The Sound of a Digital Cable: Bandwidth and Jitter

[DonH50]

Yes. Unfortunately, the CMRR is not infinite across all frequencies... The issue that has been seen is basically shorting a noisy PC ground to a sensitive DAC (or preamp) box and injecting noise into the signal circuits that reference a single ground plane. The better boxes use some sort of galvanic isolation. For the record, I do not use an external DAC; I have run into this several times with microphone preamps. Others here have seen/heard problems with USB DACs.

I have seen (measured) USB lines that have fairly high CM noise, but I do not know if that impacts the sound (I do not have a DAC box to measure and it would depend upon the USB receiver design anyway).

 

[bblue]

Yes. Unfortunately, the CMRR is not infinite across all frequencies... The issue that has been seen is basically shorting a noisy PC ground to a sensitive DAC (or preamp) box and injecting noise into the signal circuits that reference a single ground plane. The better boxes use some sort of galvanic isolation. For the record, I do not use an external DAC; I have run into this several times with microphone preamps. Others here have seen/heard problems with USB DACs.

I have seen (measured) USB lines that have fairly high CM noise, but I do not know if that impacts the sound (I do not have a DAC box to measure and it would depend upon the USB receiver design anyway).

Yeah, that computer ground to DAC (or other equipment) ground could be problematic. I've never experienced it (that I know of) but it's only logical to exist.

I wonder, if the +5 isn't used, if the ground/shield could be telescoped from the computer, but not connected to the DAC, if that might improve things?

Generally, I use a powered USB hub at the end of a USB cable where there are devices that use +5, in an attempt NOT to use the computer power source and ground. But I've never checked to see if the USB hub in some way isolates the computer ground from its ground reference. It just seems to work.

The M2Tech and Halide USB interfaces with their very short USB run don't seem to have any issues at all of that type.

Hmmm.

--Bill

 

[DonH50]

Adding length adds inductance that would help isolate HF noise. Not sure about the powered hub (I use one too); I suspect it just shorts all grounds into a single plane. However, that adds another cable, so again perhaps reduces the noise... A simple ferrite bead on the lines (signal, power, ground) would get rid of the RF noise.

 

[Part-Time Audiophile]

Sorry for the I'm-an-idiot questions, but:

If jitter is the thing to address, the primary thing to avoid in a digital cable is a less than 10MHz bandwidth?

How do you find out what the bandwidth is of a given cable?

 

[DonH50]

Jitter is but one thing that can affect the sound, and the magnitude of its impact is debated hotly.

You find out the bandwidth of a cable by knowing and analyzing its parameters, or by measuring it. Cables are not typically specified in terms of bandwidth, though there are guides available. They are specified in terms of impedance and loss. Loss sets how long a cable you can run for a given sgnal level at a given frequency (or frequency range, bandwidth).

Not sure where "less than 10 MHz" came from; the bandwidth needed depends upon the signal prototcol, transmitter and receiver properties, and a whole host of things most audiophiles need not conceren themselves about. You want enough bandwidth to recover the signal without significant errors and with low inter-symbol interference (ISI), but not so much that the wide bandwidth increases random noise and potentially allows RF noise into the system (if not well shielded). However, I have not seen a case where "too much bandwidth" exists in a digital interconnecting cable. The transmitter and receiver will set the bandwidth if the cable does not; the main drawback of a cable with "extra" bandwidth is higher cost.

HTH - Don