CatBus;372855 Wrote: > I really don't want to add to a thread that's based on such a > fundamentally flawed assumption, but I do want to address this because > I never want to give anyone the impression that I am personally > attacking them, telling them what they can or cannot hear, etc.
This was not directed at anyone on this forum, or anyone specifically. I talk to a number of industry folk on a daily basis, and well...........we all were about as frustrated the same amount about this subject. You would be surprised how often we hear this lodged against all of us. How I perceive (key word) how something sounds can not be quantified. I can measure things that I think can explain it, but not always. THD can easily be measured, and no one will dispute those results. But how it is perceived by the listener can not. If you could, no one would listen to tube amps. (I don't, so there!) But let's move to jitter, which is the genesis of this. Let me explain why I know (at least to my satisfaction) that I can. This might be protracted, so bear with me. When we first made D/A boxes, we were pleased with the results, but something was not quite right. The initial test had a 50 cm or so coax cable. Just for grins, we put in a 1 m cable. Huh, different. What if we put a different one? Again, different. Not better, just different. So what caused it? We poked and piddled, talked to some industry buddies working on the same type of product. All came to the same conclusion(s). Something did not sound right, and cables changed that around. One buddy, who did not have my background or education level, pointed out that you could listen to the PLL pin on the RX chip, and hear a distorted version of the music. Huh. I designed PLLs. Why didn't I think of that first? Anyway......guess what...............you could hear it! Wow, what if I listen to it with various cables........... What if I then measure it with various cables? Moving ahead a few steps in the measuring process, we found that everything affected that noise of the PLL pin. Cable length and impedance, type of TX and RX circuit, you name it. Not only could we measure it, relative to the "stock" configuration, we could hear the difference. And it wasn't just me and the rest of the gang at the shop. We found that *everyone* could hear the difference. And not just at our shop, on our gear. At anyone's house, and on their own gear. (OK, we had to stick our D/A in, but we used different sources.) So, to test out our latest theory, we made a special D/A box. We took one input, and applied all of our tricks to it, and left the other one stock. We schlepped it everywhere that we could. Everyone heard the difference, and heard the exact same difference. No one, outside of myself, had any idea what I had done, or what it was supposed to do. I knew what I had done, and thought that I knew how it would affect the sound, but all I can really do is speculate. It needed verification. I was able to do 2 things: 1.) When the return loss on the system was down around -30 dB, there were no further differences that could be reliably discerned. At this point, all cables and what not all sounded the same. 2.)I could measure the effect on the jitter on the recovered clock. At the levels mentioned above, it was not possible to pick out any differences in the recovered clock's noise. Mind you, this was before we made any attempt to reduce the jitter by futzing with the PLL itself. -Even with the horrid amounts of jitter still in the chain, any listener could hear the reduction of jitter.- So, we may not have reached the level at which jitter can no longer be heard, but we have found that even in the presence of large amounts of jitter that slight reductions could be heard. So, what was the next step? Build a secondary PLL. Get the jitter down much lower. Making a unit that could switch back and forth was not practical, so we took 2 identical units, and rigged one up with the secondary PLL. With a jitter level that we could measure. And compare it to the transport clock. At this point, we found out that the transport clock was not adequate. (We were shocked by how bad they were.) So, more futzing and measuring. And listening. Again, the lower we got the jitter, the better things sounded. Also, the changes that we made were discernible by anyone. Until we got down the point that we were running into the limits of our efforts. Yes, as we got lower to the floor, it became harder to detect the differences. Not everyone could hear the changes. Having done this now for over 15 years, I can confidently say that the jitter does need to be significantly lower than 100 pSec, for a 256 Fs clock. We initially thought that would be good enough, but we continued our research and were proven to be wrong. That really wasn't good enough. It is indeed important to point out that Gaussian noise affects on jitter were not as detectable as the data-correlated jitter. So, lowering the jitter at the transport end had less of an effect than on the SPDIF end. Unless the jitter on the transport on the transport was data-correlated.............which is usually is. But that was easy to fix: just get the clock out of the filter chip! So, there you have it. A summary of 15+ years of measuring and listening. We are confident that we established a level of jitter that was not discernible. Not by industry pros, insiders, or members of a secret society. Or folks "in the know" what we were up to. Just the average Joe who bought high-end gear. No knowledge of what we did, why we did it, or why it should matter. All they wanted to know was if I could do that to their gear. Depended on whether they were worried about the resale value of their gear. And my unwillingness to even try. I had other stuff I wanted to do. Like strap on the ol' feed bag. Later. Pat -- ar-t http://www.analogresearch-technology.net ------------------------------------------------------------------------ ar-t's Profile: http://forums.slimdevices.com/member.php?userid=13619 View this thread: http://forums.slimdevices.com/showthread.php?t=56712 _______________________________________________ audiophiles mailing list [email protected] http://lists.slimdevices.com/lists/listinfo/audiophiles
