frequency multiplier

[Martin Czech]

Another idea for multiplying frequency:

A sawtooth input wave is converted with a thermometer code ADC
in order to get equidistant time markers. 

This sounds horribly complicated and expensive, but it is not.  It
takes just a handfull of LF311 comparators and a 1% metall-film
resistor network (voltage divider) that creates equidistant reference
voltages for the comparators from a main reference. 

If the sawtoth wave rises one comparator after the other  will turn
on.  It is quite easy to determine the logical function for the
comparator outputs that will give a square wave with n-times the
frequency of the input saw.  With m comparators this will yield m+1
states and this will give a max. frequency of (m+1)/2. Modulating the
main reference for the resitor network will change pulsewidth and
frequency. This gives very interesting sounds, very close to some 
DX7 "FM" sounds (computer simulation).


The discharge slope of the sawtooth has of course a limited slew rate 
and this may cause some glitches in the output signal. Some ideas
to prevent this:

-The comparator outputs are open collector type and can be combined in
such a manner that disabling the one with the lowest reference will
also disable all others etc (thermometer code).

-The discharge signal of the oscillator (sync signal) could be used to
instantly reset of all final output signals for some time so that the
spikes are turned of before they can reach the output.

-Slow CMOS logic.



I had this circuit idea when I thought about modules like Buchla's
harmonics generator, and Technosauros undertone module. These modules
give harmonic/subharmonic triangle and /or sine waves. How do you
divide/multiply the frequency of these waveforms?

If the circuit described above will give some reasonable square signal
of 2x, 3x and nx the input, i.e. equidistant time markers, it should be
possible to control a polarity switch with one of these signals in such
a way that an incomming tri wave is either inverted or not inverted.
This will allready give the right slope of the wanted signal, but there
are jumps. If one adds some portion of the said square wave to this
signal you end up with a triangle signal of 2x, 3x,.. nx of initial
frequency.  This can be converted to sine by diode networks or other
methods.  Dividing works in a similar way.

Of course this way of harmonics/subharmonics generation will give some
spikes into the signal (when the sqare-waves toggle), I don't know if
it is possible to filter them to a reasonable level.  But I KNOW that a
spike right at the triangle peaks is not very good audible.  SIne waves
will require more attention to this point however.

I have not build such a circuit yet, but the switching time of the
comparators is about 100ns, this is 1000 times faster then the period
of a sawtooth wave @ 10 kHz, so the time resolution should be not such
a big problem.  15 comparators will give a 4-bit ADC, this should be
possible, even with 1% resistors. But the design will need precautions
like proper supply decoupling, grounding etc.

m.c.

------------------------------------------------------------------------
m.c. has made it finally:  3 CDs out now; 72 min. minimum; "1"
(1994-1995),"2" (95-96),"three" (96-97); experimental stuff; mostly
Eimert/Stockhausen style; but also modern popular style