Ext Word Clocks - No point unless syncing other digital gear

[labman]

Hey Dave, would this also be your suggestion if you have additional Audiowire devices connected such as a 2408mk3?

The 2408 doesn't have an internal clock. The 2408 slaves to either the PCI card, or external sync.

If you have an HD192 and 2408, and no other digital gear, make the HD192 the clock master.

Dave

What are you suggestions if you have bunches of other digital gear Dave? Thanks

[James Steele]

Hey Dave, would this also be your suggestion if you have additional Audiowire devices connected such as a 2408mk3?

The 2408 doesn't have an internal clock. The 2408 slaves to either the PCI card, or external sync.

If you have an HD192 and 2408, and no other digital gear, make the HD192 the clock master.

Dave

What are you suggestions if you have bunches of other digital gear Dave? Thanks

I think Dave would agree with what the author said basically that the time you explicitly NEED an external clock is when you DO have "bunches of other digital gear." That's really the only way to do it in that case. But otherwise, use the internal clock in your interface.

I really should change the subject header to say "other digital gear"... I said "video" but it's not really quite accurate.

[labman]

I think Dave would agree with what the author said basically that the time you explicitly NEED an external clock is when you DO have "bunches of other digital gear." That's really the only way to do it in that case. But otherwise, use the internal clock in your interface.

I really should change the subject header to say "other digital gear"... I said "video" but it's not really quite accurate.

Thanks for the add'l info James. Cant wait to read the article. Just arrived at office today.

What sample rate do you usually use with your HD192's James?

[magicd]

What are you suggestions if you have bunches of other digital gear Dave? Thanks

In my experience, the potential problem with multiple pieces of interconnected digital audio gear is this:

If you daisy chain clock from one device to the next, it's possible that over the length of that chain, there will be a timing delay or error introduced.
Lets say you have five pieces of gear. Device 1 generates clock. Device 2 passes clock to device 3 and so on. Device 5 then feeds the daisy chain clocked digital signal back to device 1. If there was any accumulated error in the clock chain, there will now be a mismatch at the digital input of device 1.

If in this case, a central clock generator is used, all five devices reference the same clock. Digital signal can be sent between any of the devices and they will all be in sync.

This is not an example of "jitter" being solved. This is an example of mismatched clock being solved. I've run across this situation more than a few times. In this setup, adding in the dedicated clock master resulted in the ubiquitous "it's like someone took a blanket off the speakers".

In my own rig, I'm running an 896mk3 with two 8Pre connected via optical. The 896mk3 is the master and the 8Pres are slaving. Since the 896mk3 has two optical outputs, there is no daisy chain and the 8Pres each get their clock directly from the 896mk3. There is no need in my setup for an external clock generator.

Dave

[1nput0utput]

I've been told that MOTU interfaces regenerate word clock in phase-alignment with the selected clock source. Based on that, you could daisy-chain word clock through any number of MOTU devices and the signal at the end of the chain would be identical to the signal at the beginning.

[James Steele]

What sample rate do you usually use with your HD192's James?

To be honest, I'm just mucking about with rock and roll songs. I'm working with another guy who is an excellent engineer, drummer, and on the MOTU-MAC list a lot-- Jim Watson. He and I have been working on a project and we've pretty much only used 24bit/44.1k. Jim has an HD192 as well. I know Jim has an Apogee Big Ben, but that's largely because he's doing film/video work and he has some digital outboard that he needs to synchronize to a common clock source.

[James Steele]

I've been told that MOTU interfaces regenerate word clock in phase-alignment with the selected clock source. Based on that, you could daisy-chain word clock through any number of MOTU devices and the signal at the end of the chain would be identical to the signal at the beginning.

My understanding is that (at least in the case of MOTU PCI interfaces) no matter how many interfaces you have they all sync together as if they were one unit. The PCI-424 card keeps all the units together. This is why if you're using external word clock, you can only designate one interfaces external word clock as the master. There's no benefit to distributing word clock from an external clock to multiple MOTU PCI interfaces. It will have no effect, except on the interface selected as the master ext. word clock source in PCI Audio Setup.

[magicd]

I've been told that MOTU interfaces regenerate word clock in phase-alignment with the selected clock source. Based on that, you could daisy-chain word clock through any number of MOTU devices and the signal at the end of the chain would be identical to the signal at the beginning.

My understanding is that (at least in the case of MOTU PCI interfaces) no matter how many interfaces you have they all sync together as if they were one unit. The PCI-424 card keeps all the units together. This is why if you're using external word clock, you can only designate one interfaces external word clock as the master. There's no benefit to distributing word clock from an external clock to multiple MOTU PCI interfaces. It will have no effect, except on the interface selected as the master ext. word clock source in PCI Audio Setup.

You are correct. The MOTU PCI system has a single clock source at any given time.

However, what 1nput0utput was referring to is clock output from MOTU interfaces. On a MOTU PCI system, there are potentially many word clock outputs. AES/EBU, optical, BNC, and S/PDIF. All these connections will output the same clock at the same time. Hence, these connections could be used for the word clock distribution to multiple devices, as opposed to daisy-chaining the clock.

Dave

[James Steele]

However, what 1nput0utput was referring to is clock output from MOTU interfaces. On a MOTU PCI system, there are potentially many word clock outputs. AES/EBU, optical, BNC, and S/PDIF. All these connections will output the same clock at the same time. Hence, these connections could be used for the word clock distribution to multiple devices, as opposed to daisy-chaining the clock.

That is very cool to know! So if I have a 24 I/O and an HD192, driving off HD192's internal clock, I could sync two digital devices, one receiving WC from the WC Out of the HD192 and another getting WC from the WC Out of the 24 I/O. Very slick. Makes me almost thing you MOTU guys know what the heck you're doing over there!

[stephen1212b]

I use an old Aardsync. Both my Traveler and 2408 MK3 sound better running off of it than off of their internal clocks. Tighter more extended bass, silkier top end. It is true that the clock nearest the ADC and DAC will have an advantage, so use the clock in your recording or monitoring interface and not the one in the interface card. However a really good external clock can still outperform a lesser quality built in clock. One overlooked reason besides jitter is absolute accuracy. All of the math assumes the proper sample rate. I tested the Traveler today set to 88100 and it measured at 88103.75. over several minutes it drifted up to 88104.67, possibly warming up. Switching to the TCXO (temperature controlled crystal oscillator) Aardsync clock and the sample rate moved down to 88100.64, and remained there. This took several seconds as the sample rate is controlled by a phase locked loop circuit. The external clock controls the rate on the internal clock, it doesn't replace it. I realize the absolute error is very small in either case, but the Aardsync still had up to 7 times less error, and the error without it changed by more than its residual error in just a couple of minutes. If you record on location in the heat or cold you can expect much larger errors. AC turning on and off or the amp heating up the rack as you listen back between takes. It all matters.

[1nput0utput]

However, what 1nput0utput was referring to is clock output from MOTU interfaces. On a MOTU PCI system, there are potentially many word clock outputs. AES/EBU, optical, BNC, and S/PDIF. All these connections will output the same clock at the same time. Hence, these connections could be used for the word clock distribution to multiple devices, as opposed to daisy-chaining the clock.

That's not quite what I was referring to. The point I was making is that you can daisy-chain word clock output to word clock input from device to device when the output device regenerates its clock signal. MOTU interfaces regenerate the clock signal (regardless of the clock source). That means you could connect a number of MOTU interfaces from word clock output to word clock input and the word clock signal at the output of the last device would be phase-accurate to the clock signal of the first device.

[burp182]

Dan Lavry is a brilliant guy and makes a compelling argument about using the internal clock vs. an external one. (For those who didn't read Dan's paper, the crux of his logic is that an internal clock runs at @2 megahertz. Why would you substitute an external one at 44.1 kHz?) A very compelling, logical argument. It just isn't borne out in some real world situations.
I'll tie this to one clock in particular. YMMV with different brands. The Antelope clocks can make a substantial difference under the right circumstances. I use Apogee, Benchmark and Crane Song I/O with Lynx AES16 cards in my system and an Antelope DA to supply all the computers and digital outboard gear that have WC inputs available. Nothing is daisy chained. Everything in this pile has a really good clock in it. So referencing to an external clock should at most make a tiny difference in sound quality. It does not. The difference is noticeable to a novice. Not subtle. And when I hook the Antelope Atomic Clock to the Isochrone WC generator, I'm going from a stunning clock to an even better reference rate. Should be a TINY difference. Nope. The image widens and the center becomes distinct, floating between the speakers in an almost "reach out and touch it" kind of sense. The sound becomes much closer to analog (in the best sense of that term).
Having said this, circumstances vary. I loaned a clock to someone with an RME interface and Mackie HR824 speakers. The difference they heard was minimal. The RME Fireface 800 has a very nice clock, but I think the listening situation had more to do with it.
External clocks are a crapshoot. Try it. If you hear a difference, then you win. If not, nothing lost. But I disagree with the orthodoxy as presented by Dan Lavry because of the non-modest difference I've heard with my own ears.

[James Steele]

Have you read the S.O.S. article? I was really referring to that instead of Dan's post. I will see if I can find time to post quotes from the article, because there were some pretty strong arguments, essentially saying that it's impossible for a decently-designed modern internal clock to actually perform better when it is having to sync to an external clock source. I'm just the messenger. I'm sure they'll get some feedback on the article.

[1nput0utput]

In any case, the measurements don't lie, right? There are quantifiable ways of measuring signal quality, and it's not possible to measure subjective qualities such as "distinct center," "floating between the speakers," or "closer to analog" (unless you're speaking of distortion or noise characteristics).

If your very expensive clock generator makes you feel better about your recordings, then you win. If it makes your clients feel better, then you win more because you can charge them more money.

[burp182]

In this case, the measurements don't always reflect real world results. Systems that measure equally do not always sound alike.
I don't have access to the S.O.S. article (sorry) but Dan's thoughts are very consistent on the subject. He clearly articulated them to me as we discussed the AD-122 ten years ago and as recently as two years ago (the last time we talked). It is well thought through, completely rational and borne out by the numbers. It just doesn't always prove to be better sounding in the real world.
I'll give you an example. In a mastering studio I was in, they owned an Antelope WC generator and were about to demo the Atomic Clock add-on. This enhancement should represent just about the smallest increment of potential improvement imaginable (an extremely accurate clock changing it's timing source to an even more accurate one). I was standing with the mastering engineer about 6 feet behind and 4 feet to the right of his listening position, both of us standing sideways relative to the speakers, he facing right, me left. When the Atomic Clock was plugged in to the Isochrone, both of our heads immediately swiveled to the speakers in a "What the heck?" moment. NOT subtle. Sitting down he remarked on the imaging improvements I detailed earlier. I sat down and agreed.
I don't believe in audio voodoo. My eyes roll back in my head as audiophiles argue the merits of $400.00 ITC power cables. This ain't that. I don't need to "feel better about my recordings". Been a professional engineer for 39 years and have confidence in my abilities. In a high resolution system, external clocking can make a distinct improvement, regardless of Dan's position on the subject. YMMV.