AW: Re: FM of Tri-Square VCO

[Haible_Juergen#Tel2743]

 > Has anyone really done it in analog yet?

I have. I think I have even posted about it to DIY, so you
could find the details in the archives.
But I try to write it again in short form, which is easy as Gene
already has given a great description:

>     My understanding is that through-zero FM on a VCO is when a descending 

>     CV into the VCO FM input causes the VCO to drop in frequency until it
>     hits zero hertz, and then as the CV keeps going more negative the VCO
>     starts up again in the reverse direction and begins increasing its
>     frequency of operation.

That's it all. Keep in mind, that "reverse direction" means phase inversion
of the triangle output signal.

>     rop to a lower and lower frequency without ever even reaching
>     zero hertz. Therefore one may consider approximating the through-zero
>     behaviour described above by building a VCO whose minimum frequency is 

>     achieved at zero volts CV sum in (rather than the minus supply) and
>     putting a full-wave rectifier on the CV input. This way a decreasing
>     CV input will cause the VCO to stop (actually just get real slow) at
>     zero volts and as it goes more negative it gets inverted by the FWR
>     and goes positive again, starting up the VCO into increasing
>     frequency.

The resistor that sets the reference current in the expo generator is
the right place to start, as it usually goes to a virtual GND summing node.
Instead of connecting this to VCC, connect it to the output of the FWR.
The FWR gets two inputs: one for the initial frequency (i.e. resistor to
VCC ...), and one for tzFM input.

>     Now the important bit here is that apparently this doesn't sound quite 

>     the same as a true through-zero FM effect. See a true through-zero VCO 

>     will go down to zero hertz and stop dead and then back up and reverse
>     its cycle.

Now the only thing that is left to do, is changing the phase of the VCO
output signal just at the same time as the FWR's input changes sign.
This is easy, because the diode drops in a opamp/diode-based
FWR cause the opamp output to make a little "jump" anyway, when
it passes thru zero. Connect a comparator to this point, and toggle
the direction of the capacitor charging with the comparator !
A possible way to do this would be using a D-flipflop in the triangle
VCO. In normal operation, an upper threshold SETs the flipflop,
and a lower threshold RESETs the flipflop. Tri/Square VCO as usual;
every 555 timer works this way.
But now for Thru Zero FM: Connect Qbar to D, and use the clock
input of the flipflop to toggle the phase.

>      enough to justify scads of complicated
>     through-zero FM VCO designs.

I don't think it's too complicated. I built an add-on to a CEM3340
VCO within a few hours. (And the 3340 makes things more
difficult here, compared with a discrete tri/square VCO!)
You can find the circuit in an old Curtis newsletter, which is part
of Barry's book on CEM/SSM chips.

How does it sound?

I wouldn't describe tzFM as a dramatic new effect. In contrary, it sounds
like linear FM should sound like - only that ordinary linear FM stops
at a small modulation index (well, when you reach zero hertz ...),
so that the effect is still rather "tame". TzFM just gives a larger 
modulation
index, and thus enters the more interesting regions.

A little problem:

The FWR does not work perfect (even with offset trimming). The result
is a *very slight*  irregularity when you just cross the border between
FM and tzFM. But you can live with that. There is also, for the same reason,
a *slight* frequency shift between very small and very large modulation
strength. "Slight" is a few cents - not the annoying shift which you know 
from
expo FM.

Hope this helped - if you want to know the details, it's in Barry's book.

JH.