Ayre wants $1.5K for DSD'ed QB-9

[Fokus]

This question is the perfect example to understand the difference between a "digital engineer" and an "analog engineer".

 

Reading this far I saw a challenge. I went over to the Stereophile measurements page, read that one, and only then continued Charles' explanation.

 

It seems that I qualify as an analogue engineer.

 

Good.

 

But I am also a digital one ;-)

[Charles Hansen]

Charles If that is how you feel about DSD then why are you upgrading a DAC to support it ? just wondering. But of course incoming revenue comes to mind.

 

Hello Mav,

 

You are free to make up your own mind. It is clear that you already have. You have decided that I have cynically added DSD support just to make more money.

 

The truth is that (as I have also said before) DSD is very much like genital warts. Annoying and hard to get rid of, but in the big picture, not very serious of a problem.

 

For the first time, if somebody purchase a QA-9 Pro and a QB-9-DSD, they can hear quad-rate PCM and DSD implemented at the highest possible level. Then they can compare for themselves and see what sounds better. ALL variables are eliminated and it is the first time in history that a truly valid comparison between the formats can be made.

 

Once people do this, and they find out for themselves the truth -- that there is no inherent advantage to DSD, then maybe we can stop being distracted by this format that has so many practical disadvantages. Once we settle on a single high-res format, it will be much easier to create a group of buyers that is large enough to be meaningful. Then instead of listening to obscure recordings of obscure musical groups, we might reach a critical mass and have a good choice of good high-res software.

 

Cheers,

[Miska]

By the way, your modulator diagram doesn't contain proper dither, hope it's still there.

 

This is so much more complicated that it required the entire resources of the latest generation Xilinx Spartan 6 FPGA that we use. There was no room for anything else. So when you flip the switch to "DSD", it actually has to reprogram the FPGA and load a new program where over 90% of the processing power is dedicated to this extremely complex algorithm. Then when you flip back to PCM, it clears out that big mass of processing so that it has enough room to implement no fewer than SIX different low pass PCM filters. And even with six different filters, there is plenty of room left over for other functions. But the DSD algorithm uses all the available power of the FPGA...

 

This tells me that you are doing something seriously wrong. My computer is not even breaking a sweat while running dithered seventh order modulator for DSD in real time. If you look at the modulator, it is actually a fairly simple filter. However, doing conversion from DSD or multi-bit SDM to PCM properly takes a lot of resources.

 

Conversion to PCM requires only one thing -- a low pass digital filter. That's it. Nothing else.

 

OK, you are doing it wrong. Sorry.

 

Sounds like you are repeating the same mistakes chip makers are doing in their implementations that makes the performance suffer.

 

And not only is this conversion much more complex than the conversion to PCM, it also adds significant amounts of OOB energy! So it's no a loss, but it is a degradation of the signal!

 

This tells me that you are an analog engineer, because you try to compare a Nyquist sampling system to non-Nyquist sampling system.

[Charles Hansen]

For high quality conversion I assume all IMD and harmonic products being at lest below -110 dB, preferably below -120 dB.

 

Higher order IMD products around 19 and 20 tones at 18 and 21 are at -80 dB level which is quite high. These are typical problems for many converters.]

 

Of course some people like sound spiced with harmonic and intermodulation distortion. To me, the performance of QA-9 is similar to a typical tube amp.

 

Hello Miska,

 

Ah, yes! The "digital engineer" emerges. Here is the one that thinks that everything can be explained with numbers. All we need is "better" numbers and then everything will be perfect....

 

Do you really believe that something with 0.01% distortion (-80 dB) is going to ruin the sound, but if we reduce the distortion to 0.0001% (-120 dB) THEN we will have "perfect" sound.

 

I hope you are enjoying l listening to your Halcro amplifiers, as they are perhaps the only ones in history that have reached such low distortion levels. Unfortunately, they didn't sound good enough that the company was able to remain in business....

 

This tells me that you are doing something seriously wrong. My computer is not even breaking a sweat while running dithered seventh order modulator for DSD in real time.

 

Yes Miska, I am sure that your Intel I7 Quad-Core processor is MUCH more powerful than the FPGA we use in the QA-9. But not everybody wishes to have a PC in their living room to be a part of their sound system. And for people that do, there are many companies that sell computers, so there is no problem. Perhaps they will even purchase your software!

 

OK, you are doing it wrong. Sorry.

 

Oh, right.....

 

I forgot. We MUST be doing it wrong! And when John Atkinson reviews the unit and says, "there was no doubt that with a 192kHz sample rate I could not distinguish between the LP and the digital rip. And believe me, I tried. I A/B'd the two versions until blood came out of my ears and I was heartily sick of this music I hadn't heard for, in some cases, decades." there is the proof we are doing it "wrong"....

 

Because after all, you are a digital engineer, and poor me. Just like my signature says...

[Hiro]

Freann,

 

Hiro is referring to the fact that some LP cutting lathes have a signal that is delayed by conversion to digital before being sent to the lathe. The lathe can adjust the width between the grooves so that loud passages have more room available, while quiet passages don't need them.

 

But this is far from a universal practice. When the original master is digital it makes a lot of sense. But when the original master was made by going to all of the trouble to use a full analog signal path, no mastering engineer is stupid enough to do that. Instead there is an extra playback head installed on the tape recorder called a "preview head". The time delay for the tape to travel from the preview head to the main playback head accomplishes the same thing without a conversion to digital.

 

All high quality albums made today use the analog preview head, regardless of what propaganda Hiro is trying to spoon-feed you...

 

Best,

 

Charles,

 

No amount of confusing rhetoric, and convincing people that white is black and black is white will help here.

 

I'm aware that there are still one or two labels in the world producing LPs using all-analog chain, just as there are probably one or two companies using ladder-type PCM converters. But this doesn't make my valid point "propaganda. On the contrary, my point point is still valid. Digitally sourced or mastered Analog records are not analog. And of course, SDM is not PCM.

[PhilipAC]

Hi Folks

 

Dumb question- all this talk of the QA9 (Pro or otherwise)- I have always assumed that I would only need one to convert vinyl to a digital file. If I have no vinyl (or presumably analog tape), then I assume I have no use for a QA9 and all I need is the QB9 or another DAC.

 

Is this correct?

[Hiro]

I wish you great success with your QA-9. It has all the elements of a great A/D converter, particularly with your expertise in front end analog design. For prosumer customers that are searching for the highest quality for transcribing their analog sources, my hope is they consider the Pro version with its DSD output.

 

Is the QA-9 A/D converter DSD256 capable as advertised in Ayre's materials, or is it all talk?

[Hiro]

I forgot. We MUST be doing it wrong! And when John Atkinson reviews the unit and says, "there was no doubt that with a 192kHz sample rate I could not distinguish between the LP and the digital rip. And believe me, I tried. I A/B'd the two versions until blood came out of my ears and I was heartily sick of this music I hadn't heard for, in some cases, decades." there is the proof we are doing it "wrong"....

 

Whatever the outcome of his little vinyl experiment, John Atkinson isn't denying DSD merits.

Cookie Marenco dems DSD | Stereophile.com

 

Someone should probably send him the recent Boston Baroque recording captured by Five/Four in DSD256 so that he can decide for himself which format, DSD or vinyl, is the closer representation of a live event.

[Miska]

Ah, yes! The "digital engineer" emerges. Here is the one that thinks that everything can be explained with numbers. All we need is "better" numbers and then everything will be perfect....

 

Do you really believe that something with 0.01% distortion (-80 dB) is going to ruin the sound, but if we reduce the distortion to 0.0001% (-120 dB) THEN we will have "perfect" sound.

 

No, but I require both good measurement results and great sound, since THD adds up I can accept highest levels in power amps, but really minimal in pre-amp stages and even smaller in ADCs.

 

My preamps are my own design and build.

 

I have four power amps of my own build, two which I designed myself, one that was designed by Douglas Self and one designed by John Linsley Hood. Best sounding is still a one of my own MOS-FET designs that has frequency response flat to 1 MHz. However the current one is based on mod'ed Mark Levinson stages.

 

Generally when I buy, I pre-filter the equipment first by measurement results and then find the good sounding ones from the ones that measure well.

 

For amps, I have found following to be good ones:

Accuphase

Musical Fidelity

NAD M3

 

Yes Miska, I am sure that your Intel I7 Quad-Core processor is MUCH more powerful than the FPGA we use in the QA-9. But not everybody wishes to have a PC in their living room to be a part of their sound system. And for people that do, there are many companies that sell computers, so there is no problem. Perhaps they will even purchase your software!

 

You can have it in a tablet now:

Convertible Ultrabook Laptop & Tablet PC | ThinkPad Helix | 11.6" with Windows 8 (US)

 

But anyway, I'm running my PC in a separate room and the processed audio is streamed to the DAC over ethernet. DAC has just a small software module running on ARM CPU.

 

I forgot. We MUST be doing it wrong! And when John Atkinson reviews the unit and says

 

Yeah, and of course all the others who think DSD sounds superior are wrong.

[PeterSt]

Wow, I have been reading this thread from of post 296 with open mouth.

And no, I should have no voice here, but I can't resist. :-)

 

Do you really believe that something with 0.01% distortion (-80 dB) is going to ruin the sound, but if we reduce the distortion to 0.0001% (-120 dB) THEN we will have "perfect" sound.

 

YES. Well, sort of. Since I can easily hear the deteriorated difference between 0.0018% and 0.0024% (for Redbook playback), yes. And I am quite sure I am not the only one, if only the nice PCM filtering isn't in the way (which it is not in my case).

To say - that when all is down to 0.0001% - all is "perfect" is another story (undoubtedly still not).

Anyway notice that with non-ringing filters the THD you look at is the THD you listen to and test signals *can* represent reality when listening.

 

Oh, right.....

 

I forgot. We MUST be doing it wrong! And when John Atkinson reviews the unit and says, "there was no doubt that with a 192kHz sample rate I could not distinguish between the LP and the digital rip. And believe me, I tried. I A/B'd the two versions until blood came out of my ears and I was heartily sick of this music I hadn't heard for, in some cases, decades." there is the proof we are doing it "wrong"....

 

It is not difficult at all to have your own LP rip sound the same as the LP itself. Just use a better ADC (allowed to be yours haha), give it the sufficient headroom for digital and stay away from any processing/enhancing sh*t.

Maybe I should add that I tend to hear the difference in exactly everything (let's say it is my job), but an LP rip is easy to have it sound equal.

If this comes across as pedantic or ignorant perhaps ... it probably is (but not exactly meant like that).

 

Because after all, you are a digital engineer, and poor me. Just like my signature says...

 

Of course all is related to your 0.01% THD etc. story, but if we are allowed to assume that this is very much audible, the prerequisite really is good figures in the digital domain first. I don't see it go the other way around (though sometimes we hope that analog is too slow to pass on HF sh*t - which is also the wrong approach).

 

Regards,

Peter

[tailspn]

I can see why you would say this. But this is the perfect example of "knowing just enough to get you into trouble".

 

Conversion to PCM requires only one thing -- a low pass digital filter. That's it. Nothing else.

 

On the other hand, to convert to DSD is FAR, FAR more complex!

 

Well, again we have to agree to disagree. But you will notice I have not disparaged, or assumed your knowledge level in any way.

 

You're cherry-picking your examples. If you investigate further you'll find the conversion, if done correctly, from the 6-bit 11.29MHz stream native to the AT1202's Delta-Sigma Modulator is hardly "a low pass filter", and you're being disingenuous telling the people reading this thread so. The modulators required for both PCM and slower sample rate DSD conversion are the same in structure, but very different in action.

 

The 6-bit 11.29MHz bit stream to DSD (1-bit, two level 2.8224MHz) is a 4X decimation and summing to 1-bit, as your figure 2 shows. The conversion to 192KHz PCM is not only a decimation of greater than 58X, but also the math processing to convert the 6-bit stream into a 2's complement binary word structure. Added to that, since 192KHz is not a integer divisible sampling rate from 11.29MHz, like 176.4KHz would be, there's the additional math processing, and round-off errors to accomplish that. It's hardly a low pass filter.

 

More importantly, if you're able to investigate the phase response of a decimation filter necessary to filter the frequency products above the Nyquist frequency of 192KHz PCM, for a conversion from a 6-bit PWM 11.29MHz bit stream, compared to the phase response of the DSD modulator, you would find a very different impulse response. This you can hear, and favors the DSD reconstruction in spaciousness and instrument details.

 

Tom

[mav52]

Hello Mav,

 

You are free to make up your own mind. It is clear that you already have. You have decided that I have cynically added DSD support just to make more money.

 

The truth is that (as I have also said before) DSD is very much like genital warts. Annoying and hard to get rid of, but in the big picture, not very serious of a problem.

 

For the first time, if somebody purchase a QA-9 Pro and a QB-9-DSD, they can hear quad-rate PCM and DSD implemented at the highest possible level. Then they can compare for themselves and see what sounds better. ALL variables are eliminated and it is the first time in history that a truly valid comparison between the formats can be made.

 

Once people do this, and they find out for themselves the truth -- that there is no inherent advantage to DSD, then maybe we can stop being distracted by this format that has so many practical disadvantages. Once we settle on a single high-res format, it will be much easier to create a group of buyers that is large enough to be meaningful. Then instead of listening to obscure recordings of obscure musical groups, we might reach a critical mass and have a good choice of good high-res software.

 

Cheers,

 

Ok, I get where your coming from...

 

Since I'm looking at your DAC, do you know when the dealers will be getting this new dac... My closest dealer for your product is going to be Audio Advice in Charlotte NC.

[Hiro]

For the first time, if somebody purchase a QA-9 Pro and a QB-9-DSD, they can hear quad-rate PCM and DSD implemented at the highest possible level.

 

You mean no other company in the world can do it better? Have you compared the QA-9's performance at 2.82MHz (64x DSD) with the Grimm AD1 converter for example?

 

And lastly, can somebody who purchases QA-9 Pro and QB-9 DSD even make the evaluation of 128x and 256x DSD captures done with the QA-9 using the matching QB-9? Have you solved that problem yet?

[elcorso]

Hmmm, I guess we are talking about semantics then:

 

"Originally Posted by Charles Hansen

When 192/24 is done PROPERLY, it sounds better than DSD."

 

When someone tells me something "sounds better", I equate that statement with an analysis based on listening, not a technical theoretical discussion such as you gave. But, even your technical explanation has flaws: the noise inherent in DSD is a technical flaw, for example, which you appear to dismiss.

In any case, I have nothing against DSD per se, but there is nothing in my experience to suggest that DSD "sounds better" than well done PCM at 24/192. I find good examples for analysis can be had from Channel Classics: I suggest that readers here might make their own comparisons. Channel Classics is now using the highly regarded Grimm Audio ADC for DSD recording, and sells both DSD downloads and PCM downloads (converted using Weiss Saracon). Purchase a DSD file from them (I would recommend one of the recent recordings done with the Grimm), and compare vs a 24/192 PCM conversion. Or, use the Korg SRC to make some PCM conversions of your own to 24/176.4, try the soft filter. And listen in your own system.

Some DACs may produce better sound on DSD, some with PCM, but if people make their own comparisons, they will have a better understanding of how things sound in their own systems.

I am not sure why anyone would want to start these types of compaprisons with a PCM source (the confusingly named "DXD") although, I suspect that "DXD" when done well is transparent anyway... but for the sake of keeping things fair for DSD, I would prefer a native DSD file be used for comparisons.

 

Bold is mine.

 

Hi Barrows,

 

Just for the records: There is not such a thing like "the noise inherent in DSD", if you don't believe me, ask some very good recording engineers...

 

The problem, as always in recordings, DSD or PCM and even analogue is the recording/mixing/editing engineer.

 

If you drive a Ferrari (I never had the chance) and the RPM meter shows red at 8,000 rpm, but you think, this is a Ferrari I will go to 10,000 rpm, and the engine crash. Who is the guilty one, you, or Ferrari engineers?

 

There are extraordinary Hi Res PCM recordings, and a lot of bad ones (HDTracks is full of it). But I can't blame HDTracks, or PCM for this, but the recording/mixing/editing engineer.

 

There are extraordinary Hi Res DSD recordings, and a lot of bad ones (from PS3 rip). But I can't blame DSD, but the recording/mixing/editing engineer.

 

"The noise inherent in DSD is a technical flaw" is to me a myth, as could be "the noise inherent to PCM is a technical flaw…"

 

I can't go to technical explanations, this is not my speciality (but Bonsai, ha, ha, ha...). For this we have Miska, with his full credentials as Digital Engineer. Also Charles with his lovely SQ music gear. But please not confuse the premises: One is market potential and other thing some 'inherency' that doesn't exist.

 

BTW, this weekend I listened to the two DSD albums DL from Opus 3 (analogue to DSD 128) at US$25.-/both, and Blue Coast Records DL Keith Greeninger "Special Event 20" at US$15 in DSD64 and free DSD128, DSD128 recording. I'm are very, very impressed with SQ, like never in my life. The listening was with the expensive (?) Playback Designs DAC, and with the US$2,500 exaSOUND DAC, both at DSD128, both with glorious sound.

 

Of course, as I said before, "one swallow does not make a summer", but I do love this 3 first albums (swallows)!

 

Best,

 

Roch

[Charles Hansen]

Hi Folks

 

Dumb question- all this talk of the QA9 (Pro or otherwise)- I have always assumed that I would only need one to convert vinyl to a digital file. If I have no vinyl (or presumably analog tape), then I assume I have no use for a QA9 and all I need is the QB9 or another DAC.

 

Is this correct?

 

Hello Philip,

 

Yes, you are correct. The QA-9 converts analog signals to digital. So if you have cassettes or LP's that you wish to convert to high quality digital, this is a great tool. BUT it is very expensive, so unless you need the very best available, you don't need this. Almost every laptop has analog inputs and a (relatively low quality, but still pretty good) A/D converter built in. If you want to digitize an LP, you will also need a phono preamplifier to boost and equalize the signal. Some companies build OK ones like this with a USB output for under $200. So it all depends on the quality level you are looking for.

 

But for playing back digital files, this type of tool is not needed at all.

 

Cheers,

[Charles Hansen]

Is the QA-9 A/D converter DSD256 capable as advertised in Ayre's materials, or is it all talk?

 

Dear Hiro,

 

For you I will make a special exception. The Ayre QA-9 is capable of DSD-16,384. No, really! And 128 channels! And room EQ that is perfect at every point in the room. And all distortion products are more than -200 dB down. And it even performs oral sex! Really! (It is an amazing bargain...)

 

Cheers,

[Charles Hansen]

Whatever the outcome of his little vinyl experiment, John Atkinson isn't denying DSD merits.

Cookie Marenco dems DSD | Stereophile.com

 

Someone should probably send him the recent Boston Baroque recording captured by Five/Four in DSD256 so that he can decide for himself which format, DSD or vinyl, is the closer representation of a live event.

 

Dear Hiro,

 

He may indeed be impressed by he DSD-256. But wait until he hears my 1000 rpm turntable!. To get enough playing time it is 10 meters in diameter. But the tonearm is so long that tracking error is no longer a problem. The sound is FANTASTIC! One time I was playing the prototype and the ghosts of the (now dead) performers emerged from my speakers. It was an AMAZING experience! I wish you could have been there...

 

Cheers,

[Charles Hansen]

You're cherry-picking your examples. If you investigate further you'll find the conversion, if done correctly, from the 6-bit 11.29MHz stream native to the AT1202's Delta-Sigma Modulator is hardly "a low pass filter", and you're being disingenuous telling the people reading this thread so. The modulators required for both PCM and slower sample rate DSD conversion are the same in structure, but very different in action.

 

The 6-bit 11.29MHz bit stream to DSD (1-bit, two level 2.8224MHz) is a 4X decimation and summing to 1-bit, as your figure 2 shows. The conversion to 192KHz PCM is not only a decimation of greater than 58X, but also the math processing to convert the 6-bit stream into a 2's complement binary word structure. Added to that, since 192KHz is not a integer divisible sampling rate from 11.29MHz, like 176.4KHz would be, there's the additional math processing, and round-off errors to accomplish that. It's hardly a low pass filter.

 

More importantly, if you're able to investigate the phase response of a decimation filter necessary to filter the frequency products above the Nyquist frequency of 192KHz PCM, for a conversion from a 6-bit PWM 11.29MHz bit stream, compared to the phase response of the DSD modulator, you would find a very different impulse response. This you can hear, and favors the DSD reconstruction in spaciousness and instrument details.

 

Tom

 

Dear Tom,

 

All I can say is that anyone who does the conversion the way you describe would be brain-dead. Why would you start off with a master oscillator that is not an integer multiple of the final sampling rate? All of our equipment has two separate crystal oscillators. One is used for sample rates that are multiples of 44.1 kHz. The other is used for sample rates that multiples of 48 kHz. I don't know of any company that doesn't do it this way, except for very cheap Chinese product that want to save $0.50 for a second crystal oscillator.

 

Cheers,

[Charles Hansen]

Hi Barrows,

 

Just for the records: There is not such a thing like "the noise inherent in DSD", if you don't believe me, ask some very good recording engineers...

 

 

Hello Corso,

 

No, actually Barrows is correct. There is a high level of noise in all DSD recordings. You cannot hear this noise, because it is above the range of normal human hearing. Nevertheless, this out-of-band (OOB) noise causes many problems, from overlosding of poorly designed playback equipment to problems with production of recordings (eg, overdubs, EQ, reverb, and other effects).

 

This is one huge reason why DSD is such a VERY small portion of the market. The other reason is that it costs a lot of money to replace all of your existing equipment to purchase new equipment, and there is basically no reason to do so.

 

Best regards,

[Charles Hansen]

Ok, I get where your coming from...

 

Since I'm looking at your DAC, do you know when the dealers will be getting this new dac... My closest dealer for your product is going to be Audio Advice in Charlotte NC.

 

Hello Mav,

 

Sorry, I don't know. I work mainly on the product designs. I would call either Alex at Ayre, 303-442-7300 x233 or Audio Advice. I know he did a demonstration with the QA-9 there a few months ago. He was playing back vinyl rips of some very well recorded jazz LP's. I think it is safe to say that anybody who was there (salesperson or customer) was shocked at the sound quality. I am sure that they will get a an update QB-9-DSD sometime this summer, but I do not know exactly when.

 

Cheers,

[Charles Hansen]

You mean no other company in the world can do it better? Have you compared the QA-9's performance at 2.82MHz (64x DSD) with the Grimm AD1 converter for example?

 

And lastly, can somebody who purchases QA-9 Pro and QB-9 DSD even make the evaluation of 128x and 256x DSD captures done with the QA-9 using the matching QB-9? Have you solved that problem yet?

 

Hello Hiro,

 

The Grimm is an excellent converter, perhaps the best on the market today for DSD. I would be glad to compare the Ayre QA-9 to the Grimm. To produce PCM, the Grimm requires an outboard box. The Ayre is more flexible and includes both PCM and DSD.

 

As you know the Ayre QB-9 D/A converter is currently limited to DSD-64, as is the Grimm. Unlike the Grimm, the Ayre is upgradeable. If there is ever a true market for DSD-128 (a minimum of 100 titles available), Ayre will probably offer an update. To the best of my knowledge the Grimm is not upgradeable. The literature I read stated that there were disadvantages to going beyond DSD-64. It did not specify what the disadvantages were, but I was left with the strong impression that they had no interest in going beyond DSD-64 whatsoever.

 

Regards,

[Hiro]

The literature I read stated that there were disadvantages to going beyond DSD-64. It did not specify what the disadvantages were, but I was left with the strong impression that they had no interest in going beyond DSD-64 whatsoever.

 

Watch this space.

[Hiro]

Dear Hiro,

 

He may indeed be impressed by he DSD-256. But wait until he hears my 1000 rpm turntable!.

 

Charles, is there any such thing on the horizon? No. And DSD-256 is already here. Ready for all record labels to use.

 

http://www.computeraudiophile.com/f15-music-general/worlds-highest-resolution-recording-16011/

[Hiro]

For you I will make a special exception. The Ayre QA-9 is capable of DSD-16,384.

 

You don't have to make any "special exceptions" for me.

 

Just make your two matching devices QA-9 (256fs DSD) ADC and QB-9 DAC compatible, because at the present moment they don't quite match.

[PhilipAC]

Hi Miska

 

I am interested in getting software to convert PCM to DSD. I believe your software does this, but only for Windows and Linux. Is there anything similar for Mac, do you know, from you or anyone else? If not, it is as simple as borrowing my daughter's PC, doing the conversions and then reloading on my Mac?

 

Philip