From nwilson@morgan.com Fri Jul 22 11:27:35 1994
Date: Thu, 21 Jul 1994 12:44:21 -0400
From: Ab Wilson <nwilson@morgan.com>
To: analogue@magnus.acs.ohio-state.edu
Subject: Re: A few oscillator sync q's

Malte Rogacki writes:

 > I KNOW what oscillator sync is. But could it be that there are different
 > types of it? I've heard the term "hard sync" a few times, does this mean
 > that there's a kind of "soft sync" and if so, what's the difference?

This is my understanding   of  it. BTW   this information  comes  from
questions asked in this newsgroup   about 8 months ago, and  twiddling
knobs of course.

Say OSC1 is the master and OSC2 is the  slave. If OSC2 is soft sync'ed
to OSC1 then OSC2 is forced to start a  new cycle every time OSC1 does
but can run freely in between. 

Now say freq(OSC2) = n * freq(OSC1).   (Does everybody understand that
little bit of  notation?)  If n is  an integer OSC2 will sound exactly
the same as if it wasn't sync'ed. This is because the sync pulses from
OSC1 will cause OSC2 to start it's cycle at the same  time as it would
have done anyway.  However if n  is not an integer  then the output of
OSC2 will be truncated at every sync pulse  and you'll be able to hear
the difference.  For n > 1 you get the sweeping harmonic type effects.
For n  much   bigger than  3 you  probably   won't hear  much  of  the
fundamental (freq(OSC1)).

If  wave(OSC2) = sawtooth  (more  crap  notation  to save  me typing),
varying  n between 0 and  1 will produce  a  sawtooth with a DC offset
that  will  get bigger  as n decreases.  If  wave(OSC2) = square, then
varying n between 0.5 and  1 will have the  same effect as varying the
pulse width. For wave(OSC2) = square and n < 0.5 (assuming OSC2 has an
even mark/space ratio) the output will be DC  since OSC2 will not have
chance to go negative before being reset again.

Now  say we have the  same setup except OSC2   is hard sync'ed to OSC1
then OSC2   produces exactly one cycle   every time  OSC1  starts it's
cycle. Note OSC2 does not run freely between sync pulses.

For n < 1 the effect is exactly the same as for soft sync.

Now say wave(OSC2)  = square, then  for n >  1 you get a strange pulse
waveform which goes 1, -1, 0. The pulse gets shorter as n increases.

Both hard  sync and soft sync have  there applications. Hard sync lets
you fuck about more  with the frequency of OSC2  before the  effect is
felt in the percieved pitch  of the sound  but the effect can be quite
subtle for large n. Soft sync  OTOH is more  noticable for large n but
you do tend  to loose the  fundamental. Soft sync'ed oscillators sound
better when mixed in with the master osc. Hard sync'ed sound better on
there own (IMHO).

FYI SEMs have both hard  and soft sync  (as well as amazing filters of
course).

 > How is oscillator sync implemented technically in analog circuits?

Beats me. To quote from the film Kelly's Heros, `I  only drive them, I
don't know what makes them go.'

	Ab.

--
