Audio over PWM

Sigma-Delta (1-bit)modulation is a preferable means of digitally transmitting audio. This is what comes directly off the CD's.

>> This is what comes directly off the CD's.

Many CD players use single bit sigma-delta D/A components, but that's not the same as what is actually on the CD.

You can make a 1st order sigma-delta loop with an op amp integrator, a comparator, a clock source, and a 1 bit DAC feeding back into the integrator.

Then, band-limit the pulse stream out of the comparator, and feed inot the mod port.

I have seen power output modules using PWM.
That is, pure PWM from CD player to speaker !
(give or take an LC output filter to reduce the broadcasting from the external wiring)

I think it is called a 'class D' output stage,
and the size of such a thing seems to be extremely small.
(hardly any need for a heatsink)

Class D amps have amazing efficiencies, and the amp size is quite small compared to the power levels they can output. However, switching frequencies need to be several orders of magnitude over the reproduced frequencies of interest. This means they are generally limited as bass amps. A few companies have had limited success with class D (generally modified, so not pure class D) full-range amps, but I believe the main market jury is still out on those.

Sorry about the CD statement. I worked on A/D's not D/A's As I recall the CD's use 12? bit words, some companies digitally resample and use sigma-delta or MASH to produce and effective 18 bits. The main advantage of the sigma-delta is that the noise shaping effectively pushes the Q-noise to higher in the bandwidth so that a simple low-pass filter can reject it, therefor higher SNR, higher "fidelity"

As far as the RF is concerned, the output of the sigma-delta or PWM needs to modulated on a carrier. OOK has bad BER, PSK suffers spectral regrowth when using non-linear carrier amps (the cheapest), therefore the best choice is a QPSK derivative or a FSK derivative, MSK etc. These all seem to converge assymtoically on the same solution at the limits of bit-rate/bandwidth, ie. no abrupt phase state changes. With all the digital com chips these days this sould not be to hard to pull off. Then there is a carrier frequency to choose, I prefer the lower freq's, however there is more interference, but filtering is easier

...and I'd prefer to choose something license exempt too...

Thanks for the comments guys. Pondering and googling will follow...

Spex

hi, transmitting audio ovr pwm is easy as long as you tolerate some noise.

using rf for phone quality bandwith only requires few components a little microcontroller with an adc module and a citizen band rf transmitter. then check for minimum dt which impose restrictions on Bw.

regards

killa