Digital Source Comparison - An audible difference?

In addition to sonic differences that may or may not result from differences in transport-generated jitter (depending on the amount of that jitter and on the jitter-rejection capability of the dac), if the connection between transport and dac is via an electrical cable you may or may not hear differences depending on:

1)The length of the cable, and the happenstance of the relation between that length and the risetimes and falltimes of the output signals of the transports. Those risetimes and falltimes are usually unspecified. See this paper for further explanation. In other words, you may hear a difference when one cable length is used (with both transports), and not hear a difference when another cable length is used (with both transports).

2)The amount of groundloop-related electrical noise current that the transport and dac collectively may cause to flow through the shield of the interconnect cable, which in turn will affect jitter. That effect is, again, unpredictable happenstance.

3)The degree of mismatch between the output impedance of the transport and the impedance (more properly, the "characteristic impedance") of the cable. That will affect waveform integrity, and may thereby affect jitter.

4)The degree of mismatch between the input impedance of the dac and the characteristic impedance of the cable. That will affect waveform integrity, and may thereby affect jitter. That effect on jitter may be sensitive to which transport is used, because the consequences of the waveform reflections caused by the impedance mismatch depend on the other three effects. In other words, an impedance match between dac and cable which is less than perfect can cause differences to be heard between transports.

Toslink is not susceptible to any of those effects, but I have no reason to doubt what seems to be the prevailing consensus that more often than not it will provide inferior performance compared to electrical interconnection.

None of those four factors have any strong correlation with component quality or component price. Transmission of digital signals between components should be viewed as a system-level issue, which is dependent on interactions between both components and the cable connecting them.

Regards,
-- Al

I presume you are talking about 44.1K/16 bit native output from the players. For that frequency and bit depth I doubt if toslink will show any difference with coax, assuming you use glass fiber toslink and not plastic optical fiber.

If upsampled digital output from the players is used, all bets are off. I did this with a CA 840c (feeding a Bryston BDA-1) and found significant loss in sound quality when using the upsampled outputs (up to 24 bit/192K)of the 840. This was compared to the 840 alone feeding the same pre-amp/amp as the Bryston. At the high frequencies and bit depth the toslink would not allow the Bryston DACs to lock, whereas coax did. When I sent native 16 bit/44.1 KHz from the 840 to the Bryston, sound quality improved significantly.

Al - I can understand the concept of reflections in a digial cable, but is there concrete evidence that these reflections will actually cause an error in the transmitted data? Can the reflections actually cause the "pulses" to be outside of the acceptable range? I would assume that the DAC is looking for specified ranges with some tolerance applied.

Hi Mceljo,

Assuming that the impedance mismatches are within limits that are not totally unreasonable (such as might occur if the connection of the shield in a coaxial cable were broken), the resulting reflections will not cause any errors whatsoever in the digital data as received by the dac.

The issue is that the waveform distortion caused by reflections, in combination with noise that will invariably be riding on the signal to some degree, will result in small but significant fluctuations in the TIMING at which the dac senses the transitions between bit intervals. That is what the term "jitter" refers to. Depending on the jitter rejection capabilities of the dac, that can and often will have audible consequences, even though all the bits will be correct. See this paper for further explanation.

Best regards,
-- Al

Al,

I passed the two main links along to an EE friend of mine that's my sounding board for all things electrical. Here's part of his response.

"Those are the first articles you've sent me where I completely agree with everything written. This guy is obviously a well-educated electrical engineer.

How noticeable corrections to these factors will be, I do not know. I don't believe that crystal oscillators used in today's computer equipment are much farther off timing-wise than a belt-driven turntable."

In any case, he generally give me a list of things he considers to be "bunk" in the things that I pass along.