting frequencies. Lets say 697Hz, 770Hz, 852Hz,
>and 941Hz and produce 3260Hz.

OK... I think I understand the desired goal better now. I didn't see
Radium's original query.

With those constraints, I don't know of any straight-forward way to
directly generate the desired output frequency, since heterodyning and
filtering out the other (undesired) mixing products doesn't satisfy
the problem description.

There are some ways of getting fairly close, I think, with the use of
some circuits which are available off-the-shelf. Whether the result
would be sufficiently accurate, or sufficiently analog, or would cover
a sufficiently wide range of frequencies in any single implementation
to be considered to be an acceptable answer to the problem would be up
to our radioactive friend.

If the incoming waveforms are sufficiently close to being sinusoidal,
one could use analog comparator/latch/oneshot circuits to implement a
frequency-to-voltage converter (or, perhaps more directly, a
frequency-to-delivered-charge converter). The amount of charge
flowing into/out of each such converter would be linearly proportional
to the frequency of the signal. Feed each input signal to one such
converter, and sum up the 4 charge-pulse chains into an accumulator (a
capacitor).

Set up another such converter, whose charge pulses are of the opposite
sign, and feed its outputs into the same capacitor.

Set up a VCO, and an op-amp integrator, with the integrator input
coming from the charge-accumulator capacitor, the integrator output
controlling the VCO, and the VCO output driving the final
(opposite-sign) zero-crossing charge pulser and also being the desired
output.

When properly calibrated (the charge-delivered-per-zero-crossing needs
to be matched quite exactly), the frequency of the VCO output ought to
be the sum of the frequencies of the four incoming signals, so that
the number of positive-going charge pulses from the inputs exactly
equal the number of negative-going pulses from the VCO. In effect,
it's something akin to an all-analog PLL with delusions of grandeur
:-)

It won't be exact or precise, it won't handle input or output
frequencies whose half-period exceeds the width of the charge pulses,
it won't work at all if the incoming waveforms are in fact "arbitrary"
rather than close to sinusoidal (in which case they don't really have
a single well-defined "frequency", of course) and it probably has
eight or nine other significant technical limitations.

--
Dave Platt AE6EO
Friends of Jade Warrior home page: http://www.radagast.org/jade-warrior
I do _not_ wish to receive unsolicited commercial email, and I will
boycott any company which has the gall to send me such ads!