On Fri, 18 May 2001, James Newton wrote:

> Bloody fascinating Scott...

Isn't it fun contorting square waves!

>
> ...killed an hour and a half on that one page and I still don't really have
> it, although I've got a sort of an "almost there" sort of buzz...
>
> ...could be just brain cells burning out...
>
> ...now where did I put those aspirin...
>
> If I got all that correctly, in practical terms, "Magic" sinewaves are
> better when you are switching lots of power (like a motor driver or power
> amp) and PWM is better when power efficiency is less important but you DO
> need:
> A) lower uP clock speed and code complexity or
> B) higher output frequencies or
> C) more even variations in amplitude such as when you need
> D) summation of multiple frequencies to one output (digital mixing).
>
> Is that right?
>
> Has anyone made a motor driver that uses "Magic" sinewaves?

Mike Keitz made one. Where's Mike?

There are some practical problems with Magic Sinewaves. On a pic at least, they
can only be implemented in software. This requires isochronous code.

There are (at least) two ways to implement them in a pic. One is to put the MS
into a ROM table, the other is to encode the MS directly in instruction delays.
The look up table requires ~10-15 cycles per edge, while the direct encoding
only requires 3 (or one if you're really twisted and demented - 1). I suppose a
co-routine solution would be appropriate too.

Another problem, or perhaps side effect, is that since the patterns are hard
coded, there is less amplitude variability/control. Also, it's more difficult to
control the generated frequency. I suppose you could add vriable dely in the
looup table, but the quantization effects would cause large frequency jumps.
(BTW, a better way to control the frequency is to PLL the RC oscillator on a PIC
- something else Mike has done).

Scott

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email listserv@mitvma.mit.edu with SET PICList DIGEST in the body