R/E/P Community

Please login or register.

Login with username, password and session length
Advanced search  

News:


Author Topic: Blue and Gold  (Read 8759 times)

Offline mcsnare

  • Hero Member
  • *****
  • Posts: 958
Blue and Gold
« on: October 03, 2004, 11:59:40 am »
Dan,
  Thanks for taking the time to share your knowledge here.
  I have a pair of Blue series D/A's. I love the sound, but was wondering what the design and or build differences are between those and your top of the line Gold series?
Thanks,
Dave McNair

Offline Ivo

  • Full Member
  • ***
  • Posts: 169
Re: Blue and Gold
« Reply #1 on: October 05, 2004, 12:19:12 pm »
I too would be very much interested to hear something about the basic design (and sound) differences between these two models (which are several thousand dollars far away from each other). I have Lavry Blues, they already sound stellar to me ... so I wonder how much further can the quality od AD DA conversion still reach. Dan, could you please qive us few basic thoughts ? Anyone having both and able to compare ? Thank you.
Ivo

VELVET MASTERING
www.velvetmastering.com

SAVITA MUSIC
www.savita.cz

Offline danlavry

  • Hero Member
  • *****
  • Posts: 997
Re: Blue and Gold
« Reply #2 on: October 06, 2004, 07:14:20 pm »
I too would be very much interested to hear something about the basic design (and sound) differences between these two models (which are several thousand dollars far away from each other). I have Lavry Blues, they already sound stellar to me ... so I wonder how much further can the quality od AD DA conversion still reach. Dan, could you please qive us few basic thoughts ? Anyone having both and able to compare ? Thank you.

I am told I should answer direct questions about my gear on the forum. I’ll try to stay as technical as I can, within the spirit of the forum. But before I answer your question I feel compelled to give some technical background.

Why is there such a difference between converters based on the same IC?
The IC makers work with a very restricted and limited boundaries. They try to fit as much as possible into a chip the size of a “pin head”. They must “limit the activity” to say l watt or so of power, or else the IC will overheat. They can not use inductors, and all the capacitors tiny (a few pF)… IC based on say a 5V supply, is not going to accommodate say +/-15V for good analog… no poly caps, no specialty items…
It is true that IC designers also have some advantages. Good matching and tracking is due to the fact that all components are “curved out of” the same materials and operate at the same environment of that tiny silicon chip. But all things considered, it is a “wonder” how well the IC makers do within their limitations!

Almost all AD and DA’s today are made around a ready made IC. In other words, some designers of your gear often do not really have much understanding of the theory behind the conversion process. Often they lack knowledge about the difficult issue regarding circuits such as the sample hold, resistor stability for PCM, the math behind high order noise shaping for sigma delta…  I only recently came to realize how much of the gear on the market is the fruit of the “glue that IC on the printed circuit board and put it in a nice looking chassis” mentality. While protesting the use of 192KHz for audio, I came across EE’s that did not understand Nyquist theorem. There are EE’s out there who don’t know that impulse response and signal bandwidth are one and the same. There are a many more EE’s today, but fewer than ever analog types. Not surprising in this computer age…  

I am not saying that the AD or DA equipment designer has no challenges. One can and should go beyond a simple copy the manufacturers “typical application” on the data sheet. The AD front end analog circuit, DA output stage, clocks circuits and more are the business of the gear designer, less so of the IC maker. The good designer will not hesitate to bypass parts of an IC, such as up sampling or down sampling filter with external hardware driven by external coefficients. Coefficients from where? Again, you can “buy a ready solution” (software to generate coefficients) or even better, do the math yourself… But you have to know math… There is a lot that can be done to make an IC based solution work well.

The above comments apply to the LavryBlue. What about the Gold series?
Say you want to go beyond what you can find in an IC. Say you want to push new frontiers, what is called state of the art. New concepts, system architecture and ideas not found on an IC (some even incompatible with IC technology) will force you to make a converter from “basic building blocks”. Such “from the ground up” approach requires years of product development. The Lavry Gold converters are such products. They are designed from the ground up. You start with quality resistors and go from there. The design is not the old design with better parts. It is a better design aimed at overcoming the limitations of the old design.

Obviously I was very familiar with PCM DA’s. I made my first audio DA while still employed at Analog Solutions – the first segmented architecture type DA. That design was later repackaged by Ultra Analog, and found it’s way to Mark Levinson, Wadia, and Pacific Microsonics. Not a bad design, but PCM had a number of flaws. So I decided to list the flaws and solve ALL of them in my next DA.  

Here are some examples:
Problem: Resistors change with temperature
Solution: Put them in an constant temperature linearly controlled oven
Problem: Resistors age
Solution: An elaborate automatic calibration scheme.
Problem: PCM performs worst at digital black (Pointed out in a paper By Dr. Lipshitz)
Solution: Create a network with “digital DC” shift
Problem: DA’s are susceptible to small jitter
Solution: Processor based re clocking scheme CrystalLock (TM) provides over 200 times improvement….

Obviously, the Gold series (from the ground up) product is a difficult way to go!
The Gold AD is great, but I made changes to allow the unit to run cooler.  The upgraded MKII+ dissipates 19 Watts (MKII was 30 Watts) and a complete AD module repackaging job with 6 heat sinks plus air vents, and a new potting material ( which allows heat to dissipate more evenly) is a dramatic improvement.
 
The Gold DA is my favorite design.

BR
Dan Lavry



Offline Tomas Danko

  • Hero Member
  • *****
  • Posts: 4733
Re: Blue and Gold
« Reply #3 on: October 07, 2004, 03:47:04 pm »
I just wanted to step out and thank you so very much for actually taking your precious time to explain all these issues for the rest of us. Especially the fact that you're apparently willing to share some of your fundamental perception of views, sharing to us some glimpses of your initial conceptual thinking when creating something as beautiful as your Gold series converters.

As much as the REP-forums are filled with amazing and talented people sharing their knowledge using a warm and friendly attitude, I have to say that you even manage to rise one notch above everybody and it just opens my eyes and mind in a stunning manner.

Off topic and all, but this time I could not resist expressing just how much I appreciate your contribution here.

Thank you, Mr. Lavry.

Profoundly yours,

Tomas Danko
www.danko.se
http://www.danko.se/site-design/dankologo4s.gif
"T(Z)= (n1+n2*Z^-1+n2*Z^-2)/(1+d1*z^-1+d2*z^-2)" - Mr. Dan Lavry
"Shaw baa laa raaw, sidle' yaa doot in dee splaa" . Mr Shooby Taylor

Offline jfrigo

  • Hero Member
  • *****
  • Posts: 1029
Re: Blue and Gold
« Reply #4 on: October 07, 2004, 10:26:03 pm »
danlavry wrote on Wed, 06 October 2004 16:14


The Gold DA is my favorite design.



The gold DA is remarkable.

Many other DAs on the market sound good, and I don't feel the quality differences are quite as profound as some do among some of the popular units. For example, in the more affordable range, I clearly preferred the Lavry blue to the Benchmark, but I didn't think it rendered the Benchmark irrelevant. It's still a good performer at a good price point.

However, when we rented a Lavry gold DA for a few jobs, it surprised me just how remarkable the difference was. With the gold DA in the room, the Benchmark really did get blown out of the water, and DAs I thought previously sounded good were called into question. Even the Lavry blue is not immune to a butt kicking by the gold. It was an eye and ear opening experience and I had not anticipated such a distinct improvement. Now I just need to hope for a winning lottery ticket!


Offline mcsnare

  • Hero Member
  • *****
  • Posts: 958
Re: Blue and Gold
« Reply #5 on: October 08, 2004, 01:51:58 am »
Dan, thank you so much. Fascinating and insightful.
Dave McNair

Offline dnafe

  • Jr. Member
  • **
  • Posts: 64
Re: Blue and Gold
« Reply #6 on: October 08, 2004, 09:38:15 am »
Hi All

Please excuse my total ignorance on this subject but I have a question.

Where in the signal chain does a DAC sit and how does it affect the final product? ie: does it improve the tracks themselves or does it improve the quality of audio to your monitors enabling you to hear more subtle nuances of the recording?

Or am I completely off base here and babbling?

Don
Far be it from me to say anything intelligent

Offline Ivo

  • Full Member
  • ***
  • Posts: 169
Re: Blue and Gold
« Reply #7 on: October 08, 2004, 11:10:31 am »
dnafe wrote on Fri, 08 October 2004 15:38



Where in the signal chain does a DAC sit and how does it affect the final product? ie: does it improve the tracks themselves or does it improve the quality of audio to your monitors enabling you to hear more subtle nuances of the recording?
Don


Great DAC enables you to hear the mix in its true complex shape with the most subtle details present - thus you are able to decide better about further mixing or mastering steps. ANd also - if you use outboard gear for processing, the higher quality your DA AD chain is, the smaller sound quality loss happens during this extra conversion.
I fully realised the importance of high end DA when long time ago I added Apogee mini DAC to my Digi 001. The difference was simply  night and day and I was wondering - how could I listen to something like that all the time ... After I replaced Apogee with Lavry Blue DA, it was a similar step above. Now everything so clear , warm and pure. Since I do not play any lottery, I will continue being happy with the blue colour  Smile
Ivo

VELVET MASTERING
www.velvetmastering.com

SAVITA MUSIC
www.savita.cz

Offline ted nightshade

  • Hero Member
  • *****
  • Posts: 1272
Re: Blue and Gold
« Reply #8 on: October 08, 2004, 09:07:32 pm »
Thanks for a very lucid and concise explanation. It was a pleasure to read.
Ted Nightshade aka Cowan

There's a sex industry too.
Or maybe you prefer home cookin'?

Offline bobkatz

  • Hero Member
  • *****
  • Posts: 2926
Re: Blue and Gold
« Reply #9 on: October 09, 2004, 12:02:46 am »
danlavry wrote on Wed, 06 October 2004 16:14


The Gold DA is my favorite design.
The gold DA is remarkable.



I do have to put my pennies together to get a gold to try out and if I like it, buy it! But I will not try a converter or any piece of gear unless I know I can afford it, or that it's within my budget. It's not fair to the company or the designer. Or at least I'd be open about it to the company that I could not buy the piece now but I would be seriously interested in auditioning it if they did not mind. (Many companies rightly shy over that one, as they have a hard time keeping demos in stock as it is)

Here's a question for Dan:

Years ago, I found that the very first 1 bit D/A designs seemed to lack something in "punch" or apparent dynamic range. I reached the conclusion that a multibit D/A (such as in my Ultra Analog custom-built DAC) inevitably sounds better. The single bit DACs are more susceptible to interface jitter than multibit. This may have something to do with my initial reactions to single bit, and I also felt that the designers getting into the topology had to learn about the sources of noise and contamination within the 1 bit system that could reduce dynamic range.

Then, as the designs got better, I finally felt after a few years that the sonic performance, particularly the dynamics of 1 bit designs began to equal and exceed even my favorite multibits.

Now, back to the Lavry Gold. Is this a 1 bit design or a multibit (ladder) design or some sort of a "hybrid" like the DCS with its 5-bit "ring DAC" approach? I wonder if the reason that so many of my colleagues think the Lavry Gold sounds better is because it reproduces dynamics (has apparent dynamic range) better than its competition. If so, why, and how? Is there anything in the measurements that may indicate this?

After all, several DACs now on the market exhibit unmeasurable jitter products and very low noise floor with and without signal. Is there anything examplary in the Lavry Gold's measurements that we should be looking for as measurably superior to other DACs?  Or does this remain as a largely subjective area?

BK
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.

Offline jfrigo

  • Hero Member
  • *****
  • Posts: 1029
Re: Blue and Gold
« Reply #10 on: October 09, 2004, 01:05:59 pm »
dnafe wrote on Fri, 08 October 2004 06:38

Hi All
Where in the signal chain does a DAC sit and how does it affect the final product? ie: does it improve the tracks themselves or does it improve the quality of audio to your monitors enabling you to hear more subtle nuances of the recording?


Both. One typically has 2 DACs in a mastering room: one DAC for the monitoring chain, and one to feed the analog processing chain. If you use a poor DAC to feed your chain, it is the weak link and a limitation on your ability to make the most of the project. No matter how good your EQ, dynamis, and AD converter, you'll be processing an inferior source. Any shortcomings of the DAC are now contained in your signal path, and anything lost is never coming back. For the monitor DAC, you can't make good decisions if you don't know what you're hearing, or can't hear it at all. You want quality full range monitors for the same reason.

Offline danlavry

  • Hero Member
  • *****
  • Posts: 997
Re: Blue and Gold
« Reply #11 on: October 09, 2004, 05:49:29 pm »
bobkatz wrote on Sat, 09 October 2004 05:02

danlavry wrote on Wed, 06 October 2004 16:14


The Gold DA is my favorite design.
The gold DA is remarkable.



Here's a question for Dan:

Now, back to the Lavry Gold. Is this a 1 bit design or a multibit (ladder) design or some sort of a "hybrid" like the DCS with its 5-bit "ring DAC" approach? I wonder if the reason that so many of my colleagues think the Lavry Gold sounds better is because it reproduces dynamics (has apparent dynamic range) better than its competition. If so, why, and how? Is there anything in the measurements that may indicate this?

After all, several DACs now on the market exhibit immeasurable jitter products and very low noise floor with and without signal. Is there anything exemplary in the Lavry Gold's measurements that we should be looking for as measurably superior to other DACs?  Or does this remain as a largely subjective area?

BK


Hi Bob,

I thought I was at least suggesting that my Lavry Gold DAC is a PCM. I guess you would call it multibit. But these days people often refer to multibit in a noise shaping loop as a few bits (in contrast to 1 bit). My DA is not based on a noise shaping loop at all. It is not sigma delta.
However it is not the old fashion R2R binary ladder either. The conversion occurs on 1 clock edge (all 24 bits), and the resistor network is not even binary. It yields a lot more codes (quantization levels) on the positive side allowing me to add a digital DC, thus get away from the having small signal levels occur at the worst possible DA region - digital black is where all the bits switch simultaneously from 10000… to 01111… That is a problem for PCM DA’s as was pointed out by Dr. Lipshitz a few years ago.
So the DA is PCM, built with special order resistor networks, kept in constant temp (linear control oven), with half the circuit dedicated for auto calibration, with Dmos (sub nanosecond) switching type deglitcher circuit, DSP based up sampling (my own coefficients), and a RAM buffer plus processor based jitter removal (CrystalLock TM).

You seem to be interested in the jitter aspect, and the CrystalLock is great for taking jitter out far better then PLL’s. But even without CrystalLock, a sigma delta DA is a lot less jitter friendly than PCM (unless there is a new improvement I am not aware of).    

It is worth noting that some DA’s do a good job of jitter rejection by insertion of an internal sample rate converter IC. They just forget to tell you that you now have an SRC in series. I will leave it to the ear people to state if that tradeoff is good or not. I suspect the answer is somewhat tied to how much jitter is present at the input. A lot of jitter may disturb a PLL, but it will also alter the process of coefficient selection in the SRC, and it can be observed on an single sine wave FFT as “widening of the main lobe”. That is why CrystalLock is better. Imagine say a 1KHz sine wave modulating jitter to 5nSec amplitude, and you end up with no SRC sound, and no PLL sound.

BR
Dan Lavry

Offline bobkatz

  • Hero Member
  • *****
  • Posts: 2926
Re: Blue and Gold
« Reply #12 on: October 09, 2004, 07:54:03 pm »
Quote:



Here's a question for Dan:





Hi, Dan! I don't know if you saw that yes, I am aware that a delta-sigma design is potentially more susceptible to jitter than an R2R ladder DAC, as was clearly shown in Bob Adams' classic paper on the subject. And yes, I am totally aware that the SRC changes the data versus a good pll. But a good PLL is incredibly difficult to design, a you know. The engineers at TC spent one man-year on the development of the PLL in the conversion system that became the TC system 6000 A/D/A converter.

I like to look at an ASRC as a "poor man's good PLL" and as the technology has improved it is amazing how it's not such a "poor man" anymore---how good an ASRC-style DAC can sound. As you know, the digital PLL in the Analog Devices 1896 is not to be sneezed at. The sound quality of the Benchmark is very very good. (And it's not all PLL, of course; every aspect of the converter must be made good, the noise floor, the initial analog PLL (which reduces jitter into the ASRC as you described) and distortion level and solidity of the power supply all contribute.

In other words, you took the no-compromise "hard route" to a DAC design, and the result can conceivably be sonic nirvana. I keep on spending "Lavry Gold" money on other devices lately, but I promise to keep my eyes on the gold when I'm ready. I don't think I'd settle for anything less when/if I upgrade DACs!

And then comes the dilemma: Should I put the Lavry Gold in front of my analog gear for a processing loop, or use it in my monitor chain  Sad   Boo hoo.....

Best wishes,


Bob
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.

Offline danlavry

  • Hero Member
  • *****
  • Posts: 997
Re: Blue and Gold
« Reply #13 on: October 09, 2004, 10:18:54 pm »
bobkatz wrote on Sat, 09 October 2004 05:02

danlavry wrote on Wed, 06 October 2004 16:14


The Gold DA is my favorite design.
The gold DA is remarkable.



Here's a question for Dan:

Now, back to the Lavry Gold. Is this a 1 bit design or a multibit (ladder) design or some sort of a "hybrid" like the DCS with its 5-bit "ring DAC" approach? I wonder if the reason that so many of my colleagues think the Lavry Gold sounds better is because it reproduces dynamics (has apparent dynamic range) better than its competition. If so, why, and how? Is there anything in the measurements that may indicate this?

After all, several DACs now on the market exhibit immeasurable jitter products and very low noise floor with and without signal. Is there anything exemplary in the Lavry Gold's measurements that we should be looking for as measurably superior to other DACs?  Or does this remain as a largely subjective area?

BK


Hi Bob,

I thought I was at least suggesting that my Lavry Gold DAC is a PCM. I guess you would call it multibit. But these days people often refer to multibit in a noise shaping loop as a few bits (in contrast to 1 bit). My DA is not based on a noise shaping loop at all. It is not sigma delta.
However it is not the old fashion R2R binary ladder either. The conversion occurs on 1 clock edge (all 24 bits), and the resistor network is not even binary. It yields a lot more codes (quantization levels) on the positive side allowing me to add a digital DC, thus get away from the having small signal levels occur at the worst possible DA region - digital black is where all the bits switch simultaneously from 10000… to 01111… That is a problem for PCM DA’s as was pointed out by Dr. Lipshitz a few years ago.
So the DA is PCM, built with special order resistor networks, kept in constant temp (linear control oven), with half the circuit dedicated for auto calibration, with Dmos (sub nanosecond) switching type deglitcher circuit, DSP based up sampling (my own coefficients), and a RAM buffer plus processor based jitter removal (CrystalLock TM).

You seem to be interested in the jitter aspect, and the CrystalLock is great for taking jitter out far better then PLL’s. But even without CrystalLock, a sigma delta DA is a lot less jitter friendly than PCM (unless there is a new improvement I am not aware of).    

It is worth noting that some DA’s do a good job of jitter rejection by insertion of an internal sample rate converter IC. They just forget to tell you that you now have an SRC in series. I will leave it to the ear people to state if that tradeoff is good or not. I suspect the answer is somewhat tied to how much jitter is present at the input. A lot of jitter may disturb a PLL, but it will also alter the process of coefficient selection in the SRC, and it can be observed on an single sine wave FFT as “widening of the main lobe”. That is why CrystalLock is better. Imagine say a 1KHz sine wave modulating jitter to 5nSec amplitude, and you end up with no SRC sound, and no PLL sound.

BR
Dan Lavry

Offline Sam Lord

  • Full Member
  • ***
  • Posts: 139
Discrete Gold?
« Reply #14 on: October 11, 2004, 10:22:25 pm »
Yo Dan, Dan!  So the core (at least) of the Gold ADC is discrete?  I'm about to keel over from desire, and I haven't even seen a picture of it!  I remember that Ultra Analog built discrete ADCs, and I think besides you, perhaps only Accuphase (in the civilian world) still does...very impressive.  Folks have been raving about the Blue on 3dB Audio and Gearslutz, so I'll troll around the Lavry website.  Congratulations and best wishes (fine forum, too), Sam    

Offline danlavry

  • Hero Member
  • *****
  • Posts: 997
Re: Discrete Gold?
« Reply #15 on: October 12, 2004, 04:17:59 pm »
Hi Sam

Sam Lord wrote on Tue, 12 October 2004 03:22

Yo Dan, Dan!  So the core (at least) of the Gold ADC is discrete?  I'm about to keel over from desire, and I haven't even seen a picture of it!  I remember that Ultra Analog built discrete ADCs, and I think besides you, perhaps only Accuphase (in the civilian world) still does...very impressive.  Folks have been raving about the Blue on 3dB Audio and Gearslutz, so I'll troll around the Lavry website.  Congratulations and best wishes (fine forum, too), Sam    


If I am not mistaken, the Ultra Analog AD design was not all discrete. It was based on a 2 chip design by Bob Adams (now in Analog Devices). He is one heck of an engineer! He is the one that introduced sigma delta to audio. His first sigma delta design offered a significant improvement in dynamic range, I think it was 108dB or so, amazing in those days. Ultra Analog did repackage it. I am more familiar with the DA because it was a repackaging of a design I did as an employee for Analog Solution.

BR
Dan Lavry

Offline Sam Lord

  • Full Member
  • ***
  • Posts: 139
Re: Discrete Gold--Thanks for some history
« Reply #16 on: October 16, 2004, 12:16:25 pm »
>>   ...If I am not mistaken, the Ultra Analog AD design was not all discrete. It was based on a 2 chip design by Bob Adams (now in Analog Devices). He is one heck of an engineer! He is the one that introduced sigma delta to audio...   <<

Dan, thank you for the clarification and history.  This forum has carried the most comprehensive discussion on the constraints and solutions in digital audio which I have yet read.  I worked on high end analog gear for some years, and appreciate the use of electrical ("analog") descriptions of difficult A/D and D/A challenges.  Regards, Sam