[sdiy] Hopf bifurcation VCOs....... my test results!

[karl dalen]

This is a really a cool cirquit!
You deserves DIY nobel price for publishing it Ian!
Why havent you done it earlier? :)

Heres my tests reults:
LM13700, batch 9642, LF353 intg, TL072 feedback.

I changed the cirguit gain a bit, 10K ,390 ohm OTA inputstage,
and 5,1 v zeners (i didnt have any other at the moment)
220k to + input at + feedback stage, otherwise all the same.

I also use one expo tied to both Iabc inputs, no reason to use separate expos.

Sinus has no dent!!!!!!!!!!!!! Absolutely Brilliant!
THD must be good!
(they are playing Maahler on the radio im getting stressed):)

I wanted to get rid of the trimmers so i just did the 10K, 390ohm 
input, still perfect symetry!!!! And that "with or without" diode
biasing, it simply dosent matter!!

Hoorrrraaaayy!...................
(i had expected some offsett problem inline with previous HB,JH posts, but no!)

Here comes the fun stuff, leave out one of the diode bias resistor,
you will then get a perfect sinus on one integrator output and 
a very good triangle on the other integrator! 
(corners nicely rounded but fall and rise very linear).
Speaking of controleable nonlinearity folks!

> The oscillator runs easily to beyond 30kHz. 

Mine to, even 34khz if stretched.

> At that frequency there is 
> some increase in amplitude, probably due to opamp phase shifts.

Mine "decreases" slightely in amplitude?!(LF353 as opamps).
Have to investigate that futher!
 
> At the low end, I checked operation at 0.5Hz.  Here there is an interesting 
> issue.  On power-up it takes the system about 2 minutes to build up to full 
> amplitude with both nonlinear circuits in place, and about twice that with 
> just one, as in the posted schematic.

With your schemo and my gain scaling i get instant start up at any frequency!
If i close the Iabc to zero the VCO oscillates with the speed of the leakage
of the OTAs, however they do it at different speeds!! Takes more 
then 10 minutes to complete a cycle.

> However, once the system is running 
> it changes frequency rapidly when the control current is changed, probably 
> because the quadrature phase relation is already established.  So I think 
> it can be swept OK.

It sweeps all right manually.
 
> The low frequency buildup can be reduced to about 15 sec by reducing the 
> feedback resistor from 1 MOhm to 100kOhm.  This increases the distortion a 
> bit, with the second harmonic now down about 30dB.

I made it to over saturate badly into crude and unbalanced
square waves, still performs as nice as prevoiusly, this is a
very stable cirquit! 
 
KD