AW: CA 3080 Tri to Sine converter question

[Haible Juergen]

	>My concern when implementing the Tri-Sin converter is "Just how
pure is the sine output?"  On the scope it >*looks* good but I lack the
instrumentation to really measure the purity.  Can anybody quantify the
purity of >the sine output for an optimized Tri-Sin converter using a 3080?

I've built several 3080-based sine shapers, and I got decent results. Expect
about 2% THD. As someone pointed out, with optimal THD you don't have
optimal waveform (i.e. sharp corner) which may be suboptimal for slow
modulations (LFO).
A far better way to use a differential pair as sine shaper is the degenerate
emitter method. Don't connect the two emitters directly, but with a small
resistor. You can't do this in a 3080, of course, but there is another cheap
IC that works fine: The LM1496.
Take the following circuit, 

http://www.it.kth.se/~e93_mda/synths/friends/haible/fs-1.modulators.jpg

and connect the "SigSin" input to GND, and you have a much better sine
shaper than you could build with a 3080.
You don't need the coresponding trimpot then, of course.

Note:
Actually, the circuit is a combined sine shaper / ring modulator, i.e. you
feed
in one sine wave and one triangle wave (of the right amplitude), and you get
a mixture of two *sine* waves (f1+f2, f1-f2) as output. The sine input has
to be
ac coupled (rference voltage is not 0V). If you ground this input (dc
connection
to GND), your sine input is a "sine" of f=0, i.e. the whole thing is just a
triangle
to sine shaper. I wanted to have both options in one circuit, because at
some
point I considered using extra ring modulators based on RC4200.
Ok, end of JH FS-1 history trivia (;->).

JH.