[sdiy] Dual Frequency Shifter update

jhaible at debitel.net jhaible at debitel.net
Mon Jul 10 16:59:00 CEST 2006


On the weekend, I have made my first test with the
dual frequency shifter.

I was really blown away by the quality of stereo
frequency shifting. (It lasted a few minutes, then the fuse
in my Matrix FX was blown away - new bigger power transformer
will hopefully arrive tomorrow.)

It makes a difference when your input signal is stereo, and all
the room information is preserved, all while the pitch and
the jarmonic structure is mangled.

>From the few minutes listening to difficult material (voices,
plucked instruments) it sounds very, very clean.
In addition to the QVCO and trimmed AD633 shown on my Matrix FX
page, im using a pair of NE572 companders.

The other experiment, upshift -> fixed filter -> downshift
was interesting as well. Most notable: the Q factor of filters
changes dramatically with the shifting, along with the apparent
center or cutoff frequency change.
And I ran into a problem that I had expected, but underestimated:
The dome filters work from something like 15Hz to 15kHz. This is
enough for the first Frequency shifter (before the fixed filter)
which performs the upshift. But the second Frequency shifter 
sees the upshifted signal, must work with higher frequencies.
In the extreme case that the input signal was shifted up over
the whole audio range, we'd need a dome filter capable of 30kHz.
(15kHz audio band + 15kHz of possible upshift).
In my test, I only had an ordinary audio range dome filter.
This means I have a strong carrier bleedthru - which is even worse
when you consider in that case, the output signal is *not*
frequency shifted (just filtered - the two shift processes compensate!),
so you don't expect any alien noises.
The good news is that this can easily be cured. 30kHz, while
twice the bandwidth of the normal audio range, is also just one octave
more - and one octave more only need one extra all pass filter
stage (for a +/- 0.5 deg ripple 90deg dome filter).

So far my intermediate results.

JH.


-------------------------------------------------
debitel.net Webmail



More information about the Synth-diy mailing list