Xref: utzoo sci.electronics:18395 comp.dsp:1387 Path: utzoo!news-server.csri.toronto.edu!rutgers!cs.utexas.edu!wuarchive!cec2!news From: fcr@saturn.wustl.edu (Frank C. Robey X5569) Newsgroups: sci.electronics,comp.dsp Subject: Re: A question about the Nyquist theorm Message-ID: <1991Mar12.232113.14764@cec1.wustl.edu> Date: 12 Mar 91 23:21:13 GMT References: <4692@apricot30.UUCP> Organization: Washington University, St. Louis MO Lines: 48 In article <4692@apricot30.UUCP> duerr@motcid.UUCP (Michael L. Duerr) writes: >From article , by wilf@sce.carleton.ca (Wilf Leblanc): >> When I bought my CD player, it said on the front panel 'Dual D/A >> converters'. For fun, I asked the salesperson what that meant. >> The reply was rather funny, and of course completely inaccurate. >> >> What does this really mean ? (I figured maybe two distinct D/A's rather >> than 1 D/A and two sample and holds ??). > >Yes, it means that. Or, more likely, one dual-channel D/A. > >There are a couple of reasons why. Sample and holds have errors known as >pedestal - a voltage step that occurs when they transition between state - >and droop, where the signal decays as it is held. .. other reasons for using dual dacs based on current technology deleted. When CD players were introduced, the D/A's used produced a lot of "glitch" energy during transitions in level. This glitch energy showed up as harmonics and other undesireable spurious and thus needed to be removed. Sample and holds (or more correctly track and holds) were used to remove the glitches. Even if they had used dual DAC's, dual track and holds would have been needed. I seem to recall that the level of distortion terms without a track and hold was about -70dB below the fundamental. This level and the noise floor varied with the fundamental frequency and the sampling frequency- the higher frequencies had higher higher distortion levels and noise. Current audio-quality DAC's balance delays (the major cause of the glitches) between bits to minimize the glitch energy. This was not true of the DAC's available several years ago. As for pedestal, it does not need to be a particularly non-linear effect. If you look at feedthrough, a linear effect, then this will only slightly reduce channel separation- to maybe 60 dB or so- not a particularly drastic problem in my opinion. I was working in at group at HP- Lake Stevans in the early 80's that was trying to get a dynamic range from a digital source around 130 dB. Around this level, many components are no longer linear. Depending upon the type, voltage, and impedance levels many capacitors and inductors created harmonics at levels far in excess of that. At that level even some resistors were not useable. I suspect that some of the high-end audio equipment is finding this problem now with the "18" and "20" bit DAC's. Frank Robey now at: fcr@ll.mit.edu MIT- Lincoln Laboratory