Jitter - gone for good?

Status
Not open for further replies.
Jump to Latest
#41
jwb said:
You're absolutely right that close-in jitter will modulate your audio outputs. Some manufacturers will specify phase noise at 10Hz, 100Hz, 1K, 10K, 100K offsets. You can get a total jitter number from these by performing a trapezoidal approximate integration.

If you're at a loss for manufacturers, you could start with Valpey Fisher.
Click to expand...

I was looking for clocks a few weeks ago and I didn't find the ideal one.

Are you saying that you can get jitter numbers from the phase noise?
Can you post a formula?

BTW lets not forget that the output should be clipped sinewave. If it's full sinewave (like all OCXOs) i'm not sure which changes need to be done in th design.

Good luck searching. If you find something really good tell me 🙂
Me, I'll add Valpey Fisher to the list.
 
#44
jwb said:
I found myself quite liberated when I finally realized that there was no reason to use 24.576MHz or 11.2896MHz clocks in audio systems where an ASRC is employed.
Click to expand...

Would these the best frequencies to run clocks for 192k and 44.1k respectively (for NOS DACs)?

I mean, why not for instance a multiple of the above (es. 22.5792 Mhz for redbook)?
 
#45
Telstar said:

I mean, why not for instance a multiple of the above (es. 22.5792 Mhz for redbook)?
Click to expand...

As I understand jwb he's saying that those frequencies aren't commonly available in very low jitter instantiations, whereas 25MHz is coz that frequency gets used a lot in non-audio applications where low jitter is important.
 
#46
Telstar said:


Yeah sure, great.
If you use an ASRC already, that is not the greater evil.
Click to expand...

DOH! Well, in my previous comments I guess I didn't do a very good job in describing exactly what the digital workstation is actually doing, my apologies to you Telstar. The dumb thing I did here was say "re-sampling", which in my stated case would NOT even be close to the function that a ASRC chip does! The algorithm that DAWs use to time compress or elongate audio is a purely off-line, non-realtime process that is related to what telecom DSP engineers call a "phase vocoder". This subject, as it regards musical audio signals, is covered in the great book written by F.R. Moore in 1990 called "Elements of Computer Music". The basis of a phase vocoder is discussed in the November 1966 Bell System Technical Journal, in a paper written by J.L. Flanagan and R.M. Golden. The application of such a process to music was actually discussed in a 1978 AES paper written by James A. Moorer, a well-known guru of computer music and digital audio. That paper is available here

Sorry to all if this reply is a bit off-topic, I just wanted to try to correct my prior poor context in saying "re-sample". Thanks for understanding....
 
#47
abraxalito said:


As I understand jwb he's saying that those frequencies aren't commonly available in very low jitter instantiations, whereas 25MHz is coz that frequency gets used a lot in non-audio applications where low jitter is important.
Click to expand...

Thanks for your help.
I think I knew what he meant 🙂
But if you could have the choice to pick a master clock frequency, i wonder which should it be.

i.e. 16.9344 is 384* 44.1k, it is indeed the ideal clock for redbook (and redbook upsampled to 176.4k?)?

I noticed that higher frequency clocks (and i'm talking about high quality parts) have higher phase noise. It is quite significant going for instance from 10Mhz to 100Mhz. I guess that from 10Mhz to 20Mhz the difference is little.

So, my next question would be what about 11.2896Mhz? or lower? Better or worse? (still talking only for 44.1k at the moment).
Is there a problem with resonance?

Most HQ clocks comes from 4 to 30Mhz, so theoretically there is a good range of choice.

This thing of the clock freq is wondering in my mind for awhile and i cannot find a definitive answer. Any point to some study or writeup would be also welcome.
 
#48
Telstar said:


Thanks for your help.
I think I knew what he meant 🙂
But if you could have the choice to pick a master clock frequency, i wonder which should it be.

i.e. 16.9344 is 384* 44.1k, it is indeed the ideal clock for redbook (and redbook upsampled to 176.4k?)?
Click to expand...

I'd choose one between 11 - 22MHz myself. Much higher and I think the oscillator would be harder to design (non-fundamental XTAL).


noticed that higher frequency clocks (and i'm talking about high quality parts) have higher phase noise. It is quite significant going for instance from 10Mhz to 100Mhz. I guess that from 10Mhz to 20Mhz the difference is little.
Click to expand...

In my limited understanding, the phase noise would be bound to increase for constant Q as the frequency goes up. But since we actually use a lower frequency, then after going through the dividers the phase noise comes down again.


So, my next question would be what about 11.2896Mhz? or lower? Better or worse? (still talking only for 44.1k at the moment).
Is there a problem with resonance?
Click to expand...

Don't understand your question about resonance. You'll need a multiple of the sample frequency unless you're employing an SRC. What multiple that is will depend on the particular DAC.
 
#49
Usually a 25MHz XO would use a fundamental crystal vibration, whereas higher frequencies would use an overtone, or a surface wave. It is rare to find fundamental crystals above 30MHz, although they can be made in the laboratory up to 1GHz.
 
Status
Not open for further replies.
Top Bottom