[sdiy] Ultrasonic VCO question for Ian Fritz (or whoever)

[Magnus Danielson]

From: Scott Bernardi <sbernardi at attbi.com>
Subject: [sdiy] Ultrasonic VCO question for Ian Fritz (or whoever)
Date: Mon, 28 Oct 2002 05:38:49 -0800

Scott,

> I direct this question to Ian Fritz, since you are the author of an
> Electronotes EN #112 article on a Ultrasonic VCO (built with a 3080,
> JFET buffer, and LM318 schmitt trigger). Anybody can answer.
> In the article, you say you compensate for the 570nS dead time with a HF
> track trimpot in series with the integrating cap. The Mikulic style
> sawtooth oscillators use the same scheme. You suggest a value of around
> 4K total.  What is the formula for calculation of this resistance?
> Doesn't seem to be t = RC,   R = t/C.

Well, it should be...

The Franco frequency compensation compensates the reset time by inserting a
series resistor R to the integrating capacitor C such that

t      = R * C
 reset

I've put a note on it on the ASM-1 homepage (http://home.swipnet.se/cfmd/synths/friends/stopp/)
if you care to read it.

I've also done the nitty gritty math to myself discover this phenomena.
I didn't know what that resistor did there, but I knew about the reset time.

What the resistor effectively do is to raise output of the integrator op-amp
linearly with frequency (since the current from the expo-pair is linear with
frequency) such that it auto-biases the waveform to reset earlier. Correctly
balanced, it removes the frequency error due to the static reset time
perfectly and thus, the acheived frequency is now linearly dependent to the
current on the input of the Current Controlled Oscillator.

If you look at the ASM-1 homepage you have enought hints to figure the rest out
in detail.

This form of frequency compensation is called Franco compensation by some, thus
relating to the work of Sergio Franco in the early 70this to the best of my
knowledge. I have not seen a good article about it and have not read any of the
Electronotes.

The compensation will distort the waveform for higher frequencies but trade
from waveform accurateness (which is bad to start with due to the reset slope,
I've seen it and it's not excessively nice) for better tracking abilities.
What can be a problem is that the compensation leads to a slightly lower AC
amplitude and raises the DC offset as the frequency raises and compensation
starts to kick in. IMHO, this is a fair trade anyway, since few depends on the
accurate waveform and spectrum distribution about 50 kHz or so.

Cheers,
Magnus