Scott Dattalo writes:

> First a question, do you need DC speed control or an amplitude
> controlled sine wave? I think the former, but just in case...
> Count on dedicating one PIC for the magic sinewave synthesizer.
> Of all the possible bit streams for a given sequence of bits, the
> number of unique amplitude controlling low order harmonic distorting
> ones is limited. Your alternatives are to 1) live with limited
> amplitude control 2) allow more lower order harmonics 3) Increase
> the length of the bit stream 4) maybe someone else has a fourth
> reason (how about get a Scenix chip then you can have magic software
> running on magic hardware).

I don't think that amplitude control is really the point of using
magic sinewaves. I didn't even realise that you had any amplitude
control at all. But you do have accurate phase and frequency control
which is important for speed control of AC motors. To implement magic
sinewaves there are only four drive transistor transitions per cycle,
so even if you want 500 Hz you have only 2 kHz switching. Although the
bit sequences that define the magic sinewaves are long it all comes
down to accurately timing the four transitions. That is, you don't
necessarily have to shift 1000-bit sequences to generate magic
sinewaves.

> I would really like to see how much more efficient the "magic"
> sine waves are compared to the less ethereally PWM generated ones.
> Has anyone done the arithmetic, or perhaps less demanding, has
> anyone else even thought about it (or even less demanding does any
> one else give a damn)? The number of transitions over a fixed time
> interval is lower for the magic sine waves. However, the intermediate
> harmonics are higher. Let me quantify: with magic you can exactly cancel
> all of the even, and the third, fifth, seventh, and perhaps ninth
> harmonics. However, the 11th, 13th, and etc. are relatively large
> (please excuse the vagueness). Now with PWM for the same frequency
> sine wave there might be a residual 2nd, 3rd harmonic, etc. but the
> amplitude doesn't increase until you approach the carrier frequency.

I'm haven't bothered to check the claims of harmonic distortion and
compare with PWM. But power efficiency is going to be heaps better
with the magic sinewaves. If the PWM frequency is 10 times the
sinewave frequency then there will be 20 transitions per cycle (cf 4
transitions for magic sinewaves). The motor will filter out the higher
harmonics so having no low-order harmonics may be all you require.

Magic sinewaves require a dual polarity supply which is less
convenient when you are operating from a solar panel.

Mal Goris
--
http://www.nfra.nl/~mgoris/