HFT

[Grant Richter]

Here at Wiard we have built 35 VCOs so far.
The design uses the Franco scheme. This was
chosen deliberately because of the waveform
distortion imitating acoustic effects.

The scheme is also temperature stable.

The hysteric comparator is a NJM318 with 70 V/us slew
rate. The exponential converter is a SSM2220 transistor
pair with a rated 0.3 ohm Rbe.

It takes anywhere between 0 to 1600 ohms in series
with the timing cap to set C8 within +/- 1 cent of reference
on a Korg MT-1200 tuner.

Replacing the 318 makes no difference. The converter pairs
are purely responsible, and they exhibit different non-linearities
over the C4 to C8 range such as over on one octave under
the next over again (i.e. "S" shaped non-linearity).

Still, a decent converter pair should give you less than
5 cents total error over an eight octave range
(with compensation).

----------
> From: Magnus Danielson <cfmd at swipnet.se>
> To: ijfritz at earthlink.net
> Cc: oakley at techrepairs.freeserve.co.uk; synth-diy at mailhost.bpa.nl;
uzs159 at uni-bonn.de
> Subject: Re: HFT
> Date: Thursday, December 30, 1999 4:53 PM
> 
> From: "Ian Fritz" <ijfritz at earthlink.net>
> Subject: Re: HFT
> Date: Thu, 30 Dec 1999 11:11:37 -0700
> 
> > Hi Folks --
> > 
> > Just to make sure this is clear, the diode plus resistor scheme has
been
> > around for many years. I picked it up from Hal Chamberlin's book, but
it has
> > appeared several other places alse. I can't find the original
reference, so
> > I'm not sure who did it originally. Franco? Hemseth? Maybe someone has
a
> > reference? Its purpose is to compensate for the emitter bulk resistance
of
> > the converter transistor, and the detailed analysis shows that it does
so
> > exactly, provided that the temperature dependence of the diode and the
> > transistor base-emitter junction are the same. I recommend not leaving
the
> > diode out.
> 
> But this is why we tend to use matched transistors with tigth temperature
> connectivity, to remove bulk resistance errors out of the exponential
> properties. But then, there are surely many ways to acheive the same
effect.
> 
> > At one time I experimented with different tracking schemes for the SAW
VCO
> > and got the best results combining both the Rbulk correction and the
reset
> > time correction (small resistor in series with the timing cap).  You
can see
> > that the design on my site has both!
> 
> Ah. We are still talking about two diffrent issues then.
> 
> I thougth someone had gone of and invented some way of adjusting the
reset
> timing error by sligth reengineering of the expo stage. While I am sure
that
> approximations can be acheived I'd still have to be convinced (I'd love
to see
> such a thing!) that you can do it and acheive the same wide frequency
> correction while keeping on scale as the resistor trick does. Fixing the
> expo stage does have the advantage of not breaking the waveform shape and
its
> DC offset and this is improtant stuff for cases where you use your
sawtooths
> trough waveshapers to acheive good canceling etc.
> 
> Cheers,
> Magnus