sean@geowhiz.UUCP (Sean Rooney) (08/13/85)
Ok, I've read enough postings about cd's without really knowing what's
going on, how about some sort of introductory, moderatly technical,
description of cd's from one of you "experts" on the subject.
Specifically I would like to know:
1. Are there really more than one laser in some of these things or is
it true (as someone has said) that one laser is used and its beam is
split (optically?). Furthermore, what do the additional beams/lasers
give you that you don't get with one?
2. What is "oversampling"? It seems that every sample on the disk
should be read by all machines, is oversampling the re-reading of
a single sample multiple times, perhaps to confirm it's value?
And if that is the case, what happens to that sample in the event
of a descrepency?
3. "Error Correcting Code" This is probably the most obtuse point to
me, does "Code" imply that there is a CPU inside these things deciding
what to do in case of a disk error? How do different manufacturers
ECC differ?
4. Output filtering, digital versus analog. I think I have a reasonable
grasp on this section. Personal preference seems to be the main
discriminator since both types of filtering can do the job.
5. Manufacturers: Apparantly there are two designs used, Phillips and Sony,
which are manufactured by a limited number of companys. These companys
products are then marketed under many names. Could someone come up
with a semi-comprehensive cross-reference list of who makes what
machine behind the trim and which design they use?
6. Moving parts: Does the disk itself spin or does the laser beam
move? If the disk spins how fast does it go and how accurate need
the motor be? Is this a point between which one can discriminate
a "good" cd player from a "bad" one?
I don't want to waste anyones valuable work time coming up with the answers
to my questions :-) but I've been feeling kind of foolish reading about
"Golden Shovel" awards for statements which, to tell the truth, don't
really sound that unreasonable to me.
An alternative to an introductory posting (or mailing) could be a recommendation
of a good literature source for the information. Note, however, that I
am not an EE and therefore not prepared for any excessively technical
Journal reading.
So go ahead and educate me.
sean rooney
!seismo!uwvax!geowhiz!seansjc@angband.UUCP (Steve Correll) (08/15/85)
> 1. Are there really more than one laser in some of these things or is > it true (as someone has said) that one laser is used and its beam is > split (optically?). Furthermore, what do the additional beams/lasers > give you that you don't get with one? In lieu of a "groove", the disk has a spiral track of pits which alternately do and don't reflect laser light. To track this "groove", the player shines three (sometimes more) beams onto the disk and adjusts their position to maximize the modulation of the center beam and minimize the modulation of the surrounding beams. Some models use multiple lasers whereas others use beam splitters. I have seen no evidence that one approach is superior to the other. > 2. What is "oversampling"? It seems that every sample on the disk > should be read by all machines... Every machine reads every sample. "Oversampling" is signal-processing jargon. In this case, it means that samples from the disk go into a digital transversal filter at 44.1k samples/sec, and filtered data comes out of the filter at 88.2k samples/sec (for 2x oversampling) or 176.4k samples/sec (for 4x oversampling). Roughly speaking, the filter generates additional points by interpolating between the original ones. > 3. "Error Correcting Code"... How do different manufacturers ECC differ? Digital circuitry within the player can correct many errors by taking advantage of redundant information included on the disk, roughly as a computer disk-drive controller or memory controller does. The algorithm for doing this is straightforward, so I doubt that players differ in this respect. However, if the redundant information proves insufficient, a player may mute, or repeat a section of the program, or interpolate; how cleverly it does so may make a great deal of difference. Similarly, if the player loses its lock on the "groove" entirely, the algorithm it uses to recover makes a big difference. On one defective disk, my early-model Technics player performed better than Sony and Yamaha players because it apparently punted the defective section and skipped to a later, intact one, whereas the other players got in serious trouble attempting to track it. In another case, the Technics compressed several minutes of music into one because it kept skipping over defects, whereas the other players merely produced ticking sounds during the music. Sometimes the Technics produces ticks, too; the behavior of players in the presence of defects is fascinating. > 4. Output filtering, digital versus analog. Digital filter systems (in practice, they're hybrid digital-and-analog systems) usually exhibit less group delay (loosely speaking, "phase shift") at high frequencies than purely analog systems, and are less likely to deteriorate with time as components age. > 5. Manufacturers: Apparantly there are two designs used, Phillips and Sony, > which are manufactured by a limited number of companys... There seem to be many designs. Old Sony models use analog filtering; new Sony models use digital filtering; all Philips models use 4x oversampling digital filters, but some other manufacturers use 2x oversampling digital filters; some manufacturers switch one D/A converter between channels whereas others use two; and so on. > 6. Moving parts: Does the disk itself spin or does the laser beam > move? If the disk spins how fast does it go and how accurate need > the motor be? Is this a point between which one can discriminate > a "good" cd player from a "bad" one? The disk spins at a varying rate so that the bits need not be crammed closer together near the center, where the diameter of the "groove" is smaller, than at the edge. The player compares the bit stream from the disk against a crystal-derived clock and adjusts the motor speed to maintain a constant bit rate. This feedback mechanism (similar to that of some "quartz-locked" turntables) largely overcomes motor inaccuracies. A good reference on the CD system is: M G Carasso, J B H Peck, J P Sinjou, "The Compact Disc Digital Audio System," Philips Technical Review, Vol. 40, No. 6, 1982. -- --Steve Correll sjc@s1-b.ARPA, ...!decvax!decwrl!mordor!sjc, or ...!ucbvax!dual!mordor!sjc
ark@alice.UUCP (Andrew Koenig) (08/16/85)
1. Are there really more than one laser in some of these things or is
it true (as someone has said) that one laser is used and its beam is
split (optically?). Furthermore, what do the additional beams/lasers
give you that you don't get with one?
Probably. What you gain with multiple beams is resistance to gunk
and scratches on the surface. The bits are recorded as tiny changes
in a reflective surface a short distance below a transparent layer.
If you have, say, three beams, each entering the covering layer at
a different angle, and you use a 2-out-of-3 vote to decide what the
bit should be, then perhaps a surface defect that blocks one of the
beams will still let the other two get through.
2. What is "oversampling"? It seems that every sample on the disk
should be read by all machines, is oversampling the re-reading of
a single sample multiple times, perhaps to confirm it's value?
And if that is the case, what happens to that sample in the event
of a descrepency?
A sample is never read more than once from the disk. However, some
machines use the same value several times running. This is one of
several ways of reconstructing the analog signal: the main other way
is to use each sample only once. The latter is conceptually simpler,
but requires extremely high-precision analog filters to work. These
are hard to make and may drift with time. Some people say they can
hear differences between the two schemes. I can't.
3. "Error Correcting Code" This is probably the most obtuse point to
me, does "Code" imply that there is a CPU inside these things deciding
what to do in case of a disk error? How do different manufacturers
ECC differ?
Every disk has redundant bits recorded on it in order to make it possible
to mask defects in the surface. These bits are recorded in a standard
format and are based on the same theory as error-correcting codes used
in various other forms of computer memory. Where players differ is in
how much of this redundant information they use. There are several
things one can do: use multiple beams, look at some or all of the ECC
information, interpolate when the signal is lost for a short time,
and so on. The intent of the ECC codes is to reduce the rate of
genuine errors to less than one per month of continuous playing.
4. Output filtering, digital versus analog. I think I have a reasonable
grasp on this section. Personal preference seems to be the main
discriminator since both types of filtering can do the job.
This is closely related to the questin of oversampling. See #2.
5. Manufacturers: Apparantly there are two designs used, Phillips and Sony,
which are manufactured by a limited number of companys. These companys
products are then marketed under many names. Could someone come up
with a semi-comprehensive cross-reference list of who makes what
machine behind the trim and which design they use?
I'm quite sure that other manufacturers are making innards, too.
I don't know who buys from whom. I'm sure the situation changes often.
6. Moving parts: Does the disk itself spin or does the laser beam
move? If the disk spins how fast does it go and how accurate need
the motor be? Is this a point between which one can discriminate
a "good" cd player from a "bad" one?
The disk spins. The stuff on the inside, near the hole, gets played
first, and the player then works toward the outside. The reason for
this decision is that manufacturing defects are more likely to be near
the outside, and disks usually aren't completely full. Thus playing
from the inside out reduces the defect rate.
The rate ranges between about 200 and 500 RPM depending on where on
the disk you are. The information density is constant per unit
distance along the track, so the disk starts spinning quickly and
slows down as it goes. The speed accuracy of the motor is unimportant,
because the bits from the disk go into a big buffer and they are clocked
out at the other end by a crystal oscillator. The motor speed is controlled
by how full the buffer is. The buffer is always necessary to allow
error correction to be done. If the motor is so inaccurate as to allow
the buffer to empty, the result would be gross aberrations in the sound
(like the thing stop playing altogether, or at least cutting out for
short periods).smb@ulysses.UUCP (Steven Bellovin) (08/16/85)
> 5. Manufacturers: Apparantly there are two designs used, Phillips and Sony, > which are manufactured by a limited number of companys. These companys > products are then marketed under many names. Could someone come up > with a semi-comprehensive cross-reference list of who makes what > machine behind the trim and which design they use? I don't believe that the lead-in statement is correct. Rather, initially Sony went with analog filtering (see Andy Koenig's answers) and a 16-bit DAC (digital-analog converter), whereas Philips -- the co-inventor with Sony of the CD -- went with 4x oversampling, digital filtering, and a 14-bit DAC. On the more recent Sony models, they've switched to 2x oversampling with a 16-bit DAC. Why 14 vs 16 bits? Cost, for the number of samples you're feeding in per second.
rfg@hound.UUCP (R.GRANTGES) (08/17/85)
[] Go to any library and look in the index of any one of the following mass market (sort of ) magazines: Stereo Review High Fidelity Audio There are many fine explanatory articles therein written for people just like yourself (and for much more simple minded ones, indeed). Also, as its within arms reach, let me suggest the (pause)... June, 1985 issue of Consumer Reports Magazine which also has a simplified explanation. Let's not burn up the net with stuff you can easily find by the ream. Thanks. -- "It's the thought, if any, that counts!" Dick Grantges hound!rfg
mohler@drune.UUCP (MohlerDS) (08/17/85)
Andy, In your note you fail to mention the key differences and reasons for digital filter - oversampling vs. analog filter - non oversampling! I am replying to add some facts to that question and ask for the netters to talk more productively on the net about CD players - you know, discussion where we all learn rather than some of us wanting to have the last word - wouldn't that be nice! The digital filter - oversampling approach allows you to spread the quantization noise over 4x the bandwidth yielding 1/4 the effective noise! It also allows you to use a cheaper or better quality 14 bit DAC instead of a more expensive or lower quality 16 bit DAC for approximately the same performance! A digital filter won't cause all kinds of ringing like an analog filter will! This is a key issue since a lower cost system has much more potential for preamp distortion due to slew-rate or speed related distortions than a high priced system, making it even more important that a low budget buyer of a CD player pick a digital - oversampled CD player! For those that aren't familiar with speed induced distortions: the higher the input level and frequency the better the preamp will have to be to avoid distorting the signal. Also an active filter is more difficult to manufacture than a digital filter! The frequency response of an digital - oversampled unit will be flatter than an analog - non-oversampled unit and it won't vary (as much) with time, temperature, humidity etc. So, you can get a better manufactured or cheaper digital - oversampled CD player with lower noise, less risk of slew-induced distortions - no ringing, and flatter frequency response (other than with a one clock - beat frequency avoidance scheme) or pick the analog - non-oversampled approach! The variation in response is audible and measurable on an analog - non-oversampled unit as is lack of filter alignment, slew rate distortions, noise level providing you know how to measure and listen for (using test disc's) these problems. If you don't believe this, borrow a scope and a test disc and look and listen since if you disagree you have either: A) haven't measured players B) listened and used music instead of test tones C) don't care D) other - I don't accept that you disagree since it has been proven! By myself and by Audio experts in labs that tried such tests and yes the levels were matched and the results were statistically significant. Also, lets say you don't care, if it is this hard to hear and measure the differences, you still get either a better built or cheaper unit with a digital - oversampled system!! For the record: 1) Some of this is an oversimplification for those that simply want to buy a unit without making a career out of it. 2) I have spent the last 3 years working on digital signal processing and digital circuit design and have measurements to documant what I say is true as does N. V. Philips! 3) I think that rather than people wasting time with this argument of what amounts to opinion, it would be far more productive to talk about facts! It would be far more informative and interesting to discuss open unsolved issues like: Has anyone started looking at rise time limitations in a CD player due to the 44KHZ sampling rate vs. the psychoacoustic fact that the rising edge of a wave-form (ictus) has a definite bearing on audible cues that make a recording sound more or less like live? Or has anyone started looking at the fact that the only way to realize the potential of the CD player is to never use a low speed tape medium in recording due to frequency modulation distortion, modulation noise and massive phase shifts induced in such units? Or has anyone looked into how much the CD players tendency to distort lower levels more than high levels degrades the sound potential of a CD player (if any?). I SUBMIT THAT THIS WOULD BE MORE PRODUCTIVE THAN BEATING THE XXIV OUT OF THE SAME OLD DISAGREEMENTS!! 4) I DO NOT MEAN TO DUMP ON ANDREW, I SUGGEST THAT MANY OF THE CD PLAYER POSTINGS HAVE BEEN INACCURATE, OPINIONATED, OR UNINFORMED AND WE COULD BE LEARNING SOMETHING RATHER THAN WASTING TIME. I MERELY CHOSE ANDY's ARTICLE TO POINT OUT ONE OF MANY INACCURACIES AND MAKE A PLEA FOR BETTER MORE EFFECTIVE DISCOURSE. David S. Mohler AT&T - ISL @ Denver drune!mohler or druxu!mohler