R/E/P Community

Please login or register.

Login with username, password and session length
Advanced search  

Pages: [1]   Go Down

Author Topic: external clocking at base rate vs. actual rate  (Read 2662 times)

user of gear

  • Newbie
  • *
  • Offline Offline
  • Posts: 18
external clocking at base rate vs. actual rate
« on: August 02, 2007, 01:58:08 pm »

Hello,

There are a number of converters / interfaces that will accept external word clock at either the base rate [44.1khz or 48khz] or at the actual sample rate, if different [i.e. 96khz or 192khz]. This is not limited to Digidesign's stuff, or their "loop sync" feature.

It appears that there are some units that will function at 192khz, but will not accept external clock at that rate at all.

I would like to hear pros and cons of clocking at base rate vs. actual [higher] rate. In general, I like to work at 192khz.

I am aware of the 96khz vs 192khz argument, and I would like to keep that argument out of this thread, except to the extent that it may be relevant to the actual question asked.

Faithful regards,


User of gear
Logged

bruno putzeys

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1078
Re: external clocking at base rate vs. actual rate
« Reply #1 on: August 08, 2007, 05:23:52 am »

Most gear, when presented with 2x or 4x word sync signals will first divide them down to 1x and run the PLL off that. It's what I do anyway. In any case the word sync input is used to synchronise a clock oscillator with a frequency much higher than the sampling rate (anywhere between 64fs and 1024fs) so whether you start with 48kHz or 192kHz doesn't really matter except for unusually badly designed PLLs.
Logged
Warp Drive. Tractor Beam. Room Correction. Whatever.

Affiliations: Hypex, Grimm Audio.

user of gear

  • Newbie
  • *
  • Offline Offline
  • Posts: 18
Re: external clocking at base rate vs. actual rate
« Reply #2 on: August 09, 2007, 03:31:10 am »

Hello,

   Thank you for repying, Bruno.  I am trying to sort out some issues before "investing".  I have a PT HD3 rig and I thought I would find a "high end" A to D, mostly to use for the final mix back into PT.
   I have a whole bunch of stuff to finish that was tracked at 192khz, so I need a unit that will work at that sample rate in order to record the 2 mix back into the original session.  Not to easy to find, especially something that supports dual wire mode.
   I have actually heard from a number of design engineers involved with high end clocking devices, saying that one is better off sending the highest clock rate to the converters that they will accept.
   Digital Audio Denmark's AX24 will not accept 192khz clock.  They require 48khz or 96khz, even when converting at 192khz.  They say that the jitter attenuation at 192khz is not good enough, and that higher clock rates are more susceptible to jitter.  That is something I had never heard before, and it does make me wonder about how well the PLL is designed.  They say 192khz no good because their "internal master clock is 266 times faster than the base rate word clock".  Does that make sense?
   Any additional information / explanation would be much appreciated.  
   Thanks again for replying.


   Faithful regards,


   User of gear
Logged

bruno putzeys

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1078
Re: external clocking at base rate vs. actual rate
« Reply #3 on: August 09, 2007, 09:13:21 am »

From "clock recovery 101": The optimal loop bandwidth settings is the frequency where the noise densities of the incoming clock and the VCO intercept.
If the VCO of the PLL is expected to be noisier than the incoming clock, use a wide loop bandwidth. In these circumstances a higher reference clock frequency is expedient. If the VCO of the PLL is better than the incoming clock (say a very nice VCXO with low-noise control circuitry), the best is to be had by making loop bandwidth narrow (<100Hz) in which case there is nothing to gain or to lose by choosing 48kHz or multiples.

A wideband PLL with a noisy VCO is a correct design point if the intention is only to multiply a presumed low-noise clock. Budget converters and on-chip PLL's (notable exception: the excellent hybrid PLL on TC's DICE chip set) will typically use the wideband version. Converters using wideband PLL's may actually turn out to perform better when synchronised to a very high quality external clock than when running off their internal reference.
A narrowband PLL with a low-noise VCO is the correct choice if you want to clean up the incoming clock as well. Converters using such a PLL usually don't care about the quality of the incoming clock.

It would appear from what you quote that DAD use a PLL bandwidth that scales with the incoming clock frequency. That's another alternative if you don't want to divide the clock back to base rate, and that's why I do prefer to divide first. Keeps the PLL design constant regardless of 1x 2x or 4x clocks.

As you can see, much depends on how the PLL is actually designed. Whether using a higher input rate improves or worsens converter performance is something you need to figure out on a case basis. What a manufacturer says about this can only be taken as applying to their product.
Logged
Warp Drive. Tractor Beam. Room Correction. Whatever.

Affiliations: Hypex, Grimm Audio.

user of gear

  • Newbie
  • *
  • Offline Offline
  • Posts: 18
Re: external clocking at base rate vs. actual rate
« Reply #4 on: August 11, 2007, 03:49:26 am »

Hello,

   Thanks again, Bruno.  I think I am understanding at least most of what you are saying.

   I guess my question [and concern] is whether or not the Digital Audio Denmark stuff will "play well with others", or whether it is one of those devices that sounds much better when referenced to its own internal clock.

   I do not see being able to clock everything to the Digital Audio Denmark AX24 device while mixing, simply so that it can run off its own clock.  My reasoning is that it does not seem proper to daisy-chain a bunch of interfaces and other gear while mixing.

   I do intend to use a high quality master clock such a a Nanosyncs HD, Iclock, or something along those lines.

   Do you see any propblems with running a system where I am sending different Fs multiples to different interfaces?  

   Do devices such as the Digital Audio Denmark AX24, that apparently scale the PLL bandwidth to the incoming clock frequency, perform less well on external sync?

   I am also thinking about the dCS and Weiss converters as an option, and possibly the Universal Audio 2192.  I know I will have to give them a listen, but I would also like to have some technical understanding when considering them [maybe I could thin out the herd, so to speak].

   I would appreciate your thoughts and recommendations when you get a moment.  Thanks.


   Faithful regards,


   User of gear


Logged
Pages: [1]   Go Up