[sdiy] jitter in oscillators for music purposes

[Magnus Danielson]

From: karl dalen <dalenkarl at yahoo.se>
Subject: Re: [sdiy] jitter in oscillators for music purposes
Date: Tue, 13 Jul 2004 15:36:21 +0200 (CEST)
Message-ID: <20040713133621.15219.qmail at web53508.mail.yahoo.com>

Karl,

> > > true, therefore it is not phase modulation allone.
> > > It's getting more complicated.
> > 
> > We see a horrilbe mix of phase modulation, frequency modulation and
> > amplitude modulation. The frequency modulation is really just a signal 
> > integrated before it contributes to the phase modulation. The total phase
> > modulation signal will contribute to the amplitude modulation.
> > 
> > The power supply DC part will affect the amplitude but not the frequency.
> 
> Are you absolutely shure on this? I would say it do affect frequency!

Ah, yes... let me explain.

The thing is, in a standard sawtooth oscillator (see ASM-1 VCO schematic for
reference) you have two internal dependencies of the positive voltage feed to
frequency, that of the current source into the expo circuit and that of the
comparator. The thing is that these cancel each other out. Very neat. When the
voltage is slightly higher than intended, then will the expo give a higher
current and the capacitor will charge quicker, but the reference comparator
will now have a higher threshold, so this cancels out. What we end up with is a
amplitude error. However, when we look at an AC signal on the powerline, then
cancelation is not perfect since the capacitor will integrate the AC signal,
so there will be a combination of phase and frequency modulation, which could
also be viewed as a phase modulated signal with filtered response.

> > The power supply AC part will contribute to both frequency, phase and
> > amplitude modulation. 
> > The modulation AC part will cause phase/frequency modulation (due to
> > the lower  corner for the integrator to actually act as an integrator)
> 
> And where would that lower corner be positioned, for lets say a TL082?
> What is this "lower corner" actually in a integrator?

Way down in frequency, but in essence the lack of infinit raw amplification
with the leakage from the loading of the cap by the op-amp (and parasitics) 
and internal leakage of the capacitor all aids in limiting the true responce
from that of an integrator. This is important to realize when discussing the
wander of VCOs, which very well can be measured for noise frequencies for very
slow phenomenas.

> > where the lower
> > frequency part is phase and the upper part is frequency modulation. 
> > The comparator noise AC part will cause phase and amplitude modulation.
> But only at ramp sample time, (ramp reset)! 

Yes, but I am in slow frequency mode here... but the picture is very complex
and things contribute at different times and compensates othertimes. Mixed mode
is a bitch.

> Let's say this would be an moog or for instance a korg core with the
> intgration cap as usuall to + refV or V supply then the cap leakage
> could play a part! but i think its so small it cant be heard!

Yes, but this is the point, we where discussing a model taking care of issues
otherwise overlooked. You just introduced an assumption which isn't very
helpfull for this discussion. 

> Then let's say this would be an moog core with the intgration
> cap to gnd would then gnd impedance variations be significant?
> (im about to say, yes).

I don't have those schematics at hand here and now and I don't have it in my
head either... but the point was to create one model, but this will not work
for all cases naturally. But let's get a deep understanding of one oscillator
at a time and then discover how we can apply it to other oscillators.

> > Simple, huh?
> 
> Yes..........no...........yes..............no.............!

Exactly my point... ;O)

Cheers,
Magnus