John Phillips wrote:

On 2005-06-14, Arny Krueger wrote:
The bestest paper about jitter is IMO Eric Benjamin and
Benjamin Gannon - "Theoretical and Audible Effects of
Jitter
on Digital Audio Quality", AES preprint 4826 105th
convention, 1998.

I have been looking at papers on audio jitter but I have
seen no clear
consensus yet on the threshold of audibility.

I don't think you'll find a consensus among many sources
because there isn't a lot of hard data, bu you will find
some credible hard data in the Benjamin and Gannon paper.

I have not read the above paper but I have seen the graphs
of
jitter audibility it contains (they are in one of the
articles he

http://forums.audioholics.com/forums...d.php?p=25135).

Post number 8 reproduces data from the Benjamin and Gannon
paper.

The graphs suggest that for 8 kHz and 20 kHz sine waves
the audibilty
threshold is at worst about 3 ns. However I have seen
curves in other
papers (I probably can find them if needed) which show the
audibility
threshold dropping down to 0.1 ns or less as frequency
increases to 20
kHz. That's 30 times worse. I believe that achieving 3
ns is
trivial, but 100 ps is much more difficult.

A 20 KHz signal has a period of 50 microseconds, so 3
nanoseconds is a time shift of 0.00006 of the period of the
20 KHz tone. This amount of FM modulation creates sidebands,
the largest of which is about 85 dB below the carrier. This
is relatively poor jitter performance

Again a 20 KHz signal has a period of 50 microseconds, so
..1 nanoseconds is a time shift of 0.000002 of the period of
the 20 KHz tone. This amount of FM modulation creates
sidebands, the largest of which is about 113 dB below the
carrier. -113 dB jitter performance is very good
performance, but it's not exceptional.

For example, the SHAC-300 Dolby digital decoder had -120 dB
jitter and the LynxTwo has -133 dB jitter.

In any case, quality audio DAC chips seem to have had
interpolating
oversamplers in them for the last few years, with the
capability to
separate the DAC master clock from the recovered data
input clock, so
a competent modern design should not present any jitter
audibility
issues as far as I can see.

Agreed.

Even beforehand a good DAC designer
could have had a two-stage PLL with a wide bandwith in the
first, a
FIFO in between and a narrow bandwidth in the second to
provide
substantial jitter reduction.

I don't think that heroic measures are required to product a
digital device with inaudible jitter. The Benjamin and
Gannon paper criticized the technical and audiophile press
for making a mountain out of a molehill.