[sdiy] Waveforms (was dx, chorus and Spock)

[Scott Gravenhorst]

I would expect that if the waves, regardless of shape are in fact 
*identical* as one moves through time that the spectral peaks would not be 
smeared or jitter in any way.  Jitter (or smearing as it would appear on a 
spectral analyzer) would be, as you said, caused by something the likes of 
say noise in the pitch CV.  I'd suspect that the higher the level of noise, 
the more smearing.  Actually what you would probably see on a spectral 
analyzer is a harmonic peak keeps moving randomly in frequency (assuming 
noise mixed with pitch CV).  Or imagine a patch where you have a perfectly 
stable pitch CV modulated by an LFO.  Here you'd see the harmonics in a 
ballet all moving right and then left again, but keeping an integral 
relationship to each other at all times.

So it seems to me that one important thing about precise and stable 
harmonics (as well as fundamental) would be super clean, noise free pitch 
control voltage and not a perfectly straight ramp of a sawtooth, as long as 
the non-straightness of that ramp is absolutely consistent for each cycle.

Looking at this from the opposite end, I once wrote a VB program that had 20 
sliders.  The top 10 controlled the amplitude of the first 10 integral 
harmonics and the other 10 controlled the phase of the harmonics.  Clicking 
the 'graphit' button would cause the program to compute and display one 
cycle of the waveform which the settings of phase and amplitude defined.  
Using the program was quite educational for me as I noticed it didn't take 
much nudgeing at all to cause rather obvious and grotesque changes in the 
waveshape, even if the changes were very tiny.  Of note actually were that 
slight changes in phase caused lumps and warts far more that I would have 
expected.

I still have the source for that if anyone is interested.

Gene Stopp <gene at ixiacom.com> wrote:
>Good info on the VCO cores, Terry! I think my question now becomes:
>
>If the input waveshape is "textbook perfect", when a resonant VCF is
>sweeping through the harmonic series, does it say "nice strong harmonic
>here, OK, moving along, hmmmm nothing here, oops, here's another nice strong
>one, moving along, nope nothing here..."
>
>In other words are the spectral spikes sharp and distinct and highly
>differentiated, and if the waveform is somehow warped they are "smeared" or
>"blurred"? I do realize that if every cycle of the waveform is identical to
>the previous one and the following one, then the harmonics should fall on
>integer ratios only (right?). Spectral garbage between the integer
>boundaries would imply some kind of intermodulation or jitter or some other
>thing happening at a frequency other than that of the waveform in question,
>right?
>
>Perhaps the spikes in the spectrum of a textbook waveform are in "pleasing"
>places at "pleasing" amplitudes. Or, perhaps the amplitude of the waveform
>as it enters the filter has a "sweet spot", not so low that the signal chain
>is noisey, and not so high that the blurring effect starts to happen. Maybe
>the filter with resonance needs a *huge* amount of headroom in which to
>perfectly pick out the harmonics. A much huger amount than traditionally
>found in most synthesizers, with the knobs set to "normal" positions.
>Certainly the state-variable hi-Q filters tend to squeal when driven hard!
>See: Korg MS-20.
>
>And yes, I do have a tendency to crank the minimoog mixer inputs to 10 and
>adjust the master volume instead. Part of the sound. Sometimes I even run
>the low output back into the external input and peg the overload lamp on
>solid. Maybe I should spend some time being nice to the filter and see how
>the harmonic-picking sounds?
>
>Hey - I wonder what a good spectrum analyzer goes for on ebay. Does anybody
>know of a good make/model with a comfortable audio-range display?
>
>- Gene
>
>
>-----Original Message-----
>From: Terry Michaels [mailto:104065.2340 at compuserve.com]
>Sent: Friday, August 16, 2002 11:13 AM
>To: Gene Stopp; synth
>Subject: [sdiy] Waveforms (was dx, chorus and Spock)
>
>
>Message text written by Gene Stopp
>> ...I wonder if the 921 waveforms
>> are just more "ideal"? Or was it some kind of "warping"? Or was it that
>in
>> combination with the 904A? And then the 902's? Or the fact that the
>output
>> went straight to the PA? I'm sure it was a combination of many things.<
>
>Hi Gene:
>
>The Moog 921 and the Minimoog oscillators both use a similar design, which
>has the collector of the expo transistor charging a capacitor, the
>capacitor is discharged by a transistor in parallel with the cap.  As the
>capacitor charges, and the sawtooth ramp rises, the collector current
>decreases slightly, because the collector to base voltage is decreasing. 
>This is due to the Early effect.  The result is the sawtooth ramp is not
>exactly a straight line, it flattens slightly as it nears the top.  This is
>plainly visible on an oscilloscope.  OTOH,  your ASM-1 VCO, and my original
>VCO design as published in Electronotes, and probably some others, all  use
>an integrator that is reset by a FET.  Because the collector voltage of the
>expo transistor in this design doesn't change as the sawtooth ramps up, the
>charge rate is constant, and the sawtooth ramp is straight as a ruler.
>
>The Moog 901 uses a true constant current source to charge the cap, so the
>ramp from that one should be pretty straight.  I don't have one on hand to
>verify that.
>
>I would guess the slightly warped sawtooth will sound a little different
>than the perfectly straight sawtooth to a discriminating ear, although I'm
>guessing the difference will be subtle.  The flattened sawtooth will
>probably have slight differences in the amplitudes of its various harmonics
>compared to a "textbook" perfect sawtooth.   It would be interesting to see
>a high resolution spectral analysis of all of these well known VCO's of the
>past.
>
>Also, the triangular and sine waveforms in these VCO's are derived from the
>sawtooth, so a warped sawtooth ramp will cause the triangular waveform to
>be slightly warped also, and the sine will have some additional distortion.
>  Again, the difference will probably be audible, but subtle.
>
>Terry Michaels

=========================================================  
- What good are laws that only lawyers understand?
- Government: The other religion.
- The media's credibility should always be questioned.
- Lambs who lie down with lions are lunch.

-- Scott Gravenhorst | LegoManiac / Lego Trains / RIS 1.5
-- Linux Rex         | RedWebMail by RedStarWare
-- FatMan: home1.GTE.NET/res0658s/FatMan/
-- NonFatMan: home1.GTE.NET/res0658s/electronics/