pmr@drufl.UUCP (Rastocny) (03/28/84)
--->a tidbit for Michael Jackson's software<--- Last Monday, I listened to Bob Carver, founder and chief engineer of the Carver Corporation, speak on the measured differences between digital CDs and analog discs cut from the same master tape. Before I begin, let me fill you in as to why Bob started this examination. Bob, being very familiar with all the technical reasons of why digital should sound better than analog, bought a compact disc player. Much to his surprise, he found that CDs didn't reproduce the instruments, soundstage, or ambience as faithfully as the best analog recordings. (Hmm. This sounds like something I've been saying for the past year. However, since I didn't back up my claims with statistics, few people took me seriously.) Being an engineer and not doubting his own ears (!), Bob set forth to find out why he heard these differences. He started by buying eight CD players; some with digital filters, some with analog filters, some with both, some with 16 and some with 14-bit converters, some with multiplexed conversion channels and some with dual conversion channels. He also bought a good assortment of phono cartridges; some with ruby cantilevers, some with line-trace styli, some cross-coils, some moving irons, some moving magnets, and some moving coils. He even used several different preamps. With minor exceptions, all achieved about the same results in the three experiments explained below. Experiment No. 1: Dragging all of these units into his lab, he set forth on finding out what electrical differences there were as the signal is delivered to the amplifier. He obtained several CDs and analog discs that were made from the same master tape, synchronized the turntable to the clock in the CD player to get exact playback synchronization, ran the stereo channels into a switch, and ran the left and right switched channels into the x and y inputs of an oscilloscope. He then switched back and forth between the CD and the analog disc. What he observed in the Lissajou patterns is kind of shown below (it's very hard to draw an ellipse on a screen, much less orient two of them at 45 degrees. I think you'll get the general idea.): y input | (left | channel)| aaaaaadaaaaaaaa | aaaaaaaaaaa dddd aa | aaaaa dddddd dddd aaa | aaa ddd dddd aaa | aaa ddd dddd aaa | aaa ddd dddd aaa | aaa ddd ddd aaaaa | aa dddd aaaaaaaaaaa | aaaaadaaaaa |--------------------------------------------------- x input (right channel) The major axis (lower left to upper right end) of the ellipses are actually at 45 degrees and not as drawn. The major axis of the ellipse shows the L+R (mono or center channel) signal component and corresponds to the horizontal motion of a stylus playing a record. The minor axis (lower right to upper left end) of the ellipses are actually at 90 degrees to the major axis and not as drawn. The minor axis of the ellipse shows the L-R ("ambience") component of the signal and corresponds to the vertical motion of a stylus playing a record. The analog and digital formats both had about the same amount of L+R signal component but the L-R information content was about 2-3dB greater in the analog disc than in the CD. (Remember, people on this newsgroup have pointed out over and over again that differences of less than 0.3dB can be detected by the human ear.) Experiment No. 2: The next experiment Bob performed was in frequency response. Since he felt that the final playback version of the analog disc more accurately represented what instruments and their harmonics sounded like in reality, he plotted the CD's response against the analog's response using a 1/3-octave real time analyzer and roughly obtained the plot shown below (note that the frequency scale is approximated. My HP doesn't draw log scales): +3 a=analog disc d=CD dddddddddd +2 ddd ddd ddd ddd +1dB ddd ddd ddd ddd 0 aaaaaaaaaaaaaaaaaaaaaaadddaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaadddaaaaa ddd ddd -1 dddddddddddddddddddd ddd -2dB 20Hz 860Hz 5KHz 20KHz There were some slight differences again between CD players but the shape was always the same at the same frequencies and the 5KHz peak never varied more than 1dB for a relative difference for the CD of +2 1/2 to +3 1/2dB referenced to 20Hz. Experiment No. 3: The last experiment he performed with a stopwatch. Using a Joni Mitchell vocal piece, he timed the decay of the last voice at the end of the song. The analog version lasted about 15 seconds and the CD version lasted about 11 seconds. Many reasons were stated as to why these differences might exist but Bob said when asked directly (in paraphrase), "I don't know why these differences exist or what part of the chain caused them... Both analog and digital master tapes seem to behave similarly... It's not my job to find out what caused it, it's my job to fix it!" So Bob bought out a mysterious circuit board and plugged it in between the CD player (a Kyocera DA-01) and the electronics (a Conrad Johnson Premier Three preamp driving a Conrad Johnson Premier One 200W/ch power amp). I had been listening to Debussy for a while trying to give this format one more chance. My ears were literally hurting after about 10 minutes and I was about to leave when Bob switched on this little gizmo. ALL OF A SUDDEN, THINGS GOT MUCH BETTER. My ears were still hurting (it took the rest of the evening before the pain subsided) but I could tell that imaging was definitely better, that the sound was much less glassy and harsh, and that the ambience was very good. Bob calls this gizmo his ---> Digital Time Lens <--- and said that this was the first time he had played it outside of his lab. He said that it basically corrected electronically for the things that he measured above. He said that the retail price would be about $250 and that it will be available in about three months. Now I don't want to say that the Carver DTL is the solution, and neither did Bob. But until people figure out what's causing these measured differences and what can be done (if anything) about them, the Carver DTL makes todays CDs very listenable. This could also be interpreted by the digital buffs as making CDs sound like analog discs. In the end, no matter which you prefer, the system should accurately reproduce the music. Those of you who know what a violin or a saxophone sounds like may prefer adding the DTL to your CD system. Others may not. However, I would recommend to all CD enthusiast that they listen to this gadget before jumping to any conclusions. If I owned a CD player, I would buy one. Bob also mentioned that he will be coming out with a CD player soon that will allow the DTL to be plugged into the D-A converter section of his unit and increase ambience even more. One interresting thing pointed out at the session was that the inner ear canal resonates at around 3KHz for most people (about 500Hz either way from this value). This 5KHz peak is roughly the first harmonic of this value and may explain why my ears hurt while listening to CDs. Please note that these are my opinions and not that of my employer. I do not own stock in Carver Corp. and I'm not being paid for this (or any other) review. Yours for higher fidelity, Phil Rastocny AT&T-ISL ..!drufl!pmr
smb@ulysses.UUCP (Steven Bellovin) (03/28/84)
That's an interesting article, and provides some food for thought. Note, though, that much of the data can be explained as infidelity in the more common medium, i.e., we're accustomed to hearing a slight drop in the response curve at around 4Khz (or wherever you drew it), and slightly exaggerated separation. And it makes Carver's DTL sound like an equalizer plus a widget that boosts the L-R signal's amplitude.
jj@rabbit.UUCP (03/29/84)
I've also heard quite a bit of discussion about the differences in L-R and frequency response. My own feeling (which has been attributed to Bob Carver <first, please, I didn't come up with it alone... I think many folks did> ) is that the ANALOG playback system has the defective frequency response, and that the differences are what is heard. It's clear, from the nature of digital, that the difference in L-R must come from the ANALOG process. The difference in frequency response (which an author shows as flat for the Analog equipment) is RELATIVE. I.E. It shows the DIFFERENCE between the analog and digital equipment, not that either one is right or wrong. Incidentally, I object to representing the digital as the WRONG response offhandedly, since the methods used to generate the digital signal are much more easily controlled, but that belongs in net.flame. Let's just say that the learned perception of the deficencies in analog recording are missing, and that the frequency shaping (which is pretty severe) that the listener is accustomed to is missing. The L-R signal is easily explained as well, because of the way that record lathes work. (It's interesting that digital techniques point out so well the differences between analog and digital reproduction, while the analog measurement techniques of 10 years ago couldn't even come close.) Oh well, goes to show: Garbage In, Garbage Out. In other words, when they finally stop compensating discs the same way that they compensate alaong records, we'll finally have listenable CD's. Not only that, it IS the recording, not the medium. Fancy that! (I'm sure we'll find a few defects in the medium, too, it's just that nobody has convinced me of any yet.) Interesting, isn't it, that this whole problem was caused by the deficiencies in analog reproduction? <Or so this author thinks, as he's said on many occasions. Shame I can't do like Bob Carver, and buy 50K$ of audio equipment to prove my point.> Cheers -- TEDDY BEARS ARE NICER THAN PEOPLE--HUG YOURS TODAY! (If you go out in the woods today ... ) (allegra,harpo,ulysses)!rabbit!jj