R/E/P Community

Please login or register.

Login with username, password and session length
Advanced search  

Pages: 1 ... 4 5 [6] 7 8 ... 15   Go Down

Author Topic: Proper word clock implementation  (Read 173859 times)

danlavry

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 997
Re: Proper word clock implementation
« Reply #75 on: November 05, 2004, 08:18:56 PM »

Dan,

You seem to have a problem with sales guys, even the ones with knowledge and integrity. Max never states anything without backing it up with engineering.


Those are confrontational comments, not technical ones.

So here is an engineer, who speaks your language and tells you, you are wrong. Sure, jitter is bad, we never said anything different. Just like you, I was convinced that internal clocking was the best thing to do. But here is what I found:

OK, lets hear why internal clock is not the best thing to use:

I have designed several PLL’s most of them hybrid and as you very well know there is not a perfect world. The VCO solution is more jittery but has a wider capture range, the VCXO solution is almost perfect but has a very narrow capture range.

Good you admit that the VCXO is almost perfect. The fixed Crystal is better than VCXO. The difference is that fixed crystal is optimized fro fixed frequency, the VCXO must operate over a tiny range such as +/-100 parts per million. So again, the fixed crystal is better, and that is what one uses (or should) for internal clock.  

You see, I worked in recording studios for two decades and found myself 9 out of 10 switching to wide capture range mode, because the situation forced me. So that is why you never find a VCXO on an Apogee device, it is simply a pretty useless PLL, not to mention the lack of being able to varispeed.

My gear uses VCXO for external lock, and we do not have customers loosing lock, not 9 out of 10, not 1 out of 1000.

BUT we need PLL’s and PLL’s have diametrically opposing requirements.

You are getting off the track. The subjects at hand are:
1. INTERNAL CLOCK vs PLL
2. THE BEST TECHNOLOGY FOR EXTERNAL CLOCK SOURCE

So you are redefining the “starting point” as a need for PLL. I say we need a fixed crystal with no PLL for best jitter. You are getting off the subject.

They need to be fast to follow sudden changes in the clock source, they need to be slow to have very low jitter…. Until I found a DDS which I really liked, one that I could use in a non-standard solution and I was able to improve on, in its specs. Let me give you some hints, its filtering, layout…all that analog crap again…

Nuts. Let me give you some hint: Before you talk about filtering, remember that a fixed crystal is a far superior filter to anything you are talking about. It is about DEVICE PHYSICS, where the alignment of the quartz molecules is such that the device yields the narrowest bandwidth possible, thus the bandwidth of operation is tiny, the Q factor is HUGE. Putting a device like that in a loop is THE BEST way to overcome power supply variations, and what you call “analog crap’. By the way, I love analog and it is not crap to me.  

A VCXO for external locking is almost perfect (YOU said it yourself). A fixed crystal (for internal applications) is so much better.

“…of course our implementation. And that is as far as I am willing to go in telling how it works. We even decided not to get a patent on it, because that would be a giveaway as well. So yes we are very proud of our work and can imagine you don’t like to see someone succeed in something you thought was impossible.”

You totally side stepped the issues of:  
1. INTERNAL CLOCK vs PLL
2. THE BEST TECHNOLOGY FOR EXTERNAL CLOCK SOURCE

During the development of or C777 PLL we also did some listening tests and found to our surprise that a lot of other converters sounded better (meaning more accurate, closer to the source) when clocking to the C777, against all theory! This was one of those moments where the engineer in me was dumbfounded, it did not make sense at all.

Ah, so we are back to marketing. I agree that the engineer in you is dumbfounded and what you say still does not make sense, nor is it to the point.  

“Yet the tests were conclusive and repeatable. So we researched that and developed a clock that not only performs great as a PLL but as a master clock as well.”

More marketing.

“Again, I would hurt my own research if I tell you why that is, we did find something and as usual it is not that complex. But why should I give it away? It is my product and it allows me to make a decent living, sorry. To give you an idea, we have the same C777 in the AD16X. There is no difference in performance whether you have the unit on internal or external clock, if you have been to the Rhode & Schwartz booth on the AES show in San Francisco you could have seen the performance yourself, since we loaned the unit to demo their new test-equipment: -111 dB THD+N at -1 dBfs. Now that is what I call “unsurpassed excellence” (to quote your website), for a 16 channel AD converter retailing for $3500.- It is up to you to beat that…At the same booth you were also able to see all kinds of other measurements, including an FFT and those who have been there, can acknowledge the “crystal” performance we get out of our C777.”

Maybe I should start talking about the music I play with my Klezmer band. A bunch of words petting yourself on the back, but NOTHING about the subject at hand!!! It certainly seems like you are running away from the technical. I know why you are.

Bottom line is Dan, technology and measurements are only one part of developing equipment. Of course we all want absolute values, facts. But experience has taught me that objectivity always comes after the subjective: why does something sound good?
Instead of: this will make it sound good. The list of examples when this happened is infinite. So I suggest, and I am told you have good ears, to listen to a Big Ben clocking a converter and then come again…you will be surprised.


Ah, the famous lecture about sounding good. I thought we are taking about JITTER. You literature claims that you C777 is “THE CURE FOR THE JITTERS”!

Now finally, about Soft Saturate and UV22, when you see our current line Soft Saturate does not exist anymore, it has been replaced by the much more subtle and better sounding Soft Limit circuitry, but that is besides the point I want to make. These are still part of our current converters, of course, why wouldn’t it be? We also continued to design our products around a converter chip! Are you going to claim that as one of your achievements as well?
Regards,
Lucas van der Mee
Sr. Design Engineer
Apogee Electronics


I did not invent the resistor, nor did I invent the IC. I did co-invent the UV22. Trying to attack me and lower me is a reflection on you.

Lucas,
I did not go after Apogee. I was talking about clocking. Max came in and now you. This is a technical forum.

Lets get back to technical. If you can not do it, I can:

THE FIRST ISSUE IS ABOUT USING INTERNAL CLOCKS VS EXTERNAL.

YOU ADMITTED THAT JITTER IS BAD. Then, you ramble on saying nothing short of a few words about filters or implementation of a non fixed source, then name throwing (Rohde and Schwartz), then how things sound.

Meanwhile, you guys are telling people that using your clock will be a “CURE FOR ALL JITTERS”. Technically speaking, even if your external clock had no jitter, THERE IS NO WAY THAT IT COULD REMOVE, CANCEL, REDUCE the noise picked up at the other end of the interconnecting cable, the receiver, the internal PLL. Your clock box DOES NOT KNOW WHAT IS AHEAD, and therefore CAN NOT COMPENSATE FOR IT!

THE SECOND ISSUE IS COMPARING YOUR CLOCK JITTER TO A GOOD CRYSTAL CIRCUIT.

Your DDS is, at best, based on a crystal that operates on an overtone, thus the “engine” of your design is already inferior. Go to analog Devices site and read about the impact of the reference clock on the DDS. It can not be corrected for, by DSP or any filter. That clock is the ONLY referance. If it moves, there is nothing else to know that there is a need for correction! The ONLY reference is your reference clock, and when it jitters, everything follows!!! Also, think about all that “analog crap” as you call it, where every device in series at the DDS demands the cleanest supply. Think about the fact that the more devices in series, the more jitter accumulates. That a Crystal oscillator can operate with one or two transistors in series and everything is designed to be FIXED, while your DDS has all the extra circuitry to deal with (for additional features) thus more buildup of intrinsic device jitter. Realize that the best way to filter the jitter noise is with the best filter – a crystal. Realize that a supper high Q narrow bandwidth device – a crystal, when operated with the proper feedback gain, can be optimized to yield a pure sine wave at one frequency, in contrast to the square wave of the digital devices.

Did you know any of it, or are you out of your league? So far, you did not come up with a single technical point to refute what I said. I am surprise you showed up here saying what you said, letting yourself be so exposed, almost all sales and no engineering. I told Max to send the engineer.

It would not be uncommon for a manufacturer to sell something with poor specs saying “it sounds good”. But it is an altogether different to call it “A cure for the jitters”. I would not argue with someone that decided they like the sound of more jitter, more distortions or anything that is a matter of taste. I do not hear you admitting that the internal clock is less jitter. That using any external clock to drive an internal PLL yields more jitter than internal fixed crystal. I do not hear you admit that a fixed crystal is less jitter than your DDS. You stated that a VCXO (pull able crystal) is almost perfect. Are you saying your DDS is better than almost perfect?

THE BOTTOM LINE:

QUESTION: WHICH SOLUTION PROVIDES LESS JITTER AT THE AD LOCATION?

1.   A REASONABLE FIXED CRYSTAL
2.   BIG BEN DRIVING A CLOCK INTO A CHASSIS EQUIPED WITH A PLL VIA A 10 FOOT CABLE?

QUESTION: CAN YOU EVER CANCEL OR REDUCE A RANDOM NOISE AT ONE END OF A CABLE WITHOUT KNOWING WHAT IT IS AT THE OTHER END?
(can you cancel, remove or reduce the receiver and PLL noise, some of it unknown, some totally random, by driving it with a clean low jitter clock? Driving it with any clock?)

That is in fact what you are selling!

I remind you that before you or your sales guy entered into this forum, no one was saying names, or directly criticizing any company. If you read the posts, for every question we have tried very hard to remain technical and elicit technical responses. We are not about subjective listening here but you can certainly find chat rooms that are. Perhaps you believe what you say as does your sales director. The facts remain that putting aside your personalized comments you have not edified or elevated the technical level.  I speak not only for myself, but for others who are reading these posts, we want very much to have intelligent conceptual inputs in the spirit of positive interaction.  I gain no joy at having to put you down.

You are missing much by not understanding analog design as are so many other young designers.  

Dan Lavry
Logged

bobkatz

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2926
Re: Proper word clock implementation
« Reply #76 on: November 05, 2004, 08:49:44 PM »

Lucas. I would like to perform some objective and subjective (blind) listening tests on the Big Ben. What I like about your post is the "against all theory" part. Because no one can refute that statement except with measurements and tests. I'd like to see the theory broken!

However (and this is important):

Will the measurements show the improvement as well as the listening tests? You describe some Rohde and Schwartz measurements. Does this include FFTs of a high frequency -1 dBFS test tone to look at the artifacts?

If I discover with simple measurements that the jitter artifacts of selected converters measure worse with the Big Ben, will you understand that I need not proceed with subsequent listening tests? Because that would refute your statement of "against all theory"?  

Are you game? Send me a Big Ben for testing. While I may seem stubborn when it comes to the technical, I am stubbornly open-minded when it comes to listening. After all, I worked for an audiophile label for more than 10 years! And you know audiophiles. I WILL separate my opinion from the facts, and if we do listening tests they will include other listeners, and will be BLIND.

May the best clock win!

Please Send the Big Ben to:

Digital Domain
931 NSR 434 Suite 1201-168
Altamonte Springs, FL 32714

It will either be purchased, or returned, in like-new condition. Promise. Furthermore, I will not write or publish anything until you have seen what I am to write, and have been given an opportunity to respond.

BK
Logged
There are two kinds of fools,
One says-this is old and therefore good.
The other says-this is new and therefore better."

No trees were killed in the sending of this message. However a large number of
electrons were terribly inconvenienced.

Curve Dominant

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 774
Re: Proper word clock implementation
« Reply #77 on: November 07, 2004, 05:11:39 AM »

Quote:

posted by Lucas van der Mee:
technology and measurements are only one part of developing equipment. Of course we all want absolute values, facts. But experience has taught me that objectivity always comes after the subjective: why does something sound good? Instead of: this will make it sound good.


Lucas,

That statement makes it appear as if you should get out of the engineering business, and be an artist instead.

Let me explain further what I mean by that:

Artists depend upon engineers to do just that: engineer.

Once you engineering blokes start to wax philosophical, you lose our confidence. Please. We have enough of the fuzzy stuff inherent in OUR field as artists. We don't look to you for the fuzzy stuff. We look to you engineering folks for solid ground.

We need solid tools to accomplish our sometimes abstract goals.

The realm of  "this will make it sound good" is the realm of the artist's subjectivity, NOT the engineer's. The engineer's mission is to provide tools and services which will establish an even playing field for ALL artists to accomplish their goals.

Once the engineer imposes his predgudice upon the medium, he immediately skews the result in a most unproductive fashion, as you have demonstrated.

I would suggest that you and Max Gutnick re-examine your role in our industry, and then re-position your company's mission in a pro-active fashion.

bobkatz

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2926
Re: Proper word clock implementation
« Reply #78 on: November 07, 2004, 10:25:53 AM »

Eric Vincent wrote on Sun, 07 November 2004 05:11

Quote:

posted by Lucas van der Mee:
technology and measurements are only one part of developing equipment. Of course we all want absolute values, facts. But experience has taught me that objectivity always comes after the subjective: why does something sound good? Instead of: this will make it sound good.


Lucas,

That statement makes it appear as if you should get out of the engineering business, and be an artist instead.




Well, I disagree. Lucas's primary contention is absolutely important. It IS the role of engineers and scientists to find out WHY something sounds good. Psychoacoustics is a science; just not as well-understood a science as the design-engineering science that we're discussing in this board.

If there is an apparent discrepancy between any or some of the measurements, it IS OUR job as engineers and scientists AND listeners to try to resolve that discrepancy. This will only improve the science and the engineering, and help to make better equipment in the long run.

For example, we need to subjectively/objectively test Apogee's contention that an external clock will improve the sound of "any converter", regardless of the scientific theory. Regardless that EVERY other respectable converter designer would disagree.

And why? Because we do not know everything about the subjective issues here. I personally disagree with their contention, not only because it disagrees with the science, but also because my past tests and experiences contradict their contentions. But I am  willing to test the box, as objectively as I can, both listening and measurements---because there we have to leave room for doubt, or we would not be true scientists.

Let's summarize: Apogee claims that the Big Ben improves the sound (and lowers the jitter, it's in their ads) of EVERY converter (except ones which contain the Big Ben ). Would it be sufficient to disprove their claim if I find a converter which does not improve with the Big Ben? Would it be sufficient to disprove their claim if I find a converter which sounds audibly worse when fed from the Big Ben? What about the measurements? What if the measurements and double-blind listening tests disagree? What then?

Responsible engineers should be prepared to deal with all of the above possibilities.

In the past, when investigating inferior converters that are susceptible to jitter, I discovered that measuring the absolute RMS (or peak to peak) value of the jitter at the converter's clock pin----DID NOT CORRELATE well with the listening experience. The spectrum of the jitter itself had an influence on the sound of the converter! In fact, to some degree, you could even "tune" the sound of a converter by the spectrum of the jitter, and whether you made it random or correlated to some degree.

But I did find that ALL OF THIS nonsense became a moot point when we finally learned how to make converters with sufficiently low measurable levels of jitter such that the jitter spectrum was insignificant. The point being that an inferior converter could conceivably sound better with the Apogee clock. (Ho hum....)

My point being that there are explanations to everything... we just have to find them! Science and engineering includes the science of psychoacoustics, after all! And we're going to have to live with fuzzy-logic for a long time, guys... until we have a much firmer grasp on the psychoacoustics.

My biases not withstanding, I promise to do a thorough examination of the Big Ben. This could take weeks to months. Every weekday I have is booked, and I've been working weekends (mastering or doing studio maintenance/updates) more often than you can believe.
Logged
There are two kinds of fools,
One says-this is old and therefore good.
The other says-this is new and therefore better."

No trees were killed in the sending of this message. However a large number of
electrons were terribly inconvenienced.

maarvold

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 853
Re: Proper word clock implementation
« Reply #79 on: November 07, 2004, 03:03:05 PM »

Mr. Lavry & Mr. Katz (Dan & Bob didn't seem respectful enough): how do we know what to listen for?  I bought Aardsync in the Digidesign 888/24 days and used it to provide a discreet Superclock feed to each 888/24 (and 1 to the USD) and I seem to remember thinking that:
1. Recordings I made of acoustic piano sounded more 'real' when played back with Aardsync as the master clock (rather than USD in the 'approved' Digi daisy-chain arrangement)
2. Apparent depth and width of reverb seemed more fully 'fleshed out' at the rear of the soundstage with Aardsync
3. The apparent frequency response in the 200-500 Hz range seemed to change by a dB or more with Aardsync instead of USD.  I seem to remember it was 'leaner-sounding' with USD

Now I own Pro Tools HD and monitor the AES (internal) out via a TacT RCS 2.0 [digital] room correction device.  Since reading this thread (in its earlier phases), I have switched, from Aardsync, back to using "Internal" of the 192 I monitor with as my master clock.  I do hear a difference (certainly more subtle than with 888/24 and Aardsync vs. USD) between Aardsync and Internal as the master clock, but which is better does seem to be much more (potentially) subjective than in the past.  Can you say what parameters, listening-wise, tell you that you are on the right track with clock integrity?  Thank you both in advance.  
Logged
Michael Aarvold
Audio Engineer

Curve Dominant

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 774
Re: Proper word clock implementation
« Reply #80 on: November 08, 2004, 03:32:21 AM »

Quote:

posted by Bob Katz:
Eric Vincent wrote on Sun, 07 November 2004 05:11
Quote:

Quote:

posted by Lucas van der Mee:
technology and measurements are only one part of developing equipment. Of course we all want absolute values, facts. But experience has taught me that objectivity always comes after the subjective: why does something sound good? Instead of: this will make it sound good.


Lucas,

That statement makes it appear as if you should get out of the engineering business, and be an artist instead.




Well, I disagree. Lucas's primary contention is absolutely important. It IS the role of engineers and scientists to find out WHY something sounds good. Psychoacoustics is a science; just not as well-understood a science as the design-engineering science that we're discussing in this board.



Bob,

Let me try to explain the context of the point I was attempting to make.

Every artist has tools at his disposal which can introduce non-linearities into the audio so long and so far as he chooses, and to the desired degree.

But control over that audio depends upon a baseline signal chain which is clean, minimal, and accurate.

What I had suggested to Dr. Van Der Mee, is that the engineer's preference for coloration of that signal chain is counter-productive to the artist's means.

Let's use DSD as an example. The one-bit conversion inherent in DSD introduces correlated distortion which cannot be removed from the source audio. Some people may find that distortion "pleasing," but those who do not are stuck with it nonetheless.

Such as it is with the euphonics inherent in external clocking. Certain demonstrations of the "Big Ben Sound" may be "pleasing" to some ears. But let's face it: It is an unnecessary step introduced into the signal chain, which introduces artifacts which once recorded cannot be removed from the audio after the fact.

Which brings us around to the point I attempted to make: Let artists distort the audio, IF THEY WANT TO. But, let's not give design engineers carte blanche when it comes to building what they percieve to be "euphonic" effects into the kit.

If Apogee's kit is essentially a distortion box, then let them sell it as that. Hell, maybe I'll buy one, if that's what I know I'm getting. I LOVE distortion. But as a composer, I would like to have a clean signal path to work with as the baseline audio. THEN introduce "euphonic" distortion effects to the degree I desire, in a manner I can control, IF and WHEN I want it, and to a degree which I can determine.

Does that make sense?

Curve Dominant

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 774
Re: Proper word clock implementation
« Reply #81 on: November 08, 2004, 04:07:00 AM »

Quote:

posted by Bob katz:
Science and engineering includes the science of psychoacoustics, after all! And we're going to have to live with fuzzy-logic for a long time, guys... until we have a much firmer grasp on the psychoacoustics.


Look here, mate. I would encourage you engineering blokes to ditch the psychoacoustics shtick, and ditch the fuzzy logic stuff while you're at it.

I'm a lyricist and composer, so it might seem weird to hear that from me.

But when I'm recording an artist, I switch hats, and then it's all about the logic and the science.

If you want to go "fuzzy," then write some songs, and you'll be an "artist" and you can get as fuzzy as you damn well please.

But when I'm recording a compelling artist performing a profound statement....I don't want no friggin' fuzz in the signal chain.

If I do, I can instantiate the SansAmp plugin, and add fuzz. But I don't want design engineers forcing it on me defacto.

Quality A-D conversion is pretty cut-and-dry business. Jitter, monotonicity, opamp rise time, capacitor dialectric absorption coefficients...y'all do your job, and we do ours.

bobkatz

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2926
Re: Proper word clock implementation
« Reply #82 on: November 08, 2004, 11:37:36 AM »

Eric Vincent wrote on Mon, 08 November 2004 03:32



If Apogee's kit is essentially a distortion box, then let them sell it as that. Hell, maybe I'll buy one, if that's what I know I'm getting. I LOVE distortion. But as a composer, I would like to have a clean signal path to work with as the baseline audio. THEN introduce "euphonic" distortion effects to the degree I desire, in a manner I can control, IF and WHEN I want it, and to a degree which I can determine.

Does that make sense?



What you are advocating is that engineers advertise their "color-making" boxes as making colors. At the end of the day, it can be effectively argued that EVERY single processor and converter is a color-making box. Just that some produce less color than others. Your argument has merit but I don't think in practicality it will fly.
Logged
There are two kinds of fools,
One says-this is old and therefore good.
The other says-this is new and therefore better."

No trees were killed in the sending of this message. However a large number of
electrons were terribly inconvenienced.

jfrigo

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1029
Re: Proper word clock implementation
« Reply #83 on: November 08, 2004, 12:48:20 PM »

re: Apogee

Dan's "bedside manner" may be brusque, but without Apogee opening their strong technical claims to peer review, we can't be sure of their merit. I understand the need to keep trade secrets, but perhaps some objective test results could be posted so that at least we could see the proof in the performance, if we can't get a decription of the innovation.


re: Eric & psychoacoustics

We can't just ditch psychoacoustics. Though it should not be used (or abused) as an excuse for bad engineering, it's an important part of making the technology work. Without knowing a lot about how we hear, one can't make good decisions about the sample rate and bit depth required for transparent reproduction, including how low dither must be before we don't hear it, and the best way to impliment noise shaping curves, and even some filter issues. The problem is when the factual basis is perverted into pseudo-science and used to falsely support fantastic claims; but this is not the only science that is used so.

Logged

Lucas van der Mee

  • Newbie
  • *
  • Offline Offline
  • Posts: 12
Re: Proper word clock implementation
« Reply #84 on: November 08, 2004, 06:04:45 PM »

Bob, we would love having you test the Big Ben. Please email Max to arrange for the unit.

But let me put a couple of things straight, because there seems to be some misinterpretation on what we claim and say.
We never said it improves every converter. We would not be able to substantiate this, since we simply haven’t tried them all, that would almost be impossible. But we did find improvements with most of the converters we tried, the exception being converters using a SRC for clocking, and got similar reports from users in the field. One of the known examples is the popular Digidesign 192 I/O unit.
However, we do not claim we lower the jitter of a converter by clocking it to a Big Ben, that is not what we are saying.
Let me quote from our website:

A Cure for the Jitters – Apogee’s new “C777” Clock
At the heart of any word clock regeneration is a phase lock loop (PLL). The PLL is a combination of analog and digital elements without the ability to dynamically adapt to the nature of the external clocks. In other words, the better the clock performance, the less flexibility it has. And conversely, the more flexibility there is in a PLL, the less effective it is at jitter reduction. Apogee’s new C777 is built to address these inadequacies. The C777 utilizes an entirely digital process that Apogee has developed using the most advanced Direct Digital Synthesis (DDS) technology available along with DSP based digital filtering. The result is the most aggressive jitter reduction ever. With the flexibility and stability provided the C777, Big Ben is able to re-clock devices with excessive jitter and function as if it were the master clock. The ultimate in flexibility combined with unprecedented levels of performance make the C777 unequaled in the world of clocking.


In other words, when you HAVE to clock to an external , jittery, source, you will find that your results will be much better if you have the Big Ben do the clocking first and clock your device to the Big Ben. This is objective, measurable and verifiable.

Then there is the much more difficult to explain phenomenon of converters starting to sound “better” when clocked to the Big Ben as a master clock only; the C777 running on internal. That is something we found empirically and like I said in my previous posting, we researched and optimized.


Eric, I have learned in my life that the danger with many (technical) engineers is that they start to think they are God; they think they know it all and can explain it all. I take a different approach, which is not unique, there are many with me. I see the scientific world as a way to explain what we experience.
The process of learning is isolating entities from its environment. By isolating we simplify and are able to make a model that resembles our experience. Yet very often we find that an entity we isolate, turns out to be much more complex than we originally thought. Or the connection we thought that was there, is not the connection, but the result of something bigger we had not identified yet. My school book example: In the medieval period scientists believed, mice would be created by old rags (Spontaneous Creation or Abiogenesis). Sounds funny and stupid, right? Still, nowadays scientists find everyday they make “goofs” like that all the time.
I have a university degree in Electronics comparable to a Masters, have a scientific background, I rely on math for the design of my converters, I use measuring equipment to get verifiable and objective test-results of what I build. But I also know there are parameters that are not being measured, simply because we haven’t quantified them yet. For that reason I trust my ears as much; they very often tell me a lot more than what I measure. A good example is the choice of op-amps in my analog design (which I love btw, Dan. I was just being ironic).
Last, the Big Ben is not a distorto box. The most common comments we get from experienced listeners, who use the BB are: converters sound less harsh, imaging is greatly improved, more depth and a much clearer soundstage. If you think that is the result of distortion, I am happy to learn more about this new insight, fill me in.

Jay, do you think we would allow everyone to use the best audio measuring gear in the world by Rohde & Schwarz (…does this get me a free unit?...) on their booth to test our equipment, if we felt our units were not up to par? After the show, the R&S rep raved about how impressed people were by the measurable quality of the AD16X.

Lucas van der Mee
Sr. Design Engineer
Apogee Electronics
Logged
Lucas van der Mee
Sr. Design Engineer
Apogee Electronics

bobkatz

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2926
Re: Proper word clock implementation
« Reply #85 on: November 08, 2004, 07:00:24 PM »

[quote title=Lucas van der Mee wrote on Mon, 08 November 2004 18:04]Bob, we would love having you test the Big Ben. Please email Max to arrange for the unit.

I may have lost Max's email, can you please provide it in a PM.

Sounds good.  The good thing is that you have come to save the day, Lucas, and that I urge you to get together with Max to change the misleading language that he uses to sell the box. It is 180 degrees apart from the language that you use in your post. I would have (little or) no quarrel if Max had used the correct technical language.

Quote:



However, we do not claim we lower the jitter of a converter by clocking it to a Big Ben, that is not what we are saying.




Well you and Max need to agree on what language you are going to use. Max was trying to tell us that engineer A and B and C are raving over the Big Ben's ability to make a converter perform better than on Internal Clock! Which as Dan Lavry points out, is technically impossible unless the internal clock of the converter under test is a piece of crap.

Quote:



In other words, when you HAVE to clock to an external , jittery, source, you will find that your results will be much better if you have the Big Ben do the clocking first and clock your device to the Big Ben. This is objective, measurable and verifiable.




Guys...the above is VERY different in meaning and intent from what Max was saying in this thread. It agrees with Bob Ludwig's reactions as well. "If you must externally clock (as in "lock to video") then consider the Big Ben.

So I think the extent of my testing should be to see how TYPICAL converters perform with the Big Ben or without the Big Ben when they MUST be externally clocked. Now we're cooking with gas!!!

Quote:



Then there is the much more difficult to explain phenomenon of converters starting to sound ?better? when clocked to the Big Ben as a master clock only; the C777 running on internal. That is something we found empirically and like I said in my previous posting, we researched and optimized.




This is a much harder contention to sell and disagrees with the theory IF a converter is well-made and contains a good internal clock. But there are a lot of interior converters on the market. We shall see.

I'm curious if the stable of converters that I can test will improve with the Big Ben:

Cranesong HEDD-192, Benchmark A/D, TC Electronic System 6000, DCS


Place your bets  Smile


Best wishes,


Bob
Logged
There are two kinds of fools,
One says-this is old and therefore good.
The other says-this is new and therefore better."

No trees were killed in the sending of this message. However a large number of
electrons were terribly inconvenienced.

Max

  • Newbie
  • *
  • Offline Offline
  • Posts: 45
Re: Proper word clock implementation
« Reply #86 on: November 08, 2004, 07:46:42 PM »

Well, I really intended to be done with this thread. I am a little confused at your responses, Bob. you said:

Quote:

Sounds good.  The good thing is that you have come to save the day, Lucas, and that I urge you to get together with Max to change the misleading language that he uses to sell the box. It is 180 degrees apart from the language that you use in your post. I would have (little or) no quarrel if Max had used the correct technical language.


What language was it that mislead you specifically as I do not see the conflict?

Quote:

Well you and Max need to agree on what language you are going to use. Max was trying to tell us that engineer A and B and C are raving over the Big Ben's ability to make a converter perform better than on Internal Clock! Which as Dan Lavry points out, is technically impossible unless the internal clock of the converter under test is a piece of crap.


I stand by this statement. it is the same thing Lucas is stating at the end of his post where he says:

Then there is the much more difficult to explain phenomenon of converters starting to sound ?better? when clocked to the Big Ben as a master clock only; the C777 running on internal. That is something we found empirically and like I said in my previous posting, we researched and optimized.

Like Lucas is saying above and as I said in my previous posts, there are many many users out there that have used a Big Ben to clock a single converter and noticed a marked improvement in the sound quality and this includes converters across the quality spectrum. Again, Bob, my point in bringing this up to you was your comment that if a converter sounds better externally clocked there must be something wrong with the converter. I do not agree with this assessment based on our results, as I am not prepared to say that many of the converter brands we tested and have heard about from end users are bad or poorly designed. To do so would be the blatant salemanship and marketing that you previously were accusing me of and I am not prepared to do that, even though doing so would most likely result in us selling more Apogee products.


Quote:

This is a much harder contention to sell and disagrees with the theory IF a converter is well-made and contains a good internal clock. But there are a lot of interior converters on the market. We shall see.

I'm curious if the stable of converters that I can test will improve with the Big Ben:

Cranesong HEDD-192, Benchmark A/D, TC Electronic System 6000, DCS


Place your bets  Smile


Best wishes,


Bob


Bob, we look forward to getting you the unit and hearing back regarding your results. While under the circumstances it will not be possible for us to be there as I would have wished, I believe the process you outlined in your previous post as to how you would go about testing will work for us. I would still appreciate interfacing with you directly on this so feel free to email me at your convenience at max@apogeedigital.com
Logged
Max Gutnik
Apogee Electronics

danlavry

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 997
Re: Proper word clock implementation
« Reply #87 on: November 08, 2004, 07:54:44 PM »

Hello Bob.
I.   Re: External box driving A/D

You said:

"Well, I disagree. Lucas's primary contention is absolutely important. It IS the role of engineers and scientists to find out WHY something sounds good. Psychoacoustics is a science; just not as well-understood a science as the design-engineering science that we're discussing in this board."

The short answer is: If this manufacturer wishes to sell a solution based on “sound better” that is fine. However, to fulfill as you suggested, an honest attempt at figuring what is going on, takes more then “we were dumbfounded” or “it sounds good”. An honest attempt could be to say “MORE JITTER but it sounds better”, which is very different from “A cure for all jitters”.

With all due respect, it is my considered opinion that your “open minded approach” postpones and delays the obvious. Their claims are akin to being able to fly by waving ones arms.  

At some point, the technical engineer should put his foot down! More jitter is worse. Using fewer bits is worse.  Audio is full of “I don’t know why but it sounds better”, which in my view is, most often, an excuse for selling lesser performance gear. That is “marketing”. Name the car that breaks down “reliant”, the car with bad gas mileage “econo-line”, and the slow one “Cheetah”.

"If there is an apparent discrepancy between any or some of the measurements, it IS OUR job as engineers and scientists AND listeners to try to resolve that discrepancy."

I do not claim to be a listener. I work at creating transparent uncolored sound that the artists then color according to their own sensibilities but measurements are indeed what I pay attention to. In this case, there is no need to measure jitter because the concept is fundamentally flawed. An external clock box cannot reduce jitter inside an A/D box.

"For example, we need to subjectively/objectively test Apogee's contention that an external clock will improve the sound of "any converter", regardless of the scientific theory. Regardless that EVERY other respectable converter designer would disagree."

At some point things get out of hand. What about a claim that 8 bits sound better? How about a claim that we need 100MHz of audio bandwidth? There is not enough time to investigate every claim made up by some sales guy. We engineers know there are some lines that cannot be crossed. More jitter is bad. Lucas admitted that in his first post.

"Let's summarize: Apogee claims that the Big Ben improves the sound (and lowers the jitter, it's in their ads) of EVERY converter (except ones which contain the Big Ben ). Would it be sufficient to disprove their claim if I find a converter which does not improve with the Big Ben? Would it be sufficient to disprove their claim if I find a converter which sounds audibly worse when fed from the Big Ben? What about the measurements? What if the measurements and double-blind listening tests disagree? What then?"

What would you do if I told you that doubling the voltage on a resistor reduces instead of doubles current flow? It would contradict Ohm’s Law! Will you bother to test that? I know I will not. As you know and stated, listening tests are most susceptible to subjectivity. Basic science and engineering is not. We are not philosophizing, we are dealing with Ohms law, Nyquist, Laws of thermodynamics, basic engineering theories.

"Responsible engineers should be prepared to deal with all of the above possibilities."

I disagree! A responsible engineer is one that while using his/her solid background can decide where to draw the line between what is fundamentally correct, fundamentally flawed and what may be fuzzy. There is no perpetual motion. Ohms law is solid. Nyquist was right. And so on.

"My point being that there are explanations to everything... we just have to find them!"

Some explanations are marketing based and go against engineering. Let’s get back to the subject at hand, and do a “think like a designer” exercise.

Problem: There is a box. It is a closed box, you can not see what is inside. It may be an AD with a perfect zero jitter clock circuit. It may be a jittery circuit. It may even be completely broken, not working. Your mission, should you accept it, is to send a message to the box telling the circuitry to do a better job. You are allowed to send any signal imaginable into that box, via a cable connection. But remember one thing, you are allowed to talk to the AD box but the box cannot talk to you.

THE BOX DOES NOT SEND ANY INFORMATION BACK TO YOU. YOU DO NOT KNOW EVEN FOR AN INSTANT, WHAT IS TAKING PLACE INSIDE THAT CLOSED BOX.

Say you knew that the PLL and clock inside the AD work perfectly. Then you should send a message with a “content to do nothing.” Say you knew that every third clock edge is late by 1 nsec. Then you should (assuming it is possible) send a message to generate every third clock 1 nsec earlier.

Sending a message to a closed A/D box to correct what it does (WHILE NOT KNOWING what it does or what its internal problems may be eg: noisy power supply) is fundamentally an impossible task. I am sorry, the Big Ben box cannot possibly do this.
.
In a very GENERAL sense one can know what is inside each A/D. There is a clock with some level of random jitter activity. Then perhaps some signal correlated jitter (or maybe not), perhaps some power line induced jitter (or maybe not), or jitter due to noise on the supply (or not)? These factors differ from AD to AD…

Conceptually and theoretically speaking, one has a chance to make some corrections when there is feedback. But here there is NO FEEDBACK. Feedback REQUIRES that the AD chassis send information back to the external clock unit, telling it what is going on at the AD. But in this case, Big Ben is driving the AD box, and there is one way communication- NO FEEDBACK. Therefore the claims stand against ENGINEERING INTEGRITY.

I will be generous and say that giving credence to such fallacies only encourages a form of marketing that confuses those who want to be well informed. By insisting that this forum be a technical one we provide a service to be trusted! This is the place where Ohm’s Law always works and jitter is always bad…

NEXT TOPIC:
II. Re: Clock itself

How good a clock can one make with the DDS technology? Again, it is a simple matter to see through the faulty logic.

To talk about secret technology of the 21st century is bogus. Hints about filters and “analog crap”  is  also bogus. How do I know so much about what Lucas did without knowing what he did? I will explain. Again, let’s “think like a designer”:

Whatever one measures or generates calls for a REFERENCE. A yard stick. Your old voltmeter had a scale, like a ruler. The “needle” moved against the scale telling you the voltage (the printed “ruler scale” on the front panel was the reference). The new digital voltmeter has a precision internal DC reference (such as a reference zener diode). Any measurement is always compared to a reference. This procedure is fundamental to science and engineering. If the reference is moving up and down by 1%, the measurements are off by 1%.

One can only get as good as ones reference. The only way to do better is to get a more accurate reference, a better, more accurate “yard stick”. If that better reference is say good to 1mV than you could know that your outcome to 1mV. One cannot correct things beyond the finest reference one has.

The same statements are true for measuring or generating current, resistance, frequency and mechanical construction to name a few..  One can have all sorts of circuitry help the outcome get closer to the reference, but one can not have a better performance then the reference itself. The DDS technology relies on a reference clock. If this reference clock can not beat the jitter of an internal crystal, all the circuits in the world can not help it. IT CAN NOT.

So I do not need to know any “secrets” those guys are talking about. No amount of DSP or hints about filters or circuits AFTER the reference clock are going to do a bit of good, and no amount of disrespectful talk towards this “theoretical guy” is going to alter the reality of what I say. The “war” for low jitter takes place at the reference clock first and foremost. Call it theory if you wish, but it is also a fact of life.

A good implementation of internal clock uses a crystal that resonates in a fundamental mode.  But fundamental crystals can not be made for very high frequencies.  DDS requires a higher frequency reference clock, thus the crystal that operates on more jitter. If the reference has jitter, you can not do a thing about it. Can you filter it with DSP? Build a notch filter to remove unwanted energy? Nuts! The notch frequency moves with the reference clock (it moves with the jitter). And so on…

Moving from theory to practice and “hands on” engineering, it is best to get a reference clock to drive a signal as directly as possible to the AD. Going from a higher jitter reference clock through all sorts of DDS circuitry, then to a driver circuit, then to a cable (possible interference), then receiver IC, then a PLL (more jitter) driving a VCXO (or anything with higher jitter then a fixed fundamental crystal)…. That can not compete with a good fixed crystal very near the destination. Check Mate!

I just explained why two of the Apogee arguments are incorrect. So why did they get a tech award? Must be marketing $ certainly not technological achievement.

At this forum technical know-how has value, without being subject to questions by listening tests which are accompanied by hype, name dropping,  subjective arguments and last but certainly not least- politics.

Best Regards
Dan Lavry

Logged

jfrigo

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1029
Re: Proper word clock implementation
« Reply #88 on: November 09, 2004, 01:11:03 AM »

Lucas van der Mee wrote on Mon, 08 November 2004 15:04

Jay, do you think we would allow everyone to use the best audio measuring gear in the world by Rohde & Schwarz (...does this get me a free unit?...) on their booth to test our equipment, if we felt our units were not up to par? After the show, the R&S rep raved about how impressed people were by the measurable quality of the AD16X.

Lucas van der Mee


I'm sure you worked very hard on your unit and are proud of its performance. I don't doubt Apogee's pursuit of quality. My only point is that it's reasonable and even expected for the audio public to say "show me." This is not to say whether the box is a huge success or or not as I haven't tried it;  just that it's a smart consumer who doesn't simply take the word of every advertiser he encounters in the magazines.
Logged

Schallfeldnebel

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 816
Re: Proper word clock implementation
« Reply #89 on: November 09, 2004, 11:36:42 AM »

When I was looking on the internet site of Apogee at the Big Ben, I got the impression this unit is much more then a simple clock generator.  Maybe I did not understand this correctly. I get the impression the BB also functions as a format converter and a sample rate converter.

If you combine those functions with an ultra low jitter internal clocking, this unit starts to work as a de-jitter generator. Mr. Daniel Weiss made such a device years ago, and called it Clockwork.

I have tested in the past intensively with the Clockwork, and it made a lot of sense hanging it between CD transports and DA converters. Often it sounded like having a better transport, or having beter DA converter.

All those studio’s which are claiming to have a better sound using the BB, are not very specific in how they use it. If they use it, to clock their workstation or recorders, it might be that what they are doing is, sending a better clocked signal to the DA converter. In my experience indeed  reclocking the signal gives better sound at the DA stage.

Or, if a signal coming from an AD converter is sent into the BB, and the BB is used to clean up that signal, so BB functions as a reclock generator, again it would be possible that the end result is sounding better. But it is not the AD conversion which sounds better, it is the final signal going to the DA converter which is cleaned up. In this case the output signal from that AD converter must have had quite much pollution.

Also when BB is used as a central clocking unit, and it is clocking an AD converter, but also your digital workstation and recorders, again you are cleaning up the whole signal path, and it is again the DA converter which resposible for the better sound, because it got a clean clock.

Only when an AD converter is clocked from the BB, and the AD converter is clocking with it’s own AES output signal your workstation, and the workstation is clocking with it’s output signal your  DA converter, I cannot believe it will sound better, at least I would be very surprised.

I have clocked several AD converters external, and the sound was always worse, and when the output signal jitter was measured, it was always 2 to 3 times higher than when internally clocked. There was one exception where it did not matter, it sounded the same and measured the same, and this experience I had with a converter from Daniel Weiss.

So we should not make the mistake when concluding external clocking of AD converter gives a better sound, that we clock all other equipment after the AD also directly out of the BB, because then you clean up the signal of the whole system, which can have a large effect on the DA converter, so it is the DA converter making us believe our AD sounds better.


Erik Sikkema

Logged
Bill Mueller:"Only very recently, has the availability of cheap consumer based gear popularized the concept of a rank amateur as an audio engineer. Unfortunately, this has also degraded the reputation of the audio engineer to the lowest level in its history. A sad thing indeed for those of us professionals."
Pages: 1 ... 4 5 [6] 7 8 ... 15   Go Up
 

Site Hosted By Ashdown Technologies, Inc.

Page created in 0.082 seconds with 19 queries.