R/E/P Community

Please login or register.

Login with username, password and session length
Advanced search  

Pages: [1]   Go Down

Author Topic: If a picosecond of jitter is bad...  (Read 3633 times)

danickstr

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 3641
If a picosecond of jitter is bad...
« on: February 28, 2007, 10:24:28 pm »

How on earth can anything possibly get recorded reliably on anything short of a stellar system?

I have seen the experts here say that most clocking devices with light pipe are not reliable and so 99% of the middle to high-end studios that are running a digital master clock are suffering from jitter issues if they go to toslink at any time.

What is the solution for, let's say, a Protools rig with an "other" converter that is using toslink inputs on the 192?

How bad will jitter be here and is there a solution that will be better short of using aes/ebu 25 pin?
Logged
Nick Dellos - MCPE  

Food for thought for the future:              http://http://www.kurzweilai.net/" target="_blank">http://www.kurzweilai.net/www.physorg.com

bruno putzeys

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1078
Re: If a picosecond of jitter is bad...
« Reply #1 on: March 01, 2007, 03:49:20 am »

How bad a certain amount of jitter is could be restated as how much better does it get when you reduce it and where the point of diminishing returns may be. How bad it objectively is, is discussed in papers referenced in answers to similar questions posted by yourself.

As for the solution: USE A HOUSE SYNC. A clock is an analogue signal. Stop trying to put an analogue signal (clock) through the same pipe as a digital signal (bits). Lock all devices to word clock. Then you no longer have to worry about which connection you use to link the data, toslink or electrical will no longer make a difference. Plus, you don't get to lose lock after passing AES/EBU through several devices (which is the real reason why external clocking was invented, not sound).
Logged
Warp Drive. Tractor Beam. Room Correction. Whatever.

Affiliations: Hypex, Grimm Audio.

danickstr

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 3641
Re: If a picosecond of jitter is bad...
« Reply #2 on: March 01, 2007, 11:15:30 am »

So just slave everything to house WC, and the jitter issues with lightpipe and AES go away, for the most part?
Logged
Nick Dellos - MCPE  

Food for thought for the future:              http://http://www.kurzweilai.net/" target="_blank">http://www.kurzweilai.net/www.physorg.com

bruno putzeys

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1078
Re: If a picosecond of jitter is bad...
« Reply #3 on: March 01, 2007, 03:17:43 pm »

They should go away entirely.
Logged
Warp Drive. Tractor Beam. Room Correction. Whatever.

Affiliations: Hypex, Grimm Audio.

Arf! Mastering

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 889
Re: If a picosecond of jitter is bad...
« Reply #4 on: March 03, 2007, 03:14:58 pm »

House sync does not work for me because my mastering workflow nearly always involves two simultaneous sample rates.  The solution, in this case, DACs that self-clock using a VXO clocked buffer.  Then, as long as there are no lock issues, the monitored audio will be as jitter free as the clock in the DAC is able to manage.
Logged
“A working class hero is something to be,
Keep you doped with religion and sex and T.V.”
John Lennon

"Large signals can actually be counterproductive.  If I scream at you over the phone, you don’t hear me better. If I shine a bright light in your eyes, you don’t see better.”
Dr. C.T. Rubin, biomechanical engineer

bruno putzeys

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1078
Re: If a picosecond of jitter is bad...
« Reply #5 on: March 04, 2007, 01:11:06 pm »

Would you be helped with a word clock generator that puts out two word clocks, 44.1k and 48k at an exact 147:160 ratio? I'm not exactly planning such a device but a synchronous SRC is in the works and I was wondering if a WCK out would help. Weiss has a great synchronous SRC too but it has neither WCK in nor out even though it generates the output rate from the input rate in a fixed ratio.
Logged
Warp Drive. Tractor Beam. Room Correction. Whatever.

Affiliations: Hypex, Grimm Audio.

Arf! Mastering

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 889
Re: If a picosecond of jitter is bad...
« Reply #6 on: March 05, 2007, 10:13:02 am »

Typically, I run at 48/44 88/44 or 96/44.  Because there is an ADC in the outboard loop I would likely use its own clock for the capture.  I can see that a house WCK offering simultaneous 44.1, 48, 88.2, 96 might be useful in a more complex set-up than mine, but for simple two-channel mastering the daisy chain approach is working well.

It looks like this:

DAW source @ 1Fs ->DAC -> Analog gear -> ADC @ 2Fs ->DSP ->SRC to 1Fs -> DAW (master print @ 1Fs)

The DAW provides the playback clock.  The VXO buffered DAC de-jitters. The ADC provides the capture clock.

1Fs can be either 44.1k or 48k.  2Fs is usually 88.2k
Logged
“A working class hero is something to be,
Keep you doped with religion and sex and T.V.”
John Lennon

"Large signals can actually be counterproductive.  If I scream at you over the phone, you don’t hear me better. If I shine a bright light in your eyes, you don’t see better.”
Dr. C.T. Rubin, biomechanical engineer

Bernardo

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 285
Re: If a picosecond of jitter is bad...
« Reply #7 on: March 07, 2007, 06:32:43 pm »

How different is daisy-chaining word clock from doing the same to AES-EBU clock?
Logged

bruno putzeys

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1078
Re: If a picosecond of jitter is bad...
« Reply #8 on: March 08, 2007, 03:12:14 am »

Daisy chaining (ie going from the output of one box into the input of the next) is not recommended unless you're sure the output clock is directly driven by the input (not regenerated). Everytime a clock is regenerated, jitter is amplified (PLL's have net gain around the corner frequencies) and accumulated. The problem works out worst for AES11 connections because more jitter is generated.
The best practice is to star distribute the house sync, although in theory it's also perfectly OK to run a single cable with T junctions past all devices with only the last device terminated. That requires that termination can be switched off, and that all devices have compatible signal level spec. And there's the rub. Word clock over coax is not standardised, everyone uses different specs for signal level and termination. Dedicated word sync generators (should) have independently switchable termination and/or drive levels on each output, making star distribution the optimum choice.
Logged
Warp Drive. Tractor Beam. Room Correction. Whatever.

Affiliations: Hypex, Grimm Audio.
Pages: [1]   Go Up