[sdiy] Xilinx 3E board comments

[Paul Maddox]

Eric,

> Good point. That gives you ~0.01Hz resolution.

yep, 0.01hz is adequate, you could easily increase the bit depth of the 
accumulator to get greater accuracy.

> It also puts Nyquist at 25 MHz, so the aliased harmonics will be down 
> ~67dB worst-case by the time they fold back into the audio range. Not 
> quite CD-quality, but better than the 45dB I was seeing.

ahh, but carefull :-)
the monowave has an update rate of 1Mhz, so a nyquest of 500khz, but you 
need to get above 4Khz pitch before you can begin to hear the artifacts.. 
but by then a 4Khz note is begining to hurt my ears...
Given that few people would play a note above 6Khz for any length of time, I 
reckon a 50Mhz update rate is as good as 'alias free' as you'll ever get.

> Then you either have to build a (fairly simple) decimating FIR to get the 
> sample rate down to the audio range to output through a good-quality DAC, 
> or use a Delta-Sigma converter running in logic at full speed like the one 
> Jim Patchell has on the fpga.synth.net site. (has anyone tried that yet?)

to my knowledge, no, nobody has tried it.. I know Jim synthesised it...

Paul