0 Members and 1 Guest are viewing this topic. Read 42041 times.
I'm not sure what the critical jitter range is yet. I have yet to see ANY correlation of jitter spectrum measurements to listening tests
The closest I have come to correlation is the jitter histogram that shows the distribution as well as the spectrum of the same. I have seen some correlation there, but it's inconsistent. When you have distribution, magnitude and spectrum all factoring in, it is hard to derive solid conclusions. A LOT more study needs to be done in this area.
My own DAC, by design, has no reclocker on the S/PDIF input. Direct to the receiver. The best DACs avoid reclocking on the inputs IMO, like the Metrum Acoustics DACs. This way they get the benefit of truly low-jitter sources.
Using the optimum S/PDIF receiver is critical though. They are not all the same. Some of them can result in much lower jitter to the D/A, others are junk.
They are 5X7mm surface-mount. Can you deal with that?
I may be able to send you a SILABS 530CB49M1520DGR - this is 49.1520MHz 3.3VDC It may have solder on the pads. Period jitter RMS spec 2psec. After my logic is added, 7-10 psec is the result.
Ah, yes.....................the Wavecrest approach.................We have one of those. TOTALLY useless when it comes to figuring out how any of this relates to what stuff sounds like. Really neat piece of gear, if you are into histograms. Not so much, once you try to relate that to the real world.(We have some really expensive 10 MHz "reference oscillators", for our phase noise gear. They measure really poor, on the Wavecrest. It will show 2 peaks, like you have a source with correlated crap. The reason why is they are built with a 5 MHz rock, and they run that through a stage that clips, to pick off the 2nd harmonic. [5 MHz is better than 10 MHz, for lots of reasons, so that is how they make their 10 MHz units.] The drawback is the noise floor is higher than some of the competitors. But, those will not do -102 dBc @ 1 Hz. Which is needed, if you are trying to measure a really good clock. The ones with better noise floors have numbers at 1 Hz offset in the low 90s, and we have "$1 specials" that approach that level, so we use the one with a crappy noise floor. Great part, for our use. As long as you don't go by what the Wavecrest says, because it has a "camel" histogram.)(OK..............I know you are probably wondering why that is. The positive and negative halves of the waveform are not symmetric, and that makes the Wavecrest goes nuts. It shows up on the + per and - per measurements.)
BTW, are all of these parts 45/49 MHz, or are they 22/24 MHz ones? I need to stick a D flip-flop in the chain, instead of a inverting buffer, to drive the rig, if they are 45/49. Just need to know what chip needs to be in the drive part of the fixture. The only drawback is the noise floor will show up around 6 dB higher if we need to divide by 2, and tell the machine to reference it to the actual DUT frequency. When you do that, all it does is shift the entire plot up 6 dB. But, the real noise floor is really lower.
NO IT WILL NOT!!!!Did you not see the phase noise plot I posted in the USB thread? The jitter is a function of the BW over which it is measured. Which, in this case, is 1 Hz to 1 kHz. (Maybe you missed the part about how we feel the noise floor is pretty much meaningless.) It has nothing to do with the divider. All it does it lower the entire plot, and therefor the dBc value, by 6 dB. It does not effect the jitter.If all that you do is measure the stinking noise floor, then, yes, the noise floor will come up, by around 2 dB, in some cases. I know, because I do this all day long.I need something that will drive a 50R load (this is RF world), and it is the only difference is a picogate inverter or a picogate flip-flop. I have made enough of these plots to know how much the noise floor goes up, IF you have to insert an additional gate. (Sometimes I do that, on purpose, so I get it reference to 11.2896 MHz, for a specific situation, without s/w manipulation.)But, whatever. I offered. Go ahead and measure the noise floor on your fancy 'scope (which works on pretty much the same method as the Wavecrest) and continue to wonder why you can not find a correlation between "jitter" and subjective listening tests. Good grief, man.This is nothing personal, but this is what I refer to when "digital guys" ramble on about "jitter", compared to what the RF world calls "jitter". Which is really the phase noise, when you are talking about clocks, because that is how oscillators are characterized.Yeah, I know......................"the spec sheet says".....................the spec sheet says diddly-squat because a bogus "jitter" number makes for a nice spec sheet and is real easy to measure. Other than that, it is meaningless, for digital audio.Yeah, I know...............in timing, jitter means one thing. In audio, even though your fancy AP-1 (or -2 or whatever) gives a "jitter" number (which has its value, in its own right), it can not look at what goes on below 1 Hz. (It wasn't designed for that purpose.) That is where it makes a difference. That can only be done by examining the phase noise of the clock. Which I have offered to do.Without sending you a bill for $100 or so. (The gear to do that is not cheap, so it doesn't get done for free very often.)Whatever.