[sdiy] FM Synth Designs

Theo t.hogers at home.nl
Tue Aug 13 16:51:18 CEST 2002

I know there are some pp on the list that studied the DX7 patent in more
detail than I did, but until they jump in I'll try my best.

The trick used in the DX is something like a*b=log(a)+log(b)
By making all values log, including the "sine table" the DX can use add
instead of multiply and there is only one log>lin conversion  needed to
generate the final output.
The log>lin conversion is done with a look up table, this might be in the
"rom"chip you describe, maybe together with the logarithmic representation
of the sine table? Combining those two on one ROM seems to make sense to me.
The alternate waveforms on the later models where made by reading the "sine
table" in different ways.
Therefore these units can still use the log+log trick and don't need
If the "sine table" is indeed on the second (rom) chip it should be possible
get the alternate waveforms on a DX7 or DX9 too by inserting some logic in
the addressing lines.

BTW There is a list of the alternate waveforms with spectrum diagrams in the
SY77 manual.


From: Colin Fraser <colin at colinfraser.com>

> ----- Original Message -----
> From: "Batz Goodfortune" <batzman at all-electric.com>
> To: "ben" <benj at iinet.net.au>; <synth-diy at dropmix.xs4all.nl>
> Sent: Tuesday, August 13, 2002 12:48 PM
> Subject: Re: [sdiy] FM Synth Designs
> > There is no outboard ROM chip if that's what you were thinking. Had
> > have been I would probably have been the first to throw in a rotary
> > and be able to select between a bank of different wave forms. The
> > wavetable/ROM is in the OPs chip. This chip is responsible for the time
> > slicing and assembling of the final output wave form and then spitting
> out.
> My understanding was that the DX7 implemented 'multiplication' to vary the
> amplitude of it's operators by adding two sine waves together with a phase
> difference proportional to the required amplitude, instead of acutally
> multiplying the wave data by the amplitude value.
> This would mean changing the waveforms wasn't viable.
> Later units with a choice of waveforms used hardware multiplication.
> Could be wrong...
> Colin f

More information about the Synth-diy mailing list