[sdiy] VCO jitter, Slop ...

Magnus Danielson magnus at rubidium.dyndns.org
Mon Jun 10 23:20:26 CEST 2013


On 06/10/2013 10:15 PM, rsdio at sounds.wa.com wrote:
> I disagree with your hypothesis (Michael?), that analog VCOs sound better or
> fatter due to jitter. My hypothesis would be exactly the opposite. Based on
> many other circuits, I'd say that digital clocks, oscillators, and PLLs are far
> more susceptible to jitter than analog VCOs and PLLs. In fact, the fix for
> jitter in a digital system is to create an analog PLL with a presumably analog
> VCO to smooth out the jitter of the digital, crystal-based clock.

While I like analog PLLs, digital PLLs can beat many analog PLLs in 
stabilizing things. It varies a lot with the specifics which is the best 
for each application.

> Jitter involves fast changes in period or frequency, and generally makes audio
> sound harsher, and possibly thinner. Drift is a much slower change in pitch,
> due to changes in resistor values or silicon gain as temperature changes. There
> are surely other flaws in a VCO besides drift, such as power supply noise or
> waveform distortion, but I would expect jitter to be either totally nonexistent
> or if it's even there then it would be the last and smallest contribution to
> the "analog" sound.

I'm afraid you use a little different terminology than you should.

Jitter is indeed the high speed variations of phase, while wander is the 
slow speed variation. The main reason for their separation in 
telecommunication definitions is due to their different sources. There 
audio equivalents is wow and flutter, It is however phase modulations 
and we often speak about phase noise, which both are more better terms 
since it doesn't in it's definition make a separation in rate.

If we with jitter only means fast modulations, then I agree that jitter 
most probably isn't the cause of "fatness" or "warmth". However, I don't 
think phase modulation is written off.

Drift is however a systematic effect rather than the random effect of 
phase noise. Linear drift is the derivate of frequency, but drift does 
not have to be linear, and when you look closely it isn't.

Other form of phase modulations can be periodic modulations like 50 Hz 
with overtones etc, but that isn't part of the drift, as drift is a slow 
drift in frequency, as in a constant movement.

Slow phase modulation can most probably cause the same psycho-acoustic 
sensation as slow phase-noise.

The trouble here is that there is a whole bunch of terms being used when 
characterizing oscillators, and the concepts does not seem to be widespread.

> If anything, it's the lack of jitter in an analog system that probably makes it
> sound fatter than a DCO. If jitter made a VCO sound better, then we could
> fatten up any CD by letting the jitter run out of control.
> In other words, I agree with David Dixon that this sounds bull unless someone
> can actually show jitter happening in a VCO, or even show a DCO with no jitter.
> Not picking on you, Michael, since you said you can't prove your hypothesis. I
> just suggest that people try to come up with a test for your hypothesis rather
> than just ramble on without any basis.

So, what about phase-noise or at least phase modulations, most probably 
slow rate? Is that as easy to throw out?

Oh, BTW. Most pink noise generators isn't 1/f noise at low rates, as the 
approximation filters is rarely made to handle very low frequencies. 
They are approximations between a lower and an upper frequency, but 
between these they can be very good. There is some really good papers on 
that from the 60thies and 70thies with modern studies in the 90thies.


More information about the Synth-diy mailing list