noise on the power rails as a cure for locking VCOs ??

[Magnus Danielson]

>>>>> "HJ" == Haible Juergen <Juergen.Haible at nbgm.siemens.de> writes:

Hi Juergen!

 HJ> Ok - before someone sues me for creating a myth, note the two question
 HJ> marks in the subject line.

 HJ> I don't know if it's true, and it may be pure speculation, but I'd like you
 HJ> to think about it and tell me what you think.

 HJ> I spoke to some very experienced service tech recently, and after exchanging
 HJ> the usual vintage synth trivia he mentioned casually that the Minimoog's
 HJ> VCOs simply do not lock to each other because of the noise on the power
 HJ> supply.

 HJ> I never heard of this before, but it sounds very resonable.

 HJ> With several VCOs physically close together, you always fight against a tiny
 HJ> amount of soft sync that limits the minimum beat rate between oscilators.
 HJ> This unwanted coupling takes place over GND and supply connections, 
 HJ> little spikes caused by the discharge of one VCO, wandering along the lines
 HJ> and reseting another VCO as well, if it happens to be so close to its own
 HJ> threshold voltage that the minimum spike lifts it over the threshold.
 HJ> Common means to minimize it is capacitive bypassing of each VCO, or even
 HJ> having separate supply voltage cables from the PSU to each VCO.

 HJ> I always wondered how the Minimoog could sound so "rich", though there
 HJ> are several different voltage rails that are shared by all 3 oscillators.

 HJ> Maybe some amount of noise on the supply rails "masks" the spikes ?
 HJ> Dithering comes to mind ...
 HJ> does this all sound reasonable?

This makes kind of sense... the noise would move the soft sync around,
infact the noise would itself modulate the oscillators. This noise
will however largely correlate between the oscillators, but it will
affect all reset points and not only those where two oscillators is
near-by each other.

 HJ> Perhaps I must withdraw my statement that 78xx and other noisy stuff
 HJ> would be no good for serious use.

Eh... noisy powersupply is one thing, but wait a minute... this noise
will stick into the audiopath here and there... why not keep noise of
the powersupply but more explicitly introduce noise (and other
signals) to the reset point instead... and now you can vary it.

Also, studying PLLs and similar constructs have shown that noise
(jitter noise in the PLL case) can actually help stabilizing the loop
under some circumstances, this is BTW true for a class of control
loops for which a ramp-oscillator is just one small subclass.

Thinking about dithering isn't a bad analogy (Tihi!) after all...

Cheers,
Magnus