Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[bobkatz]

danlavry wrote on Wed, 07 December 2005 15:21

Hi Bob, My designer perspective is that a filter is a "walk in the park" compare to many of the AD circuits.

Can't argue with a designer's perspective. When I can afford a Lavry Gold then I'll start nitpicking  . Ironically, you have to suport 192 kHz in order to prove it isn't necessary. Otherwise, any shootout against another converter manufacturer running 192K is not equal variables. The only "fair" test would be two Lavry A/D converters running different rates. Chesky wants to do a test on location and then evaluate it back at the studio. The A/D test will have one model of converter (you can guesss which) runing at  96K and another model at 192K. Parallel analog feeds on location will go to different converters running at different rates. It's just fraught with potential errors of judgment, since different brands of A/D will be used on the record side. But it will be interesting to see what the listening test shows.

BK

[danlavry]

bobkatz wrote on Thu, 08 December 2005 00:03

danlavry wrote on Wed, 07 December 2005 15:21 Hi Bob, My designer perspective is that a filter is a "walk in the park" compare to many of the AD circuits. Can't argue with a designer's perspective. When I can afford a Lavry Gold then I'll start nitpicking  . Ironically, you have to support 192 kHz in order to prove it isn't necessary. Otherwise, any shootout against another converter manufacturer running 192K is not equal variables. The only "fair" test would be two Lavry A/D converters running different rates. Chesky wants to do a test on location and then evaluate it back at the studio. The A/D test will have one model of converter (you can guess which) running at  96K and another model at 192K. Parallel analog feeds on location will go to different converters running at different rates. It's just fraught with potential errors of judgment, since different brands of A/D will be used on the record side. But it will be interesting to see what the listening test shows. BK

You said: "Ironically, you have to support 192 kHz in order to prove it isn't necessary."

Where is the logic here? Do I have to sell poison pills in order to prove that poison is bad for you? Do I need to support 100MHz sample rate in order to prove that such a notion is a crock for sampling audio? Please explain why I, a proffesional engineer, have to support something bad, in my field, in order to be able to state it is bad.

For your info, I did design a 192KHz DA, and it took a year to do so, but it was more then sufficient to tell me that 192KHz is BS. I got to see first hand that the tradeoffs where bad, and that I was operating away from the optimum rate. Only then, did I get to writing my paper about sampling rate.

Also for your info, my new Lavry Black DA does operate at 192KHz in all 3 modes - Narrow, CrystalLock-TM and wide (SRC). But I do not sell it as a 192KHz device, because 192KHz sampling IS MARKETINMG DRIVEN, not ear driven, nor is it engineering driven. Anyone with inside knowledge knows that.
I do not want to participate in taking audio down the wrong path, just to make some few extra bucks. History will tell who was telling the truth!

History will tell who was steering the ship into shore! It is unfortunate and very telling that people with great ears (such as Chesky) are finally going to some listening test regarding 192KHz, SO MANY YEARS AFTER the introduction of it. An engineering driven, or ear driven decision to go to different rate would be done differently - one would deal with theory, measuring the ear, prototype converters in lab environment, and one would move to a different sample rate ONLY AFTER something is established and proven. What happen with 192 is the OPOSITE! The semiconductor AD makers "got an order" for large enough quantity of 192KHz BEFORE and WITHOUT any credible reason to do so, and AGAINST the engineering and science.        

I have concentrated on the technical side of the issue, explaining why 192KHz is not needed, and why 60KHz is better. I concentrate on the technical because the "insiders" that know that 192KHZ was marketing driven concept, are afraid to go public, for fear of consequences (Career limiting, being laid off and so on).  

Of course there are a lot of potentials for error in listening test, and as a rule I do not talk about listening tests results on this forum. That is because I do not wish this forum to become a tool for commercially driven statements, as well as a place where subjective statements, not reflecting serious listening tests are stated.

But I have no reason to object statements regarding the proper PROCEDURE that makes up a good listening test. I certainly do not need to tell you what you already know - about the methodology of repeatable double blind listening tests. I do not mind people talking about such procedures.

My question to you is "procedural": Why should a double blind ABX test comparing some source material to say the same sound via 2 converters running at different rates, be any more difficult then comparing the results with the 2 units running at the same rate?

Yes, of course listening tests for say AD depend on other factors ranging from the DA, to the amp, speaker and so on. Yes, one does the best they can, holding all other variables constant, and even then there are potential errors. That is why repeatable tests at different locations are a good idea.

But "fundamentally speaking" (assuming that other factors do not come into play), using "the same" gear other then the units under test, one should be able to compare to the source material, and see which unit is more transparent.

Am I missing something here?

Regards
Dan Lavry
www.lavryengineering.com  

[C-J]

danlavry wrote on Thu, 08 December 2005 03:01

I have concentrated on the technical side of the issue, explaining why 192KHz is not needed, and why 60KHz is better.

Dan,

You have probably explained this many times, but what is the mathematical advantage of 60kHz SR over 44k1Hz?

C.J., Finland

[bobkatz]

danlavry wrote on Wed, 07 December 2005 20:01

BK

You said: "Ironically, you have to support 192 kHz in order to prove it isn't necessary."

Where is the logic here? Do I have to sell poison pills in order to prove that poison is bad for you? Do I need to support 100MHz sample rate in order to prove that such a

[/quote]

What I meant, Dan, is that if you believe that ultimately a listening test is required, then you have to have the supporting hardware to test it. I do not see much in your written arguments about why 192 is not necessary regarding issues of the effects of the low pass filters on the ear. The ONLY way to test that is to test (including listening) different low pass filters at 96 and 192K sample rates.

That's what I mean by the irony. In order to prove (by listening tests) that you don't need 192, you have to have a converter that does 192!

I agree that 192 is market-driven, hey it must sound twice as good as 96, right?  A powerful argument for people who listen with their eyes.

Yes, I know that the Lavry black does 192. But it isn't a Gold, is it?  .

BK

[danlavry]

bobkatz wrote on Thu, 08 December 2005 17:20

danlavry wrote on Wed, 07 December 2005 20:01 BK You said: "Ironically, you have to support 192 kHz in order to prove it isn't necessary." Where is the logic here? Do I have to sell poison pills in order to prove that poison is bad for you? Do I need to support 100MHz sample rate in order to prove that such a

What I meant, Dan, is that if you believe that ultimately a listening test is required, then you have to have the supporting hardware to test it. I do not see much in your written arguments about why 192 is not necessary regarding issues of the effects of the low pass filters on the ear. The ONLY way to test that is to test (including listening) different low pass filters at 96 and 192K sample rates.

That's what I mean by the irony. In order to prove (by listening tests) that you don't need 192, you have to have a converter that does 192!

I agree that 192 is market-driven, hey it must sound twice as good as 96, right?  A powerful argument for people who listen with their eyes.

Yes, I know that the Lavry black does 192. But it isn't a Gold, is it?  .

BK[/quote]

First, let me clarify something. The LavryBlack does 192, because most of the newer digital audio receivers and DA IC's do 192K. I did not see any reason to add circuits just for disabling it, but look at the LavryBlack front panel, the unit is designed to indicates that it receives 44.1, 48, 88.2 and 96KHz. I did not include the 2 more lamps that would indicate 176.4 and 192KHz, because doing so would support the 192KHz sampling, which by the way sounds worse and measures worse!
Imagine that, for 2 more lamps I could join the 192KHz hype, "expanding my sales"... What would I use for an excuse? Someone told me: "If I don't do it some else will". I need to respect myself, so I can not do it...

Regarding listening to low pass filters:
At what point does one stop getting the ear into the equation? How about 10MHz filter? Or 100MHz filter?
What would happen if you listened to say 100MHz filter, against a 1 inch of wire? Guess what - an opamp capable of 100MHz bandwidth would sound much worse then one aimed at say good 50KHz operation. What conclusion would that bring? That 100MHz is not wide enough?

Of course 100MHz is way off, but I used that example to demonstrate a point. The point is very specific: one hears what the 100MHz does to just a "sliver" of its bandwidth, the audible frequencies. Therefore, before you conclude that what you hear with a filter is due to increased bandwidth, you must be sure that the opamp used does not have a sound, as well as the caps and all, in a very similar circuit, power supply and "environment".

The proper conclustion is: There is an OPTIMAL POINT. More bandwidth and faster sampling is not better. Too little is not good, too much is not either. So where is the optimum? Not at hearing 96KHz audio (192KHz rate). We both know that.  

Having said that, let me add the important facts:

1. The analog filters in modern AD's are made with few poles, in the MHZ range, because the AD modulators work at over sampled speeds.

2. The analog filters in modern DA's can operate very fast as well because the DA's are upsampled, thus the image energy is at higher frequencies.

In fact, most analog filters operate very high, and the margins due to AD oversampling and DA upsampling are so large that most gear makers use the SAME ANALOG FILTERS FOR 44.1KHz-192KHz (both for AD and DA).

All that I staed above mean that we do not need to increase the sample rate to "help filter design". In fact, after I shot the argument "more dots is better", the filter argument came in and I shot it down as well, as I explained above. There were other arguments to shoot down, and at the end there is no single credible argument I saw to support 192KHz.

Again, good honest engineering, science (including listening) would call for a "systematic" approach, beginning with what is the hearing range of the ear (you can add to it "feeling sound with other body parts", I certainly do not mind including say 10Hz if we can precive floor vibrations and so on). To establish what a person can perceive, one does not have to move a whole industry to sample by X2 from where it was. In fact, one should test such theories in the analog domain, eliminating the digital audio variables all together. I would start with mics, micpre, power amp and speakers, all capable of the desired bandwidth to test. Instead of tape, or memory, I would have a sound proof room right next to the sound source (mics near the sound source, speaker at sound proof room....

Perhaps there are other better ways to test the ear. My point is - first they went to 192KHz sampling, with mics and speakers at around 20KHz, then they claim benefits, with no suporting data, no credible theory....

Regards
Dan Lavry
www.lavryengineeering.com




     

[Ashermusic]

Thank you, once again Dan, for sound science and common sense.