Soft sync and medicine to cure it.

[Magnus Danielson]

From: "Bjorn Julin" <bnillson at hotmail.com>
Subject: Soft sync and medicine to cure it.
Date: Sun, 10 Sep 2000 23:31:08 GMT

> Hi.

Hi there Björn,

> I made some measurements lately on softsync
> issues, by measuring the spikes bleeding trough
> to the common emitter of the expo convereter.

Yes, refering to the unintentional softsyncing between oscillators.

I think one has to consider both the effect of bleedthrough over the expo as
well as the voltage limit reference (not too uncommon with some resistor
divider over the power supply). It is interesting how both these effects can
work together.

If you take an VCO having a topology similar to the ASM-1 VCO for instance, you
can have the reference current resistor to balance out the error of the
reference voltage divider. This naturally requires them to see the same voltage
including unwanted power-ticks, so they should tap their voltage at basically
the same point.

> Depending on the tecknique used in the switching
> mecanism of the core (as you allready know i suppose)
> the spikes are quite different in its amplitude.

Certainly.

> Just by simple bypassing with a apropriate cap
> between the most pos and most neg rail you can
> reduce the spike significantly.

Oh yes, there are quite some spikes there, so you have to have a capacitor or
two around with extra-charge to drop in. I had the capacitor discharge on my
ASM-1 timed to be about 750 ns and the LM311 should be faster...

> To my suprice, by placing a FET in series with the
> sourcing current from the osc did not reduce the spikes?

Hum, how would that help?

> Juergen or  (others) ,did you got any improvment doing so
> in your's designs? And has any one done any soft sync measures?

Speaking only for myself, I have not done any measurements targeted towards
unintentional soft-sync from one oscillator to another. I have not taken any
special measures to avoid it, but I do have spent some time considering
generic sensitivity of power line noise. Consider an oscillator sync up with
gate/ADSR related transients, things like that.

I think the AH thread about the Bode frequency shifter which was unusable for
small frequency shift (like below 1 Hz) due to oscillators syncing to each
other relate to this discussion.

Cheers,
Magnus