How good are the digital filters in digital audio systems?
(From: Winfield Hill (hill@rowland.org).) The digital filters within a typical CD-sound sampling system are very good indeed. I’m looking at a few AES papers reprinted in the 1994 Crystal Semiconductor databook (so we’re talking “old” technology!), and I see the amazing performance possible with the linear-phase finite-impulse-response (FIR) filters in the delta-sigma A/D chips. For example, the Crystal CS5328 has a flat response to 22.5kHz and then drops like the proverbial rock to a first -105dB dip at 26kHz. Ditto for the filters in a high-quality D/A like the CS4328. Also, the in-band frequency response is very good. Passband ripple within +0.00025 and -0.0004dB to 10kHz. Hmmm, deteriorating to -0.0006dB at 17.5kHz. And for the D/A chip, a flat line on the chart (I can’t see under 0.01dB) to 20kHz with a slight 0.1 dB rise by 22kHz. Strike that “very good,” insert PERFECT. The Crystal CS5328 A/D has a very low -105dB distortion with full-scale analog input, and -125dB with -10dB in
