Digital Drive

Posted by ppopp (A) on May 31, 2006 at 20:39:35

Letter and answers from Hi-Fi World's website.

DP = David Price (editor)
NK = Noel Keywood (publisher)

TRANSPORT 2006
Can you explain why some CD transports sound better than others, and why some chose CD mechanisms designed for audio and others CD-ROMs or DVD-ROMs? According to a lot of forums the choice of mechanism should have no affect on sound quality and is therefore unimportant, as the mechanism produces no jitter or distortions and any errors are corrected by the error correction circuits? I am now a little confused with all this, I thought mechanisms had measurements and were not all equal?

Why is there VRDS, Stable platter, Philips Pro, etc., if CD-ROM and DVD-ROM give the exact same performances ? and if there are differences then in what ways can these be detrimental. And why do the CD transports I use sound different, is it some sort of placebo affect? Is a £700 DVD player as good as a £500 CD player or transport, why does my VRDS sound better than my Pioneer? What answers can I give to back up my claims of hearing a difference?
Louise

Fine questions! A short history lesson, if you will. Back in the early eighties, CD was launched and in a bid to make it as reliable as possible, most early transport mechanisms (or ‘mechs', as the trade calls them) were seriously over engineered, using metal parts extensively, often including a metal disc tray. Philips, in its early transports, also used a ‘swing arm' laser, which tracked across the disc in a way rather akin to a tonearm on a turntable. All these various things, ‘swing arms', metal diecastings, etc., etc., conspired to produce what many regard to be superb transports. The later Philips CDM9 and 12 mechs used in nineties machines, for example, didn't seem to sound half as good as the early ones, and seem built in a dramatically more flimsy way.

One exception was TEAC, which in the early nineties introduced the VRDS mech; as usual this came in cheaper variants (such as the VRDS-T1 transport) and more expensive ones (such as the P-30 transport), but in all cases they were excellent and seemed to be getting better and better whilst the Philips mechs got worse. An interesting point is that, whilst Philips stopped supporting their mechs with spares a year or three after they'd discontinued them, TEAC continue to support all their mechs to this day ? something to hold in mind when buying secondhand.

In the mid to late nineties, there was a revolution in transports; suddenly the key driver to sales became the computer market, which had a massive demand for data ‘CD-ROM' readers. This pushed costs right down, and certain manufacturers like Audiolab and Meridian, to name but two, moved into ROM transports. At this time, the accuracy of the master clocks became an issue, not least because manufacturers were now able to measure 'jitter' (i.e. time domain errors). Much was talked about on this subject, but suffice to say that high jitter mechs generally produced a vague and hard(er) sound. At last, here was something that explained the differing sound of transports!

Well, this is what many thought; but it's not the whole story. There's no denying that low jitter is where we want to be, but two transports with very similar jitter figures don't necessarily sound the same. It's possible to get very low numbers from a modern reclocked DVD-ROM, but to my ears at least, it wouldn't hold a candle to a late eighties TEAC CMK-3.2 or Sony KSS151A with absolutely no reclocking. How so ? it's down to serious build, fine tolerances and servo software designed specifically to read Red Book CDs in real time (i.e. 1x), rather than Microsoft Encarta DVD-ROMs at 48x… So where does that leave us now?

This is now a real live issue; I really don't think it's ideal to use cheapo 'parts bin' ROM mechs (and reclock them) in high end audio applications. True, notionally they work reasonably well, but I fear that designers are only doing this because of price and availability considerations, with sound a distant third? One serious high end player like the Linn Sondek CD12 uses a bespoke mech with a stunning metal disc tray that 'swishes' in and out, while another such as Meridian's 808i uses a proprietary DVD-ROM mech with a flimsy plastic disc tray that whirs in an out like a cheap PC you'd buy from Micro Mart. I think buyers should 'vote with their feet', on this...

I've been doing some serious transport listening tests of late, concluding that they do sound different ? and that the difference is not always a subtle one. Generally, well engineered bespoke Red Book CD mechs from the eighties sound deeper, more three dimensional and musical than any of the new breed of clocked ROM drives. Although the latter can be very detailed across the midband, I find them quite two dimensional and lightweight at frequency extremes. Any of the big Jap battleship players of yore (Pioneer PD91, TEAC VRDS25, Sony CDP-X77ES) show the latest mechs a thing or two about bass…

How so? Well Louise, all I can say is I don't know. I am not a transport engineer, and so can only report my findings. Rest assured, however, that I'm putting all of these thoughts and experiences together, and will be researching a feature on this very subject soon. In the meantime, any readers' experiences on this subject are welcome.
DP

There's much debate about this. The traditional reductionist engineering view is that 'bits are bits' and digital is, by its very nature, perfect. This is one view, a rather simple one, that is still floating about. It is still vigorously defended by those who need a simple belief system.

Luckily, many digital engineers nowadays acknowledge that there are indeed differences in the sound quality of digital devices like transports, discs, tape machines and what have you. They can be quite large too. In response to this a fair bit of hypothesising and measurement has been used to try and pin down what is happening, so far with limited success.

Initially, errors and error correction were invoked, and indeed they can be substantial our measurements show, especially with poor CDs possessing pin holes in the reflective layer (you can see these by holding a CD up to bright light).

Then jitter was brought into the picture as a likely culprit. It certainly does affect sound quality, adding to that peculiar coarseness and greyness of tone of digital, but it isn't the only problem. Many engineers feel that re-clocking eliminates it, whilst others feel it only disguises timing problems, by re-translating them into another domain, so coarseness becomes vagueness or imprecision for example.

Then there's noise. Digital systems generate unpleasant forms of noise as a result of jitter and quantisation error, laser read error and what have you. This may well affect matters. We accept that noise in the form of dither smooths digital by randomising errors such that they cancel, having a remarkable affect upon low level signals our measurements show. But there is, subjectively, 'smooth' noise, like pink noise for example, and 'ripping' or 'sharp' sounding noise and it could, conceivably, be the presence of different types of noise at low level that are introducing perceived differences. High quality transports, like those from Teac, reduce all forms of error and noise, and this does translate through to better sound quality. By what mechanism we are not sure, I believe I am right in saying (there's always something I haven't read !).

I suspect it is the sum total of all these effects, plus others we don't know much about perhaps, that influence what we hear. I always recall that Yost, of the Parmly Hearing Institute, claims we can hear down to the movement of a hydrogen atom! The human ear is capable of remarkable sensitivity and resolution under certain circumstances. At present a lot of effort is being put into digital technology, yet we know at least one highly qualified expert in the field who claims it will never sound good until we use sampling rates far higher than today's. In the meantime we have to live with what we have, which in CD is first generation technology from the 1970s. Happily, a lot of companies and engineers are trying to improve matters and there are some real digital gems out there, amongst the prosaic. NK