Re: Q: Magic Sinewave short sequence

On Mon, 02 Apr 2007 09:13:26 -0700, John Larkin
<jjlarkin@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> wrote:

On Mon, 2 Apr 2007 17:48:29 +0200, "Henry Kiefer" <otc_friend@xxxxxxx>
wrote:

"John Larkin" <jjlarkin@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> schrieb im Newsbeitrag news:cn6213dm2uc3ginrnhi7ga2bov8ssveljd@xxxxxxxxxx
| >sigma-delta in an FPGA, and feeding it a sine wave in digits. The output
| >would then be a string of bits that would provide your required bipolar
| >signal in real time.
|
| Right. A counter would scan a sine lookup table, producing, say,
| 12-bit sines. A delta-sigma stage then decimates that to a 1-bit
| stream whose duty cycle tracks the sinewave. I think Xilinx has
| appnotes.
|
| We've done something like this to force a 12-bit dac to have 16-bit
| resolution.

But how will this suppress the lower harmonics? I think the sprectrum of your solution is nothing else than a digital signal
spectrum: amplitude(n) = 1/n * f^n and n=2k-1

- Henry

Well, you'd have to read up on the delta-sigma math. But the summary
is that the process "noise shapes" the output spectrum, shifting the
noise energy (and distortions) up in the spectrum, where the high
stuff is easily filtered out. That's how a 1-bit dac can make good
audio, or how a cheap d-s adc can deliver 24 bit data.

John

Oh, yeah, if you used d-s, you can stick a multiplier after the sine
lookup table and get high-resolution amplitude control for free.

John

.

Quantcast