Digital Drive: RE: How can a transport make such a big difference? by Thorsten

Hi,

It must be understood how Audio CD (as opposed to data CD) works.

It includes a basic CRC Code (to check data integrity), but no data redundancy, which would allow misread data to be reconstructed. If a "bad" word is detected it is "interpolated" from the last good and next good word, there is no attempt to re-read the data (as it would be with a Data CD) or reconstrut missing data.

The CD is a thin sliver of plastic and can deform as a result of being exposed to external acoustic waves, other vibrations or even tolerances in the drive mechanism.

While the laser system can track some of this, there are limits. The typical low cost linear drive systems with geared systems include "slack" and are especially limited. The old Philips Swingarm mechanisms were much better in this respect.

Once the CD is enough out of focus, the data will read incorrectly. Data will be interpolated. When we did the AMR CD-Players we had to look into this in more detail than I cared to learn about.

This explains the result of many mechanical tweeks, be it Teak VRDS or Pioneer Stable Platter mechanism, C.E.C Belt drive with a huge heavy clamp, they cancel many of the mechanical issues and prevent mis-read data and thus "interpolated data" which is not what was there originally.

The second element is the clock. A crystal oscillator can be "pulled" away from the nominal frequency a considerable degree by changing the power supply voltage.

Many transports and CD-players power the clock from the power supplies that also power the CD Processor IC, and even motors and servos. You can see the modulation of the clock power on a good oscilloscope, it's enough to mess up the jitter performance of the clock, even if with a clean power supply it would be adequate.

So the clock should have a really stable, low noise power supply that has at least a separate winding, better separate transformer to power it, so it is isolated completely from the rest of the circuit.

The third factor is the actual SPDIF Output stage. The quality of these varies widely. Actual output impedance, stability of output impedance with frequency, cabling etc. make a big difference on the recovered jitter.

Ideally the SPDIF signal is reclocked directly at the output jack using our super clean clock and then suitably adapted to correct levels and impedances within a few mm of the BNC jack (RCA Jacks are not as good on impedance match).

Which brings us to cable and DAC. Ideally a DAC should be designed to reject all source clock jitter on the SPDIF input. It is possible to do this, but rare (it's somewhat non-trivial, having done it several times on different hardware platforms).

In reality receiver chip, actual physical interface, connectors etc. play a role. Impedance mismatches cause excess jitter on the recovered clock, as does degraded waveform etc. Some times a cable with a "wrong" impedance actually works better than the correct impedance, because it actually matches the real circuit better.

So, in something that should be "in theory" a reliable and purely digital "It's all numbers right?!" pathway there is a huge amount of actually ANALOGUE and MECHANICAL problems which if unaddressed mess up our pure numbers.

Before switching to PC and USB as digital source (which has another set of problems, though most much less severe than SPDIF) my preferred transport was a modified Pioneer first generation DVD Player.

The DVD Player stored the data fromm the CD in RAM, it was reading the CD to immediately fill the RAM buffer to max and was able to attempt re-reads in case of CRC errors. This dealt with most mechanical issues in a cheap and easy way.

The output in the original design was reclocked and was transformerless and buffered by a discrete buffer, basically a video circuit almost exactly the same as the analogue video outputs and the waveform was exemplary.

The modifications were a new "super-clock" (for 44.1kHz CD Replay only) with it's own transformer and power supply and a new analogue stage (discrete, LC Audio) also with it's own power supply and transformer for the on board Legato Link DAC which I rather liked the sound and BNC connectors for Audio & Video.

It was the best transport for CD I found, and it ruled at many an audio meet. IIRC the unit new was 399 UK Pound, but I bought it discounted. The Video performance was also rather nice. For 480p SD TV.