[net.music.synth] AM Synthesis theory, AX80, JX8P

knudsen@ihwpt.UUCP (mike knudsen) (01/10/86)

About cross-modulators (X-Mods), which were standard synth modules
back in the Moog days under the name "ring modulators" (after
their circuit implementation, a ring of 4 diodes.
These do for AM what Chowning and Yamaha did for FM:

A cross-modulator is a multiplier, which ideally can
be set up as either straight AM (meaning one of the oscillators
bleeds thru or is mixed with the product) or as "double sideband"
meaning that only the pure product is passed.
Actually, there are four possibilities for bleed-thru: OSC-1,
OSC-2, NONE, or BOTH.  I believe the Roland JX8P has more of these
options than the AX80, but all you really need are NONE and OSC-1.
(Another observed option, SYNC, merely seems to phase-lock the OSC's.)

In the NONE mode, feeding in two sinewaves will output ONLY
the sum and difference frequencies.  If the freqs are F and N*F,
N an integer, you'll get out (N-1)*F and (N+1)*F.
With OSC-1 bleedthru, you'll also keep the fundamental, F.
This is handy for getting odd harmonics (for clarinets,
French baroque organs).
With non-integer ratios you get non-harmonics (bells and other
weird effects on the old Moogs).

Example: You want the fundamental, plus the 3rd and 5th harmonics,
to get a very pure, hollow clarinet/rohrfloete sound.
(Traditionally got by filtering a square wave.)
Set OSC-1 for 16' (call this "F"), OSC-2 for 4' (4*F),
and bleed thru OSC-1.  You will get F (bleed), 3*F (4-1),
and 5*F (4+1).  Try it, you'll like it.
Next, detune the 4' oscillator up a 5th, to 6*F;
you'll get harmonics 5 and 7 instead of 3 and 5.
Finally, drop the 4' to 8' but keep the 5th (3*F);
you'll get F, 2*F, and 4*F, or equivalent of 16', 8', and 4'
with only two OSCs.  A clean, basic organ sound.
(Yes, the AX80 OSC-1 has a sub-osc.  Don't confuse me with facts.)

This backwards arithemetic (using odd harmonic inputs to get even
outputs and vice-versa) is similar to that of FM synthesis.

Of course, AM modulation works with any waveform, giving an
output with all possible sums and differences of all the
two inputs (set-theoretic product of two sets).
This opens up all kinds of possiblities...

I think the AX80 lacks a sine wave, tho the JX8P has it.
Roland's literature brags about "AM synthesis", which for
once is an accurate piece of advertising.  Many JX8P patches
rely heavily on the X-Mod.

The JX8P and AX80 have separate VCAs for each OSC.  Be interesting to
find out where they are placed when X-Mod is used --
is OSC-2 enveloped before the modulator?  Is OSC-1's envelope VCA
then applied to the modulator output?
Here's a chance for owners to find out how complete their
manuals really are!

DISCLAIMER:  I just got a Korg DW-8000, which lacks an X-mod,
relying on its prestored digital waveforms -- shucks!
This posting is based on my fooling around in stores with the
other two synths, and my theoretical knowledge of EE and Ham Radio,
plus the FM synthesis program I wrote for my Color Computer
(realtime joystick volume and timbre control).
Flames accepted AFTER you try my suggestions on your own synth
and prove I'm all wet!

	-- mike k

PS: I second the requests for more general purpose
(brand-name-independent) patch ideas.  Here now, I just gave
out some ideas I can't even use myself!
What did YOU get for Christmas?