We previously went through this crap with CD players.
The sampling frequency was chosen to be 44.1 kHz, well
beyond the range of human hearing. No filtering would
be needed.

Except for one thing...when they played the CD back
unfiltered, people would find their tweeters melting
for some weird reason....44.1kHz! at huge powers!

Out came the drawing board and complex analogue (and
expensive) filters were designed until one day some
smart engineer discovered they could double the freq.
in a computer and put out 88.2 kHz sampling noise and
use a less efficient and less expensive filter.

Well, the audio hype that came out then was "2 times
oversampling" followed by 4x, 8x, 16x & 32x
"oversampling". Shister and ignorant marketing people
explained this as "reading the CD 16 times repeatedly
and eliminating the digital errors" "You can eliminate
scratches with this"

In reality the "oversampling" technique was the
development of a digital filter that made the analogue
filter into a simple capacitor to eliminate the
sampling noise.

Any square wave can be filtered enough to produce a
pure sine wave. The trick is the cost. Huge core
inductors and capacitors to handle and smooth out big
quantities of power cost money to design and money to
produce. Not to mention the sheer weight of the beasts.

Multistep waveforms can be filtered much easier. This
is analogous the "oversampling" technique used in CD
players of years past. Digital filtering is much easier
and cheaper than the equivalent analogue filtering.
It's not like an inverter, these days, doesn't have a
computer chip inside then anyway.

How little distortion do you need anyway? Most of it
can be accomplished inside the computer and then just
amplified to useful power levels. At a cost, of course,
and a marketing tool for more money...always.


"daestrom" wrote in
message
...

"philkryder" wrote in message

oups.com...
Steve - How many equal "steps" are necessary for
the MSW inverter to be
a sufficiently close approximation to a "rotary"
sine wave?


Careful. Even true 'rotary' generators don't always
put out a true sine
wave.

Even the very, very large commercial generators used
in power plants, don't
put out a 'pure' sine wave. The number of stator
slots and rotor geometry
cause a small amount of harmonics. The exact
connections of windings is
often used to help improve the fundamental and
minimize some of the higher
harmonics (6th, 9th and 11th are some of the more
troubling ones).

After it passes through several transformers, getting
to the substation,
most of the harmonics have been filtered out by the
characteristics of the
transformers.

So the question, as usual, boils down to 'how good,
is good enough?'

daestrom