lauren@RAND-UNIX.ARPA (07/19/85)
Actually, "head" motion is not that major a problem in terms of overall speed, but is still significant. Consumer CD machines are spec'd for about four second access, while Sony's professional machines are spec'd for around two seconds. Part of the way Sony achieves this is by moving the disc, not the optical head assembly, on their professional units. There is a lot of very finely aligned optical equipment (lenses, mirrors, prisms, etc.) that must sit very close to the disc at the exact track being played. They (or the disc) can only be moved so fast safely. Also very important is the significant settling and search time associated with these units. CD's are actually amazingly "imprecise" in some ways. The center spindle hole is always somewhat eccentric from the exact center of the disc. The surface of the disc is much less uniform than is actually required for the primary lens to be within the correct focal distance at all times for playback. There are a variety of other variations as well. To get around all this, there are all sorts of servo loops that are constantly riding the lenses and head/disc assembly up and down and right and left to keep the disc in "focus." Whenever you move the assembly more than one track, it takes a finite period of time to re-achieve optical lock. Even more critical is the fact that to seek on a CD you basically must take the time to read the disc as you move along--there's no simple servo track you can latch onto to exactly determine your position independent of the normal reading cycle. There are tricks that are pulled to try speed up this process--but those very tricks are what brought us to the point of 2-4 second seeks, instead of much longer ones. The whole CD system was based on the concept that you didn't need the kind of high-speed positioning accuracy of magnetic media systems, since you could instead take your time and read the disc as you went along for positioning and address info. The limitations of the optics are very significant. I don't doubt that it's possible to improve somewhat on the two second seeks of the Pro players. But I don't expect the kind of speeds that would be required to make CD's really competitive, in terms of seek times, with conventional hard-surfaced magnetic media. --Lauren--
herbie@watdcsu.UUCP (Herb Chong [DCS]) (07/22/85)
sony's new consumer audio player 650ES has a seek time of approximately 1 second. supposedly it uses the laser module from the portable and a much higher gain control system. Herb...
mjs@eagle.UUCP (M.J.Shannon) (07/22/85)
I, too have been looking at CD-ROMs and CD-RAMs (coming in 1-2 years; some are
available now), but I see them not as an active disk, rather as an archival
store (like we currently use magtape). For archive storage and retrieval,
these media would seem ideal. The relatively long access times are of little
importance in this application, but the capacity and media lifetimes are ideal.
To give you an idea of the capacity, I present some `back of the envelope'
calculations:
75 minutes playing time (upper limit on current audio discs)
60 seconds/minute
44100 samples/sec
2 bytes/sample
2 channels (stereo)
----- (multiply)
793800000 bytes/disk
or approximately 760 Megabytes per disk (that's 76 10-Meg winnies, folks)! The
upper limit of 75 minutes is due to the higher level of surface defects on the
outer edge of current discs, but I'm betting on technology to clean up that
surface to make the outer inch fully usable (for the record (disc?), the audio
discs I've listened to that approach the upper limit have no errors). Also,
the calculation above doesn't count the 6 bits of ECC code per sample (i.e., a
16 bit sample is represented as 22 bits) on the grounds that a data application
would need at least as much ECC. My memory is a little hazy on the numbers
cited, but I believe them to be accurate.
--
Marty Shannon
UUCP: ihnp4!eagle!mjs
Phone: +1 201 522 6063brando@linus.UUCP (T. J. Brando) (07/23/85)
In article <9@brl-tgr.ARPA> lauren@RAND-UNIX.ARPA writes: >Actually, "head" motion is not that major a problem in terms of >overall speed, but is still significant. Consumer CD machines are >spec'd for about four second access, while Sony's professional >machines are spec'd for around two seconds. How is it, then, that Sony's (unprofessional!) CDP-302 has an access time of less than 1 second! (This comes not from specs but from direct observation!) -- Thom Brando {allegra,decvax,ihnp4,philabs,utzoo,uw-beaver,...}!linus!brando.uucp
rap@oliven.UUCP (Robert A. Pease) (07/24/85)
> I don't doubt that it's possible to improve somewhat on the two > second seeks of the Pro players. But I don't expect the kind of > speeds that would be required to make CD's really competitive, in terms > of seek times, with conventional hard-surfaced magnetic media. > > --Lauren-- Lauren, I don't dispute what you say, but the above statement reminds me of a design engineer I knew at one of my previous jobs. He told me in no uncertain terms that there would NEVER be a 64K memory chip on the market because 64K was the maximum addressable space of a microprocessor and you had to leave room for a power on ROM. Now maybe there never will be a CD that can compete with hard-surface magnetic media, but how much you want to bet that after these things get popular the R & D will provide about half second access times max. -- Robert A. Pease {hplabs|zehntel|fortune|ios|tolerant|allegra|tymix}!oliveb!oliven!rap
lauren@RAND-UNIX.ARPA (07/24/85)
There is certainly work going on at trying to speed them up (particularly the NEWEST consumer models, since people have been bitching) but the exact seek times will still depend heavily on TOC contents. With some contents, you can run in what's called "high-speed-kick" mode throughout the search. With some you can't, and you end up approaching the worst case situation--which presumably is what we're most concerned with. Of course, you indeed COULD put multiple lasers in the thing and multiple head assemblies if you're willing to pay enough for them--presumably some fanatic will insist on doing this at some point. Note that a typical popular music CD is little more than half as long (if that) as the disk capacity. Few discs (comparatively) run over an hour (manufacturing costs go up past 60 minutes). For typical length discs (even regardless of exact TOC timing which affects whether you can use slow or high speed kick during any particular search) there's a lot of excess capacity with which you need not deal (equiv. of having a "narrow band" of data). Classical music discs with only a few bands may be longer, but the lack of dense banding allows for faster access through precalculated high speed searches. --Lauren--