A warning - there is a fundamental problem with the AS3340. Changing the pulse width causes shifts in output frequency.
AFAIK this is not a problem with the CEM3340 Rev G, nor the V3340, unless anyone knows better. I haven’t tried the V3340.
The simple workaround is to take the AS3340 triangle output and slice it at different voltage levels with a comparator. That gives you a variable width pulse. For example, Frequency Central’s CEMOsc uses this technique, using a spare op-amp. The AS3340’s own PW input and pulse output are unused.
If you are using a quad op-amp like the TL074 to buffer the 3 outputs (or in Sam’s case, two TL072’s), the spare op amp can be used as a comparator.
I see that Sam is using one half of a TL072 as a ‘CV Adder’ non-inverting voltage follower to control the PW input to the 3340. If you instead produce the pulse externally as described above, you could use that op-amp as the comparator instead, with a little juggling.
Alfa Rpar’s ‘solution’ is to produce a hybrid device , the AS3340-HYB, which contains an AS3340 derivative die and a comparator die, but it’s expensive and uses more power - you may as well use a CEM3340.
There are other minor differences between the CEM3340, AS3340 and V3340. I recommend Electric Druid’s website, he sells all 3 versions and the spec sheets are there. He also has an interesting article about historic designs using the CEM3340.
If building Sam’s VCO, I recommend sticking with the CEM3340.
A couple of comments on Sam’s design:
First, the ‘square’ output of Sam’s design is more accurately described as a pulse output, as the pulse width is of course variable.
Secondly, it is good practice to put capacitors across the supply lines near ICs to reduce noise and interference, say 100nF between ground and each supply rail. Also a couple of fatter electrolytics, say 10uF/25V, between ground and the supply rails would improve performance. These components cost pennies.