James Richard Tyrer wrote:
This is an idea that I came across.
Now we have our 24b/channel 192Ks/s audio and I presume that we would
like to listen to it and retain the very high S/N and low distorting.
I found this paper which was actually presented at an AES convention.
Digression: is anybody a member?
http://www.hindawi.com/GetPDF.aspx?doi=10.1155/2007/94386
Distortion-Free 1-Bit PWM Coding for Digital Audio Signals
So, I was working on this since there are other ways to do this. I was
working on linear interpolation since this can be easily done with a
FPGA using the Bresenham Line-Drawing Algorithm. Even some overlap with
the graphics project.
One slight problem. Despite the fact that this paper was presented at
the AES convention, it is bunk.
This still makes no sense to me, but this:
http://ir.sun.ac.za/dspace/bitstream/10019/79/1/JacoD.pdf
claims to have the math to show that harmonic distortion is generated.
My rational is based on a basic analysis of superposition.
1. If the sampled signal is run through a LP filter then the original
signal is fully recovered.
2. The total power of the PWM pulses is directly proportional to the
aptitude of the sample.
3. The only thing that changes is that the power is provided in a window
in each sample of the sampled signal rather than uniformly throughout
the entire sample period.
4. (This one needs to be proved) All of the frequency products from this
are sidebands of the carrier frequency of the PWM, and therefore are
above half the carrier frequency.
Perhaps #4 is only true if we use separate positive and negative pulses.
I always thought that this was a better design since if done with
commutating diodes, it eliminates the distortion cause by shoot through
prevention.
I suppose that I will have to do the math for that since he didn't.
--
JRT
_______________________________________________
Open-hardware mailing list
[email protected]
http://lists.duskglow.com/mailman/listinfo/open-hardware
