mulletchuck
Well-known member
http://www.soundonsound.com/sos/jun10/articles/masterclocks.htm
please discuss!
please discuss!
external clock
- Thread starter mulletchuck
- Start date
Help Support GroupDIY Audio Forum:
jdbakker
Well-known member
Best magazine article I've read about clocking in a long time, and I don't say that lightly. Solid theory backed up by measurements. I have nothing major to add.
JDB.
JDB.
I would wish this article to put an end to all these BS testimonies about the magic clock that "took the blanket off my monitors".
MikoKensington
Well-known member
I'm of the opinion that quality cabling will get you as much "improvement" for much less money. Even still, you probably wouldn't notice a difference unless you were made aware of any change.
My rig at work runs on a house black burst from the video facility next door. I hosted a demo of the 10M and I can honestly say it did make a difference. Noticeable? Not really. Improvement? I can't say. Just different. In a large session, at 96khz with a number of converters running hardware inserts, it seemed to make more of a "change" than in a smaller track count session running at 48khz.
I'm not sure I get behind this article as a definitive test for clock performance, however. Am I wrong in thinking a clock would be better tested with multiple tracks of high sample rate audio instead shooting through 1 tone at a time? Perhaps, using different clocks on a mix, and using null testing to observe any differences?
I'm thinking of building a mobile rig with multiple converters, and the thought of a clock purchase is confusing me. Should I even bother, or should I spend $1000 on outboard gear? (that's mostly rhetorical)
My rig at work runs on a house black burst from the video facility next door. I hosted a demo of the 10M and I can honestly say it did make a difference. Noticeable? Not really. Improvement? I can't say. Just different. In a large session, at 96khz with a number of converters running hardware inserts, it seemed to make more of a "change" than in a smaller track count session running at 48khz.
I'm not sure I get behind this article as a definitive test for clock performance, however. Am I wrong in thinking a clock would be better tested with multiple tracks of high sample rate audio instead shooting through 1 tone at a time? Perhaps, using different clocks on a mix, and using null testing to observe any differences?
I'm thinking of building a mobile rig with multiple converters, and the thought of a clock purchase is confusing me. Should I even bother, or should I spend $1000 on outboard gear? (that's mostly rhetorical)
Andy Peters
Well-known member
MikoKensington said:
You need clocks on the ADCs on the way in, and the monitoring DAC needs to be clocked as well. And as the article points out, you're best off using the ADC's internal clocks unless you have multiple units which need synchronizing.
Once the tracks are captured, the ADC clock is irrelevant. The mixing in the DAW is not affected by, and has no effect on, converter clocking. Your proposed test won't prove anything.
-a
jdbakker
Well-known member
MikoKensington said:
What exactly would that show that you don't get from a high frequency single tone test?
JD 'not rhetorical' B.
[given that jitter artifacts are fairly predictable, and don't have the dependence on signal envelope that, say, amp nonlinearities or transformer coloring have]
MikoKensington
Well-known member
Andy Peters said:
jdbakker said:
Sorry if I don't understand guys. So when using AD/DA converters to patch in analog inserts in real time (delay compensated) on a mix, and using a DA for monitoring, the clock is irrelevant?
Don't get me wrong, I think clocks are largely snake oil and holy water. I just thought null testing would illustrate any changes, or lack thereof, to the sound of a mix. To prove or disprove the purported dramatic effects of luxury clocks.
And to JDB, you're right. My test wouldn't "show" me anything. But I also don't record music comprised of single high frequency tones. So that illustrated test doesn't "play" me anything.
jdbakker
Well-known member
MikoKensington said:
Not at all. I believe Andy was talking about ITB mixing/effects. When patching analog stuff in clocks do matter.
MikoKensington said:
I get that. My point was that the spectral smearing you get from clock jitter is predictable enough that once you have the single-tone response you can then apply that to any audio signal you like. There's software out there that can do that for you, so you can take your tracks and see/hear how it ends up. I've seen one or two Matlab packages that were used for AES papers, and I believe there's at least one plugin offering this functionality.
(While I do record and mix music, I tend to work more on the converter design side, and for repeatability my tests are comprised of single high frequency tones. By the time I'm satisfied on that front my tin ears can't hear anything I can attribute to jitter in actual program material. For similar reasons I'm not in the market for a master clock, and while I have stumbled upon jitter-applying software I don't use it myself so I can't readily give you any links).
JD 'cross purposes' B.
Nevertheless, these tests are valid as can be. We all know that music is complex, but we also know that it is pertinent to consider it as a sum of simpler tones; although a single-tone test won't permit evaluation of artefacts that happen in the presence of complex tones, it is very often significant enough to trace some problems. When a unit passes the single-tone tests, it is not a proof that it will happily pass complex tones, but when the unit fails on single tone test, it is enough proof that it does not perform well, which some of the examples in the article have clearly shown.MikoKensington said:
You cannot dismiss single-tone tests, although I agree that they are not sufficient to qualify a design.
jdbakker
Well-known member
Not to harp on this too much, but:
I would argue that jitter is one of the few cases where a clean1 single tone test outcome guarantees that there are no jitter-related artifacts in complex signal playback/recording. At that point there may still be other problems (most notably intermodulation) that do only manifest with complex input waveforms. And of course, if the single tone reveals that there is jitter, one might still like the sonic impact.
JDB.
1 in this case clean means: no measurable spectrum smearing above the noise floor, measured on a signal close to Fs/2 but lower than the start of the converter's anti-aliasing filter's rolloff.
abbey road d enfer said:
I would argue that jitter is one of the few cases where a clean1 single tone test outcome guarantees that there are no jitter-related artifacts in complex signal playback/recording. At that point there may still be other problems (most notably intermodulation) that do only manifest with complex input waveforms. And of course, if the single tone reveals that there is jitter, one might still like the sonic impact.
JDB.
1 in this case clean means: no measurable spectrum smearing above the noise floor, measured on a signal close to Fs/2 but lower than the start of the converter's anti-aliasing filter's rolloff.
In all practical terms, you are right. There is however, one very dubious theoretical case where it would not be completely true.jdbakker said:
Let's assume jitter is cyclic at 1 kHz and you do the single-tone test at 1 kHz (both being perfectly synced). The test will show only mild THD increase, mainly 2nd harmonic. Running the test at a different frequency would show a different form of distortion with enharmonic sum and difference products.
Oh, in fact, is that example really hypothetical? What about interface-induced jitter?
jdbakker
Well-known member
abbey road d enfer said:
Riiiight.
While interface-induced jitter is real, you would be very hard pressed to find any DIT/cable/DIR combination where the recovered clock in this scenario would have jitter with energy only at 0 and 2Fc. So yes, my money's on hypothetical.
JD 'straw man' B.
[besides, who in their right mind clocks converters directly from interface-recovered clocks these days?]
rascalseven
Well-known member
I configured my system according to Dan Lavry and Bob Katz's recommendations on this thread:
http://recforums.prosoundweb.com/index.php/mv/msg/14324/0/0/0/
I have the original Apogee AD16 and DA16 with my Digi 192 Digital interface (PT HD3 Accel system). I had been clocking everything off the computer (Digi interface) in daisy chain (WC in, WC out...) function. After reading the above thread I switched to clocking from the AD16 using a BNC 'T' connector to connect to the Digi interface and then terminate at the DA16 word clock in (which is internally terminated 75R - I checked it with my multimeter and scoped it as well to see the quality of the clock waveform).
I must say there was a clear sonic improvement over the previoius connection. It was like getting new converters for the cost of the 'T' connector!
The only problem I have now is a practical one: I just need to make sure I set the AD16's clock for the clock rate of the session document -- a friend who uses my facility tracked and edited a lead vocal overdub at 48k (the AD's clock) when the session was actually a 44k1 session. He edited the vocal and burned a rough mix for listening back only then to realize the mistake when it hearing it back at 44k1.
Other than having to keep an eye on that, the new clocking has been a wonderful improvement!
JC
http://recforums.prosoundweb.com/index.php/mv/msg/14324/0/0/0/
I have the original Apogee AD16 and DA16 with my Digi 192 Digital interface (PT HD3 Accel system). I had been clocking everything off the computer (Digi interface) in daisy chain (WC in, WC out...) function. After reading the above thread I switched to clocking from the AD16 using a BNC 'T' connector to connect to the Digi interface and then terminate at the DA16 word clock in (which is internally terminated 75R - I checked it with my multimeter and scoped it as well to see the quality of the clock waveform).
I must say there was a clear sonic improvement over the previoius connection. It was like getting new converters for the cost of the 'T' connector!
The only problem I have now is a practical one: I just need to make sure I set the AD16's clock for the clock rate of the session document -- a friend who uses my facility tracked and edited a lead vocal overdub at 48k (the AD's clock) when the session was actually a 44k1 session. He edited the vocal and burned a rough mix for listening back only then to realize the mistake when it hearing it back at 44k1.
Other than having to keep an eye on that, the new clocking has been a wonderful improvement!
JC
it's not the first time I hear about this happening, in particular with PT. I must say I find this very strange; how can one not recognize a sound playing a full tone too sharp and 10% above tempo?rascalseven said:
MartyMart
Well-known member
abbey road d enfer said:
Ahem, I had to correct a set of masters from Mel-B's album ( Spice Girls ) that was all pitched up due to someone NOT
spotting this !!
It happens, they were a "pro" mastering house and I'm NOT !! ... go figure ...
MM.
You mean you actually listened to the playback? ;DMartyMart said:
JR
PS: Sure clocks matter, but this gets swept up in audiophoolery's constant second guessing and low opinion of original designs. Unfortunately you need to have a clue to recognize that you don't have a clue, and can't get there from there...
HerculesVR
Member
nice topic``
Every master clock equipments based TCXO(some device using OCXO) then easy to get super high quality clock performance, whatever frequency and jitter(phase noise). But those equipments output is word clock (fs / LRCK), cable delivery will bring additional jitter, so the key is equipment who is word clock receiver how to reduction additional jitter.
Actually master clock equipment is not necessary, if your digital audio equipments have reclock design such as 2nd PLL or others.
Every master clock equipments based TCXO(some device using OCXO) then easy to get super high quality clock performance, whatever frequency and jitter(phase noise). But those equipments output is word clock (fs / LRCK), cable delivery will bring additional jitter, so the key is equipment who is word clock receiver how to reduction additional jitter.
Actually master clock equipment is not necessary, if your digital audio equipments have reclock design such as 2nd PLL or others.
rascalseven
Well-known member
abbey road d enfer said:
Well, at the risk of looking like a fool, I have had two experiences, personally, with sessions that that were brought in for vocal overdubs, and the tracks didn't sound strange at all at the higher speed (I had never heard them at proper speed). Fortunately only one of those sessions resulted in an actual 'mistake' in the recording (I caught the other before hitting 'record').
The more relevant and likely mistake in clocking is when an engineer creates a completely new session at 44.1k, for example, but the converter is at 48k and the system is slaved to it. They do all the drum/rhythm tracking only to get back to their place, open the session for overdubs or mixing (now playing at 44.1), and be freaked out that everything is slowed down.
Just a few hoops to jump through to remedy that situation, but a hassle nonetheless. And yes, it can make you look stupid to a client!! :-\
JC
JBVries
Well-known member
Cheap Big Ben
;D
http://cgi.ebay.com/Apogee-Big-Ben-192k-Master-Clock-/270668593468?pt=LH_DefaultDomain_0&hash=item3f051af93c
;D
http://cgi.ebay.com/Apogee-Big-Ben-192k-Master-Clock-/270668593468?pt=LH_DefaultDomain_0&hash=item3f051af93c
