simple way to add harmonics to a fundamental signal

[Martin Czech]

> >3x is harder to obtain. You could try to feed a sine into a Tchebycheff
> >polynome function.  3x would require a 3rd-order polynome.
> >
> >The question is how to get the x**3, x**2 functions needed.
> >Multipliers are expensive, that's why I thought about a nonlinear diode
> network.
> 
> What about LM1496-type ringmodulators, I've a neat schematic for a
> dc-coupled ringmodulator at my website, two of these properly cascaded will
> give you x^2 and x^3 at the simultaneously. The LM1496 is pretty cheap, and
> you'll need many diodes 
> to have similar precision.

True and not true.

I was more looking at higher order polynomes, say up to 7th or 8th.
It is not so clear that the error will be small using multipliers.
OTOH a nonlinear diode network may be not so bad, since every x**n has
it's own network, so no cascading errors.  I wanted to cut the diode
number by two, using a rectifier/polarity switcher scheme, thus only
one quadrant computation is needed.

I'm currently thinking about a programm to compute the optimum
number of diodes for a given relative error. I think 1-5% error
will be good enough. 

I wonder if the multiplier error will be in the same range,
for some reasonable temperature range (10C-55C Ta).

I have to point out : I have nothing soldered yet.


m.c.