SoftwireEngineer;192370 Wrote: > Right, the stock SB3 digital output's jitter is not that great.
Well, actually it's not so bad. I tried to do some kind of measurement, using the test sound pattern first proposed by Julian Dunn. There is a good description of this method on TNT-Audio here: http://www.tnt-audio.com/clinica/jitter1_e.html In the attached zip, you may find some result of this experiment. These are FFT analysis of the analog output of different configurations, when playing this test WAV. Files names are self-explanatory. The ideal system should show a perfectly pixel-narrow vertical line at the 11025 Hz frequency, with two spikes around -130 dB at frequencies about 230 Hz apart. I used a RME Multiface to perform the test, in 24/96 mode, the test file being a plain 16/44. For the FFT, I used the Wavelab software. The sample size is 256K, with a Kayser-7 windowing (this is for reference only, don't ask me to explain why this setup, I got it from help files, I'm not that kind of specialist). I put a copy of the test WAV here: http://cjoint.com/?efarjcVzYW The file is only a 220 KB zip, but it gracefully expands to a 160 seconds WAV. WARNING: do be careful, don't try to play this wav on your hi-fi, most of the tweeters will not be happy with a 0 dBFS 11025 Hz sinus.... There is a close relationship between the RMS jitter and the total area under the curve, around the 11025 Hz peak. Shame on me, I was stupid enough not to save a copy of the stock SB2 curve, so there is some bias in the comparison: a stock SB3 versus a modded SB2. And I'm too lazy to roll-back the mods on this SB2. The D-clock I used is from NewClassD, designed by Lars Clausen, formerly LCAudio. As you may see, there is a perceptible, although not spectacular, improvement when the Lavry DA10 is in the path. I must add that the SB3 tested is from stock, with it's own wall-wart. On the other hand, the SB2 was powered by a workbench linear power supply, but with the lid open, so it was exposed to potential parasitic signal. A wired ethernet was used for all tests. To offer some reference, I added the test result of a Terratec Aureon Firewire, this audio peripheral being credited with good reviews. The driving firewire came from my VAIO Sony laptop. As a matter of fact, this test is not intended to evaluate other characteristics of the system, such as THD, DAC linearity, and the like. >From these tests, I feel it obvious that the stock SB3 is fairly good, regarding jitter. SoftwireEngineer;192370 Wrote: > As another poster 325xi, mentions, I am very suspicious about the > quality of the USB approach. Empirical Audio does mods in this area and > they seem to throw lots of money and tweaks into getting a good quality > out of it. I investigated the USB approach earlier and finally went > with the Squeezebox route. Same for me. The question is: does USB audio devices actually request data from computer? As far as I know, the answer is a big NO, at least for the so-called "class compliant" devices, i.e. those not requiring a driver under Win XP. Instead, all these devices are doing their best to cope with a data flow dictated by the computer. It's not different with Mac or Linux. On the contrary, over the network, the Squeezebox is the boss, summoning the computer to keep filling in an ample buffer. JLM +-------------------------------------------------------------------+ |Filename: Jitter.zip | |Download: http://forums.slimdevices.com/attachment.php?attachmentid=2650| +-------------------------------------------------------------------+ -- jlmatrat ------------------------------------------------------------------------ jlmatrat's Profile: http://forums.slimdevices.com/member.php?userid=10656 View this thread: http://forums.slimdevices.com/showthread.php?t=34132 _______________________________________________ audiophiles mailing list audiophiles@lists.slimdevices.com http://lists.slimdevices.com/lists/listinfo/audiophiles