Today's Transport War: Significant Differences?

I'll be trying the North Star 12S interface shortly. I have had the transport for a while, and now just bought the DAC. It will be interesting to see how it compares with my existing set-up.

The term "hard-drive" based source is a bit misleading. The hard drive is only a place to store the data. It might be stored in RAM in the future or 3-D magnetic memory. The term should be "computer-based" source.

You can read this paper comparing computer-driven to CD transport:
http://www.positive-feedback.com/Issue22/nugent.htm

In a nutshell, the computer has the capability to "feed" the audio device in any number of ways - streaming at high speed, native speed, bursting or network packets. Many of these data transfer techniques have intrinsic buffering in them, allowing the D/A converter to be fed with a fixed clock rather than a PLL. Doing this with a spinning CD is much more difficult. The Lavry concept is a good one. I'm not convinced that it is jitter-proof however. I have at least one customer that replaced his Audiophile USB with my Off-Ramp Turbo 2 and reported that the sound was more clear and focused driving the DA-10. This should not have been the case if jitter were completely rejected. However I do believe that Lavry's DAC is one of the better designs on the market for jitter rejection, even though I have not heard one yet.

The point is that there is at least the opportunity to generate a non-PLL clock to drive a D/A chip directly with Computer-driven audio. Not so with a transport. You might say: well what about a word-clock driven back to the transport? It turns out that the reality of this is much more difficult than the concept. Most modern D/A chips are not even clocked on the word clock, it is the bit-clock or the master clock. To drive a DVD player for instance, you would need a 27MHz clock from the DAC to the Transport. Then you must divide this down inthe DAC to synthesize the clocks needed for the D/A converter. Then, in order to make this work, the D/A must resolve the phase difference of the Transport signals and the local non-PLL clock. The phase of these signals will change with the digital cable length, delays in the transport circuitry and clock generation etc.. The total timing budget is about 75 nsec for 24/96 signals, assuming S/PDIF interface. The slop in the reading of the bits from the optical head will probably eat-up most of this budget. The only way to pull it off is probably to FIFO buffer the data coming into the DAC to allow it to slop around. I'm not saying it is impossible, just more difficult and involves tight coupling of the transport and DAC. Mixing and matching transports and DAC's would not be possible. The Meitner system got around all of this by using an I2S interface. This is much simpler, but still relies on a PLL.

Steve N.

The Noiseball (aka computer) converts audio data many times, uses questionable performance clocking, has switching mode power supply and uses poor interfaces (mostly USB and Wi-Fi packets or combination of both) to transmit audio data to an external DAC. It does not really matter if the USB signal is converted to S/PDIF, I2S or anything else, it still comes from that same USB port! Same applies for Wi-Fi connection where the middle of the audio track may be sent first and then the beginning of it and so on in random packets which are then decoded and reconstructed by a DSP (talk about error correction). There are some nice pro-audio computer cards available allowing much cleaner data transmission and slave mode to the DAC clock, but please realize that all you slave to the DAC is the clock of the computer card DSP, nothing else.

Even a cheap $149 universal player spins the CD, SACD or DVD-A at higher speed allowing for memory buffering using both FIFO and large SDRAM buffers which results in “jitter free” clocks and “bit perfect” data. It is not true that the DAC in a CD/DVD/SACD player has to be PLL-ed (and what’s up with the “evil PLL” thing? Latest PLL techniques are great achieving as low as 30pS jitter!). Good example that comes to mind would be the famous, faulty and long discontinued Philips SACD1000. This player has its main 16.9344MHz non-PLL audio master clock (low noise powered) next to its DACs, the perfect scenario. This non-PLL, low jitter master clock oscillator is also clocking the audio DSPs. Then, PLL is used to generate the 27MHz clock for the video circuit. The SACD1000 spins the CD at much higher speed and uses combination of FIFO and large SDRAM buffers. The “jitter free” and “bit perfect” PCM data/clock transmission in the SACD1000 is carried by our beloved I2S but without any data conversion or long cables and connectors resulting in about 5-6 inches cleanest possible signal path. WOW, isn’t this amazing! It looks like the perfect CD player, the perfect solution, right? Well, I am sure that many around here know for a fact that the stock SACD1000 performance is nothing to write home about, although it was a nice machine for its time. The moral of the story; not everything is as easy and simple as it looks.

IMO and IME, a carefully optimized (key word), vibration/error free “classic” digital front-end using memory buffering and latest error correction engines, with built-in dedicated, specially designed DSPs, shortest signal path, no data conversions to completely irrelevant to audio data transmission formats, ultra low jitter clocking and low noise power supplies simply can not be outperformed by computer based audio, at least not for now. Hopefully in the future the time will come when proper technology will be developed so one is able to load favorite music onto a computer based audio solutions and play it back while achieving the same audio quality as with the classic dedicated stand alone digital front-end is possible today. So let’s stay optimistic.

Regards,
Alex

Aplhifi: if I'm understanding you correctly, you're saying that a vibration/error free front end as you suggest reads different data from a disc than would a lousy CD-ROM drive in a computer? Has this been tested, e.g. is it possible to somehow capture the digital data from a such a transport and compare it to a standard rip from a CD-ROM drive on a computer? Now that would be an interesting experiment! :)

If they are indeed identical, what then is producing the difference in sound? You stated that jitter is not much of an issue anymore, what else could be a factor if we assume that the DAC is identical?

I'd really welcome a discussion on this, it's so hard for me to understand what the problem really is. It should be so simple I think, but demonstrably it isn't since I've too found significant changes between various CD transports connected to the same DAC using the same cable and playing the same disc. How is this possible?

Thanks!

Grant a friend of mine whose products were on display at RMAF was brought down the Nova Memory Player you mentioned above. He heard this player on his system setup at the show after hours. The Nova player was compared heads up to the 10k player they were using in their listening room..he said he wants to purchase the Nova..The copy he made on the Nova and then played back on the other player sounded much better than the original cd played back on either machine. Electro mechanical jitter error correction all of that the Nova guys claim is pretty much gone.Playback from the memory chip internal of the Nova player is even better yet. The Melos crew is back! I recently purchased an Altmann Dac from Germany. Sounds wonderful. Altmann claims with his circuit redo's that the transport makes little or no difference..my listening comparisons so far are making me listen and look harder at transports differences when played thru this little 12volt wonder.Tom