Originally Posted by Mike_WI
Has any third party meaured jitter in various CE devices with a "Materials & Methods" section defined?
The closest thing to a somewhat widespread measurement is the Miller Audio Research test system. This was developed by Paul Miller, who is now editor of Hi-Fi News and Record Review (UK). He uses the system to measure the equipment they test.
John Atkinson of Stereophile also bought one of the MAR analyzers about ten years ago. So he also uses it. Therefore there is a fairly large database of equipment that has been tested with this system.
However, there are several problems with the system:
a) It uses a test signal developed by the late Julian Dunn called JTEST. It was specifically designed to stress the S/PDIF connection. Nonetheless, it has been used extensively to test many other kinds of digital devices, including those for which it is poorly suited.
b) The test results you get depend almost as much on the test equipment as on the device under test (DUT). The original MAR system used a 16-bit data acquisition card in a desktop computer. This gave relatively poor performance and was later replaced by a 24-bit acquisition card. This newer card gives better results but is still trying to make very low-level measurements whilst inside the noisiest possible environment -- a computer!!
Stereophile added an Audio Precision 2722 to their test arsenal a few years ago and the quality of the measurements improved so dramatically that JA decided to compile the first two years worth of new measurements in one place. It explains the test, how it works, what to look for, and then gives the results in order of performance, with the best at the end. The highest performance came from the Ayre C-5xe, which virtually looked like the computer-generated "ideal" graph at the beginning of the article. You can read it here:http://stereophile.com/features/1208jitter/
You can look at some of the Hi-Fi News test reports if you go to the MAR website and jump through some hoops (like registering):http://www.milleraudioresearch.com/avtech/index.html
c) There is a strong temptation to try and take the results of this test and spit out a single "jitter" number. The MAR test set does this in some secret algorithm. However, the algorithm used has changed over time and the resulting numbers cannot be compared unless made by the same person with the same equipment within a relatively short time span.
On top of that there is a measurement limit of 120 psec for 16-bit data, which really isn't very low. In theory, 24-bit data can be measured to sub-picosecond levels EXCEPT that the JTEST does not look at low-frequency jitter components at all!!! It turns out that the low-frequency jitter components are probably the ones that the ear is the most sensitive to.
So in practical terms, the Miller test as used by Stereophile and Hi-Fi News can tell you when something is horrible. But it can't tell you when something is operating properly versus operating superbly.