CEM3340 sync results

[Haible Juergen]

Hi,

last night I have finally made an A / B comparison between
the CEM3340 + external transistor sync and "normal" sync
of a saw VCO.

The 3340 circuit (VCO 1) can be found at
<http://www.it.kth.se/~e93_mda/synths/friends/haible/hj2vco.gif>.
The 6-position switch is in its most right position, selecting the
external pnp transistor that is connected to pin 10 of the
3340. This is very similar to what was suggested in the 3340
data sheet, and what is implemented in Rev. 3 Prophet 5,
OB-Xa, OB-8 and probably in some other synths, too.

The Saw VCO circuit (VCO2) is a clone of the Oberheim SEM VCO.
A picture of the original should be at the Synthfool page. In this circuit
the VCO's capacitor is discharged by an external sync pulse, just
as it would be in it's next own cycle.

I hooked up the two VCOs for identical frequency, and for identical
modulation depth (minimal beat rate, without sync). Then I sync'ed
both VCOs to the same (3rd) VCO and started to compare the
outputs of VCO 1 and VCO 1.

I compared them by ear, by cancellation, and by scope.

For comparison by ear, I applied various envelopes to both VCOs
at the same time, and switched between the VCO outputs.
I used long, manually triggered envelopes as well as short LFO
triggered ones.
At first, I thought there was no audible difference at all. After doing
the other tests, I heard a difference in sound, too, but it was
very minimal, and hard to decide which one would be "better".

For comparison by cancellation I inverted the output of VCO1
and connected the two VCO outputs to the two ends of a
potentiometer and monitored the signal at the wiper. I adjusted
the pot for best canellation of the two signals. Without modulation
(i.e. at the same frequency as the sync'ing 3rd VCO) I got almost
perfect cancellation. There was a minimal difference between the
two saw waves, because VCO2 had a tiny flat portion after the
capacitor discharge. The triangle-based 3340 doesn't have this, so
one could even say the 3340 has the more perfect saw wave. But
the difference was minimal, and I had to turn up the volume to
hear the difference signal at all.
Things changed with modulation applied. As I increased the
frequency of both VCOs simultanously, there was a noticable
difference signal. Still rather quiet, but not neglible. (I made sure
that this was not due to mistracking of the VCOs in the unsync'ed
case - otherwise I'd have different waveforms in the sync'ed case
even from two identical VCOs).

So there was clearly a small difference. I checked with the scope,
and I saw what I had seen before, but never spent too much thought
about it:
The 3340 + single transistor does indeed emulate the ordinary
saw-type sync function. BUT the level to which the capacitor
is discharged / charged does not exactly result in a "reset to
zero" for the saw case. In fact there is some "overshot" that looks
like a capacitor in a saw VCO would not be discharged to 0V,
but rather be slightly reverse charged to -200mV or so. On the
saw output of a a sync'ed 3340 you have several "normal" saw 
cycles, then one unfinished cycle as the sync pulse strikes
(just as in the normal saw VCO), then one normal cycle
with *slightly* larger than normal amplitude, and so on.

So here you have the difference. It's clearly there, but it's hardly
audible. And it's so far from the 3340's internal "hard sync" that
one can say that this one external transistor does an excellent
job to emulate "normal" hard sync.

How comes this difference ? I can only guess that it has to do with
the level to which the external pnp is tied. It's probably hard to hit 
the exact point of discharge asymptode to get a perfect "return to
zero" after the triangle-saw conversion. But I would not be surprised
if you could hit that exact point if you use a trimpot and an opamp
buffer to adjust the voltage to a slightly different voltage near -5V.
I have not tried this, and I won't, because for me the sync is close
enough, but I encourage you to try it if you want to know for sure.

JH.