Original Page

What Hi*Fi "Digits is digits- well almost..."

Reproduced from What Hi*Fi magazine, March 1999, page 49. Copyright exists with the original author and / or What Hi*Fi magazine.

My comments in red italics. -E. Woudenberg


Digits is digits- well almost...


Our recent reviews of various CD-RW and Minidisc recorders have attracted a flurry of correspondence, much of it via e-mail, denouncing the results we reported. 'How can different CD-RW discs/Minidisc blanks/CD-RW recorders sound different?' they howl, adding, 'surely the machine/disc combination either records the ones and zeroes or it doesn't. After all, different floppy discs don't make word-processor documents read better or worse, do they?
It's hard not to argue with that last bit of logic. But experience has taught us that, just as different CD players impose their own sound on a recording, so digital recorders, and even the various brands and types of blank media available for them, can make a difference. Don't take our word for it - even our on the couch crew heard differences when they listened back in January. The recordings they auditioned were made on the same brand of disc, with the same source component spinning the original recordings and the copies played back on the same CD player, and our panel didn't even know which equipment they were listening to.

So what's going on? Digits are either there or not, right? The answer to that is 'kind of...' since all digital systems rely on error correction to get the sound from the disc to the analogue outputs in a recognisable form. The less hard the correction systems are having to work to reconstitute the original sound, the better the reproduction becomes. What? These error correction systems all operate in the digital domain. So long as all errors can be corrected (that is, so long as the block error rate stays below a fixed threshold [i.e. 220 per second for CD's CIRC and MD's ACIRC systems]) then the final, error corrected sample values (i.e. numbers) will be the same no matter how ``hard'' the calculation was. If there is some failing in the error correction system, this should be easy to demonstrate by comparing sample values produced by different discs. In any case, I cannot take as serious an experiment that tries to compare two binary files soley by ear.

What's being corrected is faults in the data, caused by anything from scratches on discs to mistracking of the laser pickup, from fluctuations in disc speed to wobbles in the spinning disc, and from low reflectivity causing misreading to vibrations caused by someone walking across the room. So far, so good And that's before you get into electronic failings such as jitter...

Trouble is a CD or MD player looks dead simple: you bung in a disc and it just records then plays music, just like these words are about to be saved to a hard disc and later they'll come back on the screen exactly as they were typed.

But the fact of the matter is that CD players, and digital recorders, are all about high precision engineering operating in a fairly hostile environment. True. For example, the laser pickup system in a CD player, or the write-head in an optical or magneto-optical recorder, needs to move in three dimensions, alter it's power and focus, and deal with a disc spinning at a constantly changing speed, and do all that on a microscopic scale. Fine. Thus anything that can make this task easier, be it discs with greater reflectivity, more even spirals of pits, or even a more consistent optical layer, is likely to give a better sound. Nooooooo! Why do we skip this important point: CD and MD readout systems [before and after error correction] produce numbers, not sound! Get the numbers right (and the clock in certain cases) and you are done! If you want to discuss D/A converters you can certainly have a great all night party, but you cannot argue about the quality of numbers that come from digital audio systems. They are abstract entities, a one is a one is a one.

No, all digital equipment doesn't necessarily sound the same, however much logic might suggest otherwise - in fact, it's a miracle most of it is so consistent...

Please, What HiFi, invite some experienced enginneers over for a demonstration of this effect. Most I know would love a chance to investigate such a wild anomaly!

There's more to those noughts and ones than meets the eye.

Aye! But not the logic analyzer!

Further fun from What Hi-Fi, December 1998, Page 91


We used a variety of discs from the major manufacturers - Maxell, TDK and Sony - throughout this test, and have to report that the differences we heard, while noticeable, are much less than between, say, different audio cassettes.

For most of the recordings we used a mix of standard and premium discs. The differences within the Maxell and TDK ranges were hardly mind-blowing - - in both cases the mainstream `colour' discs performed extremely well, revealing differences between mains and personal machines clearly.

Switching to the upmarket gold-coloured discs from the two companies brought a little extra focus and clarity and a shade more bass weight, but the differences were small enough for most of the personal recorders to overlook them - they only became really apparent when the discs were fed into a recorder like the Sony MDS-JA20ES, and even then the quality from the TDK MD-RXG and the MAxell Gold MD was only on a par with the standard Sony disc.

We also had a brief play with Sony's expensive ES MiniDisc and, yes, the ES disc does bring out the best in machine slike the upmarket Sony and TEAC recorders, but otherwise we'd stick to the standard discs for go-everywhere day-to-day recordings.

Return to the MiniDisc Community Page.