Printable View
I would agree it's worth a try - using a dedicated complete supply for a CDP clock makes a difference, with or without a better crystal (no noise in return path from other shared devices as stock), so it will improve quality of crystal reference. Points is whether that translates to an audible improvement in a DD turntable. Only one way to find out :)
Richard
The crystal will always be the limiting factor.Quote:
Originally Posted by UV101
Any difference a supply would make would almost likely be in the noise floor. We have not found (with our uber-exotic test equipment) all that much difference in supplies. But, you must note that we do not measure anything above 1 kHz, when it comes to jitter frequency. Which is the area one would be looking, if they were interested in the noise floor.Quote:
With the various clocks I've had and built for CDP's I've definitely found that the power supply makes a difference. I realise that better oscillators with real interesting parameters quoted like phase noise will ultimately out perform something like the eBay oscillators with nonsense performance figures quoted. I would say that the quality and suitability of the supply rail will allow an oscillator to perform optimally.
Having said all of that..............
There are lots of folks claiming to hear differences in supplies, and I would not dismiss their claims. Only to point out the differences between a really good crystal, and the ones 99% of us have, is much more dramatic.
Now, as for the actual schematic...........
Yes, I contend it won't make much difference, unless you are really overdriving the crystal. With the common ones, the crystal will swamp any effects of the design. When it comes to really, really good ones, then everything matters, if you want it to perform at its best. In this case, what is perhaps the most important parameter is drive level. Too cold, and it will be on the noisy side. Too hot, and you run the risk of not only fracturing it (which I have never seen), but more importantly, it will have stability problems. So, there is a narrow range of drive levels, that will work. Unfortunately, it takes equipment most companies do not have. (Which is good, because we pad our income measuring stuff for other companies.)
One very peculiar effect we have noticed, and lots of our "clients" think we are nuts (we are, but for other reasons) is that if the drive level is not right (which usually means too low), you will see a lot of mains-related spurs, in phase noise plot. In all likelihood, they are not spurs, but mains noise that is picked up by the circuitry. So, if you see lots of 60 Hz (in our case), you know the drive level is not optimum. A slight tweak can reduce it by over 10 dB. In specific, it would mean changing the series R, in the typical IC Pierce circuit, by only a few ohms.
Which reminds me....................the one you see, in 99% of the publications.................
Wrong, in one way or another. You need to create a 90 degree phase shift, from the output of the gate, to the input of the crystal. The only way to do that is with a RC network. Yes, you could just use the R of the gate, but as low as it is, you are going to need a really big cap. And the gate won't like looking into that low of a Z. So, that circuit must use a series R. It also helps to keep drive level at a safe difference.
So..........as your homework......................stick 100R in series with the output, back to the crystal, and first shunt cap. Figure out how much phase shift there is, at the operating frequency, with only the usual 33 pF cap.
Then try it with something in the range of 1-10 nF. Tell me which one makes more sense.
Has anyone else considered this?