Like watchn a multiple tennis tennis match, i feel like I took in too much.
So far this is the last quote I enjoyed most:
The issue with sigma delta dacs is that they take 16bit/24bit input and converts them to Bitstream data @ 5bits(5bits for sabre/hugo/dave or less bits for other designs) 2.8MHz(for Ess Sabre)/104MHz(Dave). From what I understand this conversion process is quite destructive(time domain or samples wise) and is a lossy/decimation process. Also there's alot of complex feedback algorithm at work here(in the case of hyperstream), these feedback systems/noiseshaper algorithms are not fully understood by the dac designers(check out video of rob watts explaining noise shaping simulation below, perplexing even at his level). So much "black arts" involved in designing sigma delta dacs when a pragmatic dac designer can just stick to high precision R2R to get really good signals out of the decoder and design whatever filters codes/analog stage necessary to get good sound. The amount money spent on "DSP cores/FPGAs" noise shaping system is getting more expensive than just getting a highly precision R2R chip, not to mention with all these high Mhz cores, they might leak higher emi/rfi(electrical noise) into the surrounding audio components and require more filtering/pcb noise mangement/power conditioning and etc etc(increased cost/design time).
Rant: effing money grubbing texas instruments killing off the pre Burr-Brown true R2R chips.
Esstech patent(relates to Sabre Hyperstream modulator)
should have done a few takes,
so he could smooth out what was trying to be said..
Edited by Maxx134 - 6/11/15 at 12:23pm