rcd@opus.UUCP (Dick Dunn) (06/28/84)
ONCE AGAIN, we rejoin Charles Pearson. First, we see Pearson battling the awesome foe, Audio Theory: >For all of those out there who insist upon praising CDism and >their (non)square waves by relying on the FFT sine-wave generation >approach have missed something intrinsic in their approach. Charles, Ivor Tiefenbrun (sp?) loves you for writing things like that. (For all I can tell, he could be PAYING you for writing things like that.) I hold no brief for (or against) the CD, but I tend to get a little annoyed when someone who (apparently) knows virtually nothing about signals, frequency response, Fourier analysis, and the like, makes ignorant statements about CDs and square-wave response, and continues making them in the face of massive refutation. Face it, there have been almost 20 articles that I've seen here in the past week or so, from numerous contributors, trying patiently to explain to Pearson that he's in left field. I haven't seen ONE article coming to his defense - small wonder; almost everything he's said has been wrong. I'm all for moving the whole square wave discussion to net.flame. The ratio of technical content to emotion in Pearson's postings ( 0 : a bunch) certainly doesn't merit keeping them here, and the people who have refuted his arguments could well spend their time and patience on matters of greater substance and interest (though I thank them for taking the time to TRY to straighten him out). >This method will produce a 'square' wave with maximal distortion >at the points of the square, not a uniformly decaying sine wave >distortion. In case you're wondering, this is an explanation of how Fourier analysis of a square wave gives the wrong answer. My best surmise is that a sketch of a square wave with a (dull) No. 2 pencil gives the right answer. But I promised more than the square-wave topic... >Since it is almost universally accepted that the sample rate >of CDs is as low as possiple (or even too low already) I >thought that you would like to know that at least one SONY >CP player is playing in multi-plexed mono... cutting the >sample rate in half. I must be getting rusty - I just can't keep up on terminology. First I can't figure out what is meant by "soggy sound". Now I have to admit that I don't know what is meant by "multi-plexed mono"...Does this mean (1) that the upper frequency limit of this alleged CD player is 10 KHz? (2) that it doesn't use the same CDs as other players, since it only has half as many samples? (3) that it's a mono player??? (4) that Pearson, once again, doesn't know what he's talking about? It was actually interesting, reasoning out the term "multiplexed mono". Now I know that "multiplexing" means combining signals into a single composite signal (rough, colloquial interpretation), and that "mono" refers to a single "channel" or signal. Thus "multiplexed mono" means combining one signal into one signal. Damn clever idea, and I bet it doesn't take much circuitry to do it! (I can forgive Pearson's typos if he'll try to introduce some real semantic content for the rest of us. I'd also like a little substantiation of his arguments - e.g., WHICH Sony player, and very briefly, how does it differ from others?) Then there's the defense of tube equipment. Sure, there is good tube equipment, no question. Is it as good as solid state? Well, "consensus" is a risky term to use around audio, but if there is a consensus it favors SS. There are certainly differences in sound, though the ends of the reproduction chain (mikes, cartridges, speakers) tend to overwhelm them. But the idea that you can hear the transistors switching had to be hatched out of pure imagination. You can run either SS or tube amps in class A, A-B, or B. If the design is screwed up, A-B or particularly B can mess up the signal with "switching" transitions. (Tubes DO have capacitance too, you know...) The problem can be fixed for either tubes or SS, to the extent that the only reasons for running class A are along the lines of a cold house or snob appeal. And if we're going to try to compare tubes and SS, let's try to make the comparison technically sound. Lumping bipolar transistors with, say, MOSFETS (of various configurations) is nonsense; their behavior is as different, for reasons fundamental to the design of the circuit, as the behaviors of tubes (shall we say indirectly heated pentodes?) is different from transistors. (Hey, if tubes are such a damn good idea, where are all the bombardment-heated cathodes?) -- Dick Dunn {hao,ucbvax,allegra}!nbires!rcd (303)444-5710 x3086 ...Lately it occurs to me what a long, strange trip it's been.
charles@sunybcs.UUCP (Charles E. Pearson) (06/29/84)
Gee... Maybe I should have called it time-slicing.
Multi-plexing is a computer term. The concept easily
extrapolates to its use in other fields.
Since it is too complex an idea, maybe I had better explain it.
Time x, sample only channel a.
Time x+1, sample only channel b.
Time x+2, sample only channel a.
Time x+3, sample only channel b.
.
.
.
Time x+n, sample only channel a.
Time x+n+1, sample only channel b.
This clearly reduces the effective sample rate by dividing it by 2.
At any given time Z, sound is only coming from channel Q.
Rather simple an idea... multi-plexed mono.
The hope is that the time-slicing between the 2 channels will be
in-audible, but still very effectively reduces the sample rate of
re-production from any (every) standard CD with this logic.
But then you say, double the speed of the reading/processing and you
recover the loss... If so, it would also make reasonable the
doubling of the sample rate of all other units operating in true
dual channel mode, or the doubling of the resolution.
Not bloody likely.
Charles E. Pearson
UUCP: {allegra, seismo}!rochester!rocksvax!sunybcs!charles
decvax!watmath!sunybcs!charles
ARPA & CSNET: charles.buffalo@rand-relay
Physical: University Computing Services
4250 Ridge Lea Road
room 28
SUNY Center at Buffalo
Amherst, NY 14226ron@brl-vgr.ARPA (Ron Natalie <ron>) (06/30/84)
Charles, has it occurred to you that they might just run the D to A at twice the speed so they can get the same number of samples per channel out. It seems that perhaps it is cheaper to make one D to A twice as fast, as it is to make two D to A's at the original speed. -Ron
acscmjm@sunybcs.UUCP (07/03/84)
[ ]Help!!! I'm TRAPPED in a VAX 11-780!!!! > Time x, sample only channel a. > Time x+1, sample only channel b. > Time x+2, sample only channel a. > Time x+3, sample only channel b. > . > . > . > Time x+n, sample only channel a. > Time x+n+1, sample only channel b. > This clearly reduces the effective sample rate by dividing it by 2. This is correct (except instead of 'x+2' 'x+2k' where k is the sample time, but this is minor) > At any given time Z, sound is only coming from channel Q. > Rather simple an idea... multi-plexed mono. Ahhh, but you are forgetting something... it is NOT multiplexed mono, because the information corresponding to channel A and channel B is DIFFERENT, though interlaced. > But then you say, double the speed of the reading/processing and you > recover the loss... That's right. You double the resolution and then halve it by alternately sampling the other channel, you get the same resolution. (2 * 1/2 = 1) > ... It would also make reasonable the > doubling of the sample rate of all other units operating in true > dual channel mode, or the doubling of the resolution. You can double the sample rate without doubling the size of the medium or halve the time because there is now only ONE channel to deal with. Mike Moroney (respond via news, since i am moving on to bigger and better things)
rfg@hound.UUCP (R.GRANTGES) (07/26/84)
Sorry, Charlie. Your logic is impeccable (well almost), but , as usual, your conclusion is incorrect. The info rate coming off the disc is the same whether the D to A converter is time shared or not. The D->A converter does have to operate at twice the rate of either of two converters (when two are used), but that's about all. Sample and hold ckts stretch the pulses back to regular width. I undrstand that essentially the same D->A converters are used whether one or two, so the conversion is about as linear in one case as the other.And the sampling rate is the same 44khz in both channels. I might expect there to be a mite more crosstalk in the shared configuration, but I haven't noticed it in any tests. In any event, the result is the best, quietest, cleanest audio I have ever heard either way, and I have been listening a long time. It's still hard for me to believe "they" had the guts to market something this good. Where was their vice-president in charge of lousing up designs? Certainly every Americun firm has one. Don't the Japanese? Perhaps he was out to lunch. :-) Dick Grantges hound!rfg