Jump to content
IGNORED

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


Recommended Posts

Yes, DST is like a zip file to get both stereo and multi-channel DFF files onto a SACD disc.

 

You're right Russ, I completely forgot about DST, since is buried in the SACD authoring process. Thanks for the correction!

 

Tom

Link to post
Share on other sites
I hope that you take more care with your products than you do with your posts. Figure 11 in the Stereophile article is a measurement of the dCS converter. You know the one that uses the 5-bit "Ring DAC"?

 

You would have to ask dCS about their designs. I know nothing about them.

 

Oh damn, sorry, I really shouldn't be posting at 3am... ;)

 

4x rate ADC PCM conversion noise floor shouldn't limited by SDM (even 1-bit) at 256x rates (12.3 MHz). IIRC, dCS was running at 64x?

 

Even 32-bit PCM is not enough to represent 256x DSD noise floor from 0-20 kHz.

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to post
Share on other sites

Hi Charles,

 

Apologies to my fellow CA board members, but it's rare to have this kind of access to the man who designs gear that I plan to audition, so here goes:

 

I read your earlier comments re power conditioners, and this is a topic very dear to my heart. I am in the (enviable ?) position of being able to start all over in early 2014 - from the wall socket out - but the downside is that it will be in an apartment block in one of the craziest cities on earth - Bangkok. 230V of mains power that seems to have been laid by epileptics on crack, but it has been reliable on each of my 2-3 month stays over the past ten years.

 

It would seem that I can base a system around the QB-9, Ax7-e and my laptop : with the possible exception of a pair of Ayre power cords, I'm hoping that I can avoid the need for expensive power ancillaries from the likes of PS Audio.

 

- am I being overly simplistic in my assumptions ? I have no idea what the limits of your passive conditioners are.

 

- I assume that you prefer the DAC to be plugged into a different outlet to the amp, and not via a powerboard ? The laptop will only be plugged in when I need to recharge it, and I wont have any other appliance running from that outlet, although there will be the usual maze of AV gear nearby, and I wont be turning off the fridge to listen to music :D

 

Thanks for your time,

 

Ned

Just one more headphone and I know I can kick this nasty little habit !

Link to post
Share on other sites
Oh damn, sorry, I really shouldn't be posting at 3am... ;)

 

That was my initial impression. But, since I was up until 5 AM (making cables), I thought it best to not pour any more gasoline on this.

 

Glad someone has a sense of humor about all of this.

 

Back to our regularly scheduled whatever it is.................

Link to post
Share on other sites
Charlie Hansen: Now I know that Ayre is a hardware company, and I understand that you guys might rather sell hardware than do this, but I wanted to plant a seed anyway. Have you considered developing SRC software using your relayively unique digital filter designs? I would think that Ayre SRC SW might be a nice thing to offer. I for one would love to be able to convert DSD-PCM at 176.4 and 352.8 using your filters, and there would likely be some professionals who might like a nother option for SRC SW. I could also see some computer audiophiles oversampling 44.1 files to, say 352.8 using your filters (there are some 352.8 DACs around these days) to take advantage of your filter designs.

 

Hello Barrows,

 

It's something that I've thought about but it's not a high priority. I really don't think there would be much demand for it. Let's take the example of a DSD-to-PCM converter.

 

First of all, I've never heard a DSD recording that blew me away. When it first came out we borrowed an SCD-1 and a few discs and took it to our listening room for a few days. It sounded good, but not great. There was a certain identifiable "sameness" to the high-frequency reproduction. It was hard to know what to attribute that to. In the early days ALL SACD's were made with the Sony-owned hardware that was loaned to the studios. So it could have been that. It also could have been the player.

 

The more I found out about it, the more obvious it was that the SCD-1 was a LONG way from a state-of-art design. Plus we naturally listened to it in the balanced mode and only found out many months later that the single-ended outputs sounded better. The DAC chips had balanced outputs, but that was then converted to single-ended. The balanced analog output was created by adding another (crappy, as they mostly are) op-amp in the signal path to create an inverted phase for the "balanced" output. (I'll never understand the mind of the Japanese engineers...)

 

My next experience was when we made our universal player. I thought, "Let's convert that DSD to PCM and get rid of all of that horrible out-of-band noise. So we made a converter with FPGA's. It sounded pretty good, but the native DSD sounded a tiny bit better, so we left it alone and played the native DSD signal. No point to put a lot of hardware in there if it didn't help the sound.

 

Of course, that was nearly a decade ago, and we know a LOT more about digital filters now than we did then. For this past year's CES I bought a bunch of high-res downloads from HD Tracks to show off our new upgraded players and DACs. A few of them were Rolling Stones albums that had been converted from DSD to PCM. I thought they sounded fairly ordinary and I was confident I could make a much better sounding converter than whatever they used on those. (At one time we owned a few of the original Stones DSD releases, but when they get carried around the world for use in demos, they tend to get lost. But although I had never been impressed by them, I didn't remember them sounding as ordinary as the PCM conversions did.)

 

So let's say we made a DSD-to-PCM converter that was completely transparent -- that trained listeners could hear no difference between the DSD file and the one converted to PCM. Who would buy such a thing? HD Tracks didn't show any interest and I offered to make them one for free! Anybody that is a "true believer" in DSD wouldn't touch it with a ten-foot pole. So all that leaves is a few people that have DACs that they absolutely love but aren't DSD-capable. But where in the heck are they going to get their source material? Most of the good stuff has been out of print for over a decade. So I think it's pretty much a dead-end.

 

~~~~~~~~~

 

Regarding an external "upsampler" box, I can't see much of a market for that either. I think that the first digital filter in a string has the largest influence on the overall sound. So it is certainly possible that someone could improve the sound of their DAC by purchasing an Ayre-designed "upsampler". But I don't like putting two digital filters in a row, just on basic principle. One of the problems with digital filters is that 99.9$ of them are concatenated 2x filters, as that is the cheapest way to do it. But every filter you add compounds the rounding errors that are inevitable. So we do all of our filters in one pass.

 

And if someone really thinks that we make the best sounding digital filters would really be better off buying an entire Ayre DAC, instead of creating some sort of "Frankenstein" that only gives you part of the benefits. I think the bottom line is that I tend to look at things from a different perspective than a lot of people. There were a bunch of people who wouldn't buy a QB-9 because it only had one input. That was a real head scratcher for me. I would ask people what other input they would want and what they would use it for.

 

The most common thing was they wanted a TosLink or S/PDIF so that they could connect therir DVD player. I was like, "Why not just watch movies from your computer?" And they would have some weird reason that never made sense to me. So instead of people going, "Wow, this is great! I can get $10,000 sound for under $3000 by leaving out a bunch of features that have limited performance and limited utility!", they would rather that we added a couple of relatively bad sounding inputs with inherently flawed designs and paying twice as much for the unit.

 

Same thing with our phono stage. I'll put it up against anything under $10,000. But people have one of two attitudes:

 

a) It only cost $2500, so it can't be any good.

 

b) It doesn't have remote control loading (that only gets used once every few years when they change cartridges) or some other similarly useless whiz--bang gizmo and they feel better paying twice as much for competing product loaded with features of questionable value even if it doesn't even sound as good.

 

Oh well. Maybe someday it will all make sense to me....

 

Cheers,

Charles Hansen

Dumb Analog Hardware Engineer
Former Transducer Designer

Link to post
Share on other sites

Thanks Charles. Right now I am upgrading an XMOS based USB-I2S interface I have here with new firmware for DoP. If this works I will be able to compare DSD vs DAD to PCM conversions I have done with Korg's Audiogate (to 176.4, using their "soft" filter option and "aqua dither"). It will be interesting for me to listen to the results. I have some SACD-DSD rips here, some decently done titles like Wish You Were Here, and some of the RCA Living Stereo stuff, and Mercury Living Presence. I also have some DSD downloads from Channel Classics, including DSD masters made with the Grimm.

 

BTW, I support your choice of a single input DAC, I even feel there are likely performance advantages to the single input. I would rather spend $ on great sound than mostly useless features. I have a great deal of respect for how your company operates, and the decisions you make. Please, continue to plot your own course, and not let consumer's "wants" influence you unduly to compromise your designs.

SO/ROON/HQPe: DSD 256-Sonore opticalModuleDeluxe-Signature Rendu optical--Bricasti M3 DAC--DIY Purifi Amplifier-Focus Audio FS888-JL E 112 sub-Nordost Tyr USB, DIY EventHorizon AC cables, Iconoclast XLR & speaker cables, Synergistic Orange Fuses, Spacetime system clarifiers.                                                       

                                                                                           SONORE computer audio

Link to post
Share on other sites
Even 32-bit PCM is not enough to represent 256x DSD noise floor from 0-20 kHz.

 

Everybody --

 

This is getting completely out of hand. This is such blatant misinformation that I cannot allow this to stand uncorrected. If you want to post complete nonsense like this, please put it into some other thread. This thread is about the upgrade to the Ayre USB D/A converters. I will not tolerate any more of this misinformation in this thread.

 

First of all, a 32-bit PCM signal would theoretically have a wideband S/N ratio of 194.4 dB.

 

Of course this is useless to know as it is completely meaningless. No analog electronics will have a dynamic range anywhere close to this. The human ear is incapable of hearing this. In fact, if you were exposed to a signal of 194 dB SPL, you would be instantly killed -- literally. Even below this SPL, the air pressure would be going from 2 atmospheres to a total vacuum thousands of times per second. So to even talk about things like this are completely stupid.

 

Second of all there is a great deal of confusion about the difference between a wideband S/N ratio and what the reading on a spectrum analyzer shows. Let's use the misleading graph shown in the Positive Feedback Online article as an example:

 

dsd noise.jpg

 

In this completely misleading graph, they are supposed to be comparing the dynamic range levels in 3 sample rates of PCM (96/24 = red, 192/24 = orange, 384/24 = green) against the noise level of DSD (blue).

 

There are many, many things wrong with this graph:

 

1) The graph started out as showing the theoretical noise level of a DSD modulator as calculated in a computer program (almost certainly MatLab). In this graph a higher level of the line means a higher level of noise. Then some made-up numbers representing a theoretically perfect PCM system (with a number of errors -- some relatively minor and others huge) was drawn on this noise graph. But the graph was re-named as the "theoretical dynamic range".

 

This is silly. If one were graphing the dynamic range of something, the graph would be turned upside-down, so that a higher level of the line would represent a higher dynamic range. But here the higher level of the line represents the lower dynamic range. So from the very beginning this graph is difficult to interpret and therefore misleading.

 

All three of the PCM graphs show the dynamic range of the PCM rising to 0 dB above the Nyquist frequency (Fs/2). While this is true, the way the graph is drawn it leaves the reader with the impression that the noise is rising. The noise does not rise. Instead, the signal and the noise both fall to zero. There is no out-of-band energy (signal or noise) to disrupt the operation of the amplifier or damage the tweeters with PCM.

 

This is in direct contrast with the DSD graph, where the noise level does rise with frequency. The Scarlet Book specification for SACD mandates a 3rd order filter (-18 dB/octave) on all SACD players so that this out-of-band (OOB) noise does not damage the amplifiers and/or loudspeakers.

 

2) The theoretical signal plotted for the PCM systems is for the full record + playback chain. In contrast, the computer-generated level of noise from the DSD modulator shows only the record side -- just half of the overall equation.

 

3) The person who drew this graph does not understand the difference between the wideband S/N ratio and the noise floor of an FFT plot. When measuring the wideband S/N ratio, the system's noise floor is measured with a wideband voltmeter. This level is then compared to the maximum output level to calculate the S/N ratio.

 

But when the DUT (Device Under Test) is connected to a spectrum analyzer using FFT's to look at the spectral distribution of the noise, the apparent noise floor is lowered, typically by some 30 to 50 dB, depending on how many points were used to acquire the data required to calculate the Fourier transform. This phenomenon is known as "processing gain".

 

Briefly explained, the spectrum analyzer only "looks" at the noise contribution from a very small portion of the overall bandwidth of the DUT. As more points are used, each point measures a narrower and narrower portion of the spectrum and the noise contribution of this narrow portion is smaller and smaller. The formula for the processing gain is:

 

Processing Gain = 10 log (M/2), where M is the number of points used in the spectrum analysis.

 

With modern test equipment a common number of points used is 2^15 = 32,768, so the processing gain is -42.2 dB. The graph below summarizes this information. (It is an older graph when test equipment did not have so much memory and shows the processing gain for 4096 points.)

 

processing gain.png

 

The graph shows the wideband S/N ratio caused by the quantization noise of a 12-bit converter. The formula for this noise is 1.76 dB + (6 x N), where N = number of bits. This formula is for a converter without any noise shaping. Note that the noise floor is flat, as no noise shaping is utilized. This works out to 74 dB of wideband S/N ratio, but when measured with a 4096 point FFT, there is a processing gain of 33 dB.

 

To the uniformed reader it would appear that this converter has a S/N ratio or 74 + 33 =107 dB, but this is just nonsense. If a 12 bit system would give us 107 dB of S/N ratio, we could use that for recording high-quality digital audio!

 

This is the same point that I made back in post 199:

 

http://www.computeraudiophile.com/f6-dac-digital-analog-conversion/ayre-wants-%241-5k-dsded-qb-9-a-15650/index8.html#post227635

 

But apparently some people are having difficulty with this concept. If this does not make sense, please take some time to read up on this subject. There is an excellent short paper (7 pages) on this topic available from Analog Devices:

 

http://www.analog.com/static/imported-files/tutorials/MT-001.pdf

 

Please read this enough times until it makes sense to you.

 

Finally Miska states that a DSD-256 signal will have a higher signal-to-noise ratio in the audio band than a 32-bit PCM system. I say "Bullshit."

 

In the first place, there is no real system of any type in the world, DSD or PCM that will have a true S/N ratio of anything even close to 200 dB.

 

In the second place, I am sure that Miska has performed some silly computer simulation of the FFT noise floor of a hypothetical system. As anyone can see in the computer simulation of the noise floor of the standard DSD record - only side in the PFO article, the FFT noise floor in the audio band varies between -200 and -150 dB. But this includes an unspecified amount of processing gain due to the number of points used in the simulation. This is probably in the neighborhood of 40 to 50 dB more than the broadband S/N ratio.

 

If we used the same smoke and mirrors with a 32-bit PCM system we could (stupidly) say that it would have S/N of 250 dB! Of course this is not any sort of reflection of reality whatsoever.

 

Furthermore, we do not know what kind of noise shaping was used for Miska's simulation. One can apply higher and higher orders of noise shaping. This will lower the noise at the low frequencies,, at the expense of increasing the noise at the high frequencies. I have seen many curves for systems using DSD-128 and DSD-256. All of the ones that I have seen have the same noise floor in the audio band of approximately -120 dBFS. The only difference is that the frequency where the noise starts to rise rapidly shifts upwards one octave with each doubling of the sampling rate.

 

So with standard DSD, as Sony has defined it, using 7th order noise shaping and 64 Fs sampling, the noise starts to rise rapidly above 20 kHz. Then with DSD-128, the noise remains low up to 40 kHz before rising rapidly. And finally with DSD-256 the noise remains low up to 80 kHz before it rises. But I have never seen anyone try to reduce the noise in the audio band. This is because even with a sampling rate of 256 Fs, a 1-bit system is still very limited by the amount of noise present.

 

If one looks at the graph in the PFO article, it is easily seen that by 1 MHz that the FFT noise floor is only about -30 dBFS. Now we don't know how many points were used to generate this plot, but we can be fairly certain that it was enough to provide at least 40 dB of processing gain. So in real life, when we add this measurement artifact caused by measuring with an FFT, we can see that the true S/N of this system is actually about +10 dB!! That is to say, that on a broad-band basis, there is actually more noise than signal in this system.

 

No wonder the Scarlet Book mandates a -18 dB/octave filter in the player starting at 50 kHz. This means that by the time you get up to 500 kHz, when the FFT noise floor is only -25 dBFS or so and the true S/N would be about +15 dB more noise than signal, that the analog filter in the playback chain has reduced the noise level by -60 dB (plus whatever other rolloff exists in the analog circuit signal path).

 

~~~~~~~~~~

 

And with that, I think that we have had MORE than enough discussion about the technical shortcomings of the DSD format. The fact is that it can sound very, very good when implemented properly. And the specifications mandated by Sony luckily turned out to make it a better sounding system than the Redbook CD system.

 

And it is also fortunate that by digging deeper, and trying to understand why a system that measures so poorly can sound as good as it does has led us to the point where we can make huge improvements to the PCM format, particularly at the quad-sample rate. And since there will never be a mass exodus in the world to change over to a one bit system, and recognizing that for better or worse, we live in a world that is now and will always be (for the foreseeable future) dominated by PCM, it good to make PCM as good as we can.

 

When each of you gets a chance to hear how good properly implemented quad-rate PCM sounds, I am sure that you will be very pleased. I don't think you will feel shortchanged in any way that such good sound is available from PCM. In fact you should be rejoicing that it is so, as that will continue to be the mainstream format that 99.9% of the world's music will be available in.

 

And if you happen to have some DSD files, whether ripped from an SACD or downloaded from some small specialty audiophile label, the Ayre will play those as well. And as I mentioned before, if there ever gets to be even 100 titles available in DSD-128, we will offer on optional upgrade for the Ayre USB DAC's to play back these files. We certainly won't include it in all of the units as it is quite obvious that this will always remain (in the immortal words of Woody Allen in "Bananas") "a travesty of a mockery of a sham of a mockery of a travesty of two mockeries of a sham" of a format. But if there is enough source material to warrant it, we will make such a capability available.

 

Now let's quit arguing and go listen to some music, dammit.

Charles Hansen

Dumb Analog Hardware Engineer
Former Transducer Designer

Link to post
Share on other sites
Thanks Charles. Right now I am upgrading an XMOS based USB-I2S interface I have here with new firmware for DoP. If this works I will be able to compare DSD vs DAD to PCM conversions I have done with Korg's Audiogate (to 176.4, using their "soft" filter option and "aqua dither"). It will be interesting for me to listen to the results. I have some SACD-DSD rips here, some decently done titles like Wish You Were Here, and some of the RCA Living Stereo stuff, and Mercury Living Presence. I also have some DSD downloads from Channel Classics, including DSD masters made with the Grimm.

 

BTW, I support your choice of a single input DAC, I even feel there are likely performance advantages to the single input. I would rather spend $ on great sound than mostly useless features. I have a great deal of respect for how your company operates, and the decisions you make. Please, continue to plot your own course, and not let consumer's "wants" influence you unduly to compromise your designs.

 

Hello Barrows,

 

Thanks for the kind words. If you knew me, you would know that I am far too stubborn and hard-headed to do things in any other way than what I think is proper. And as with everyone, our greatest strength tends to also be our greatest weakness... I just design and build equipment that I would enjoy owning and hope that there are enough other people in the world that appreciate our approach so that we can remain in business. So far, so good... :-)

 

Good luck with your DoP upgrade. I'm not much of a programmer, but the main programmer at Ayre did the work on it himself (I think he was hoping Gordon would do it so that he wouldn't have to!) and he said it really was quite straight forward and didn't require too much brain damage....

 

And I must say that there are several SACD's that I wouldn't mind getting rips of. In all likelihood it is quite possible that no later release will be made of many of these albums, and the SACD's will represent the highest quality versions that are available. The only hope for something better to come along in the foreseeable future is Pono. I hope that succeeds. They have built quite a few features in there that everybody should be happy with it. It would satisfy the needs of the critical audiophiles while meeting the masses on the level that they are used to.

 

But perhaps most importantly, it will remove Apple from the equation. The problem today is that Apple is making more money on music sales than anybody else. This is very bad for the future of recorded music, as there isn't enough money left for the artist to make a living, the labels to find and develop new talent, and the studio to record the music in a truly professional fashion. Instead Apple gets the lion's share for reducing the total income, by selling individual tracks. While I can't blame people for just wanting to purchase the "good" songs on an album, that (more than anything else, IMO) is what has sharply reduced the overall revenue of the labels. But if we don't figure this out soon, there won't be any labels left to save...

 

Cheers,

Charles Hansen

Dumb Analog Hardware Engineer
Former Transducer Designer

Link to post
Share on other sites
4x rate ADC PCM conversion noise floor shouldn't limited by SDM (even 1-bit) at 256x rates (12.3 MHz). IIRC, dCS was running at 64x?

 

The dCS converters use a 5bit @ 64fs SDM front end indeed. Modern PCM converters aren't considered DSD-like for nothing.

Link to post
Share on other sites
Instead Apple gets the lion's share for reducing the total income, by selling individual tracks. While I can't blame people for just wanting to purchase the "good" songs on an album, that (more than anything else, IMO) is what has sharply reduced the overall revenue of the labels. But if we don't figure this out soon, there won't be any labels left to save...

 

You're free to blame Apple, but the greedy major record labels putting out half-assed albums with only 1 or 2 good songs are the ones who should take a lion's share of blame for this situation.

Link to post
Share on other sites
This is because even with a sampling rate of 256 Fs, a 1-bit system is still very limited by the amount of noise present.

 

Charles, with the usable bandwidth up to 100kHz, I would say, the 1-bit system is terribly limited.

 

BTW, could you give me a list of speakers capable of reproducing 120dB of DR and with frequency response going to 100kHz?

Link to post
Share on other sites

The most common thing was they wanted a TosLink or S/PDIF so that they could connect therir DVD player. I was like, "Why not just watch movies from your computer?" And they would have some weird reason that never made sense to me. So instead of people going, "Wow, this is great! I can get $10,000 sound for under $3000 by leaving out a bunch of features that have limited performance and limited utility!", they would rather that we added a couple of relatively bad sounding inputs with inherently flawed designs and paying twice as much for the unit.

Charles,

I have to disagree with you about above statement:the customers do count and your company has followed the audiophile demands/requests.

Let's take a look at your great DX5 unit.

It is DAC,cd,SACD and blu-ray player all in one.

The question is why if I can get QB9 dac for 2750$,c5 mp for 6000$ and oppo 105 for 1100$ total 9850$:cheaper than dx5.

Answer convenience of one vs 3 units.

Same applies to DACs.

Customers want to connect their computers,DVD players,satellite to DACs in order to have better sound.

In Addition your company released recently two products:

Ax5 and vx5.

Again one would buy Ax-5 for convenience of one unit rather than 2:vx-5 plus soon to be released preamplifier.

By adding spdif and possibly toslink Ayre will corner the DAC market-qb9 sounds great.

Link to post
Share on other sites
By adding spdif and possibly toslink Ayre will corner the DAC market-qb9 sounds great.

 

Hello Pawel,

 

Thank you for the kind words and the strong belief in Ayre. The problem with adding S/PDIF (or any derivatives, such as AES/EBU, or TosLink) is that it is a lose-lose situation.

 

This format is inherently flawed, and jitter is inherently added to the signal when using it. Once can spend more and more money to improve the system to any arbitrary degree, but no matter how much is spent, it will never equal the performance of async USB.

 

So we are faced with several choices, and I find none of them attractive:

 

1) Add a low-cost S/PDIF-based input(s). The performance of this would significantly lower than the USB input. In this situation I would worry that people would purchase the product and then be disappointed in the performance of those extra inputs.

 

2) The other extreme would be to spend a lot of money to improve the sound of the S/PDIF-based input(s). In this case the sound would be much closer to the sound of the USB input, but still not equal it. But taking this approach would easily double the price of the product (or even more!).

 

3) Choose something in-between, where the additional inputs have some performance improvements, but not so many as to increase the price quite so much. Would this solution be the best of both worlds, or the worst of both worlds?

 

I don't think there is a clear answer to this problem. I think that some customers would prefer solution 1, others solution 2, others solution 3, and still others the existing solution. The truth is that no one product (or company) can be all things to all people. To do that a product would have to look better than all competing products, sound better, have more features, and be sold at a lower cost. Clearly this is impossible. That is why it is good that there are many companies making many different products.

 

For example, we don't want to make DAC with a 9-bit, 25 MHz delta-sigma input. But for those customers that want one, apparently there is a company that will meet their needs. So then everybody gets what they want -- just not all from one company.

 

I think there are already too many products on the market with mediocre performance. There is no reason for Ayre to make another product with mediocre performance. So we only make a product when we believe that we can make something that is significantly better than what is currently offered by other manufacturers. To do so means we have to come up with a new idea that gives us a unique advantage over the existing products. It would be trivial to add a S/PDIF receiver chip to the QB-9 and have a low-quality digital input. But there are already hundreds of such products on the market. We will only release a product like that when we have a way to do it that performs better than existing products without costing an absurd amount of money. That is our philosophy. Some people like that philosophy and others not so much. For people that don't like our philosophy, there are many other choices in the market place and I am sure they will find something that meets their needs.

 

Best regards,

Charles Hansen

Dumb Analog Hardware Engineer
Former Transducer Designer

Link to post
Share on other sites
Charles, with the usable bandwidth up to 100kHz, I would say, the 1-bit system is terribly limited.

 

BTW, could you give me a list of speakers capable of reproducing 120dB of DR and with frequency response going to 100kHz?

 

Hiro,

 

There is no need to create a new version of a non-standard digital encoding system to reach that performance level. 192/24 PCM easily reaches (or exceeds) the levels of performance of DSD-256, but without adding the extreme amounts of out-of-band noise that all variations of DSD do.

 

Regards,

Charles Hansen

Dumb Analog Hardware Engineer
Former Transducer Designer

Link to post
Share on other sites
Hiro,

 

There is no need to create a new version of a non-standard digital encoding system to reach that performance level. 192/24 PCM easily reaches (or exceeds) the levels of performance of DSD-256

 

4fs PCM? Keep dreaming Charles...

 

Not even 8fs PCM (DXD) can touch DSD256fs.

Link to post
Share on other sites

Charles,

 

I talk about my DSD modulators, performed in digital domain with data sourced from a 64-bit floating point signal generator. I specified 0 - 20 kHz bandwidth. If I convert it back to 32-bit PCM of said bandwidth the dynamic range is limited by the 32-bit presentation, rather than the oroginal DSD data. So I either get 32-bit quantization error or dither visible.

 

There's nothing analog in entire chain. So performace of real world DSD systems is certainly not limited by the digital domain presentation.

 

My current seventh order SDM is optimized for 0 - 20 kHz performance at 6.1 MHz. I have also internal use variant optimized for 12.3 MHz rate. I don't know nor care how others design or optimize their modulators. This modulator id naturally mostly used to play back RedBook and hires PCM content to avoid using DAC chips's modulator.

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to post
Share on other sites

BTW, I wonder if the QB-9 already reached the absolute level of human perception :) why are even other manufacturers still upgrading their USB DACs? The MSBs, the dCSs, Metric Halos etc...

 

Look and learn MSB and dCS, that's how a SOTA DAC looks like.

 

92912ayre3.jpg

Link to post
Share on other sites
Hiro,

 

There is no need to create a new version of a non-standard digital encoding system to reach that performance level. 192/24 PCM easily reaches (or exceeds) the levels of performance of DSD-256, but without adding the extreme amounts of out-of-band noise that all variations of DSD do.

 

Regards,

 

Until it hits the DAC where it in 99% of world's playback systems faces poor oversampling digital filters and modulators. Good quality DAC can be made much cheaper and better performing by leaving those out completely and performing processing in a computer instead where there's much more processing power, especially unbeatable MIPS/$ ratio. And much less noisy digital electronics close to analog electronics.

 

Same goes for PCM conversion in ADC too.

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to post
Share on other sites
Hiro,

 

There is no need to create a new version of a non-standard digital encoding system to reach that performance level. 192/24 PCM easily reaches (or exceeds) the levels of performance of DSD-256, but without adding the extreme amounts of out-of-band noise that all variations of DSD do.

 

Regards,

 

Why do you constantly conveniently ignore what happens when your PCM reaches the DAC?

 

There are very few true PCM DACs on the market. For most DACs PCM is as far from the native format as it can get.

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to post
Share on other sites

Zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz....

Charles Hansen

Dumb Analog Hardware Engineer
Former Transducer Designer

Link to post
Share on other sites

By the way Charles, my "simulations" are either measurements of real world DAC analog output or digital domain analysis of player output when subjected to 64-bit floating point input file.

 

I already posted measurement of one PCM DAC output in this thread, clearly showing images repeating every 352.8 kHz up to about 4MHz and also visible noise bump of DAC's modulator at much lower level.

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to post
Share on other sites

When I use PCM DACs, I perform upsampling and ninth order noise shaping to get extra 40 dB SNR in audio band by trading about 10 dB of SNR above audio band. Gives pretty good figures for 32-bit output.... At least no need to worry about loss of dynamic range due to digital volume control.

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to post
Share on other sites
When I use PCM DACs, I perform upsampling and ninth order noise shaping to get extra 40 dB SNR in audio band by trading about 10 dB of SNR above audio band. Gives pretty good figures for 32-bit output.... At least no need to worry about loss of dynamic range due to digital volume control.

Off topic, Just wondering, Miska when are you going to design and build a dac... you seem to have the knowledge.

The Truth Is Out There

Link to post
Share on other sites

Hi folks! So, I suggest reading this:

 

http://www.grimmaudio.com/whitepapers/dsd%20faq.pdf

 

Grimm has an interesting perspective on DSD 128 and higher… This seems sensible enough to me, suggesting that there is no analog circuitry which could even take advantage of the higher resolution of DSD128 and higher...

 

BTW, Miska, it seems from your last post that you claim to have measured S/N ratios of ~200 dB at the analog outputs of a DAC? Sorry, but that is impossible for me to believe.

SO/ROON/HQPe: DSD 256-Sonore opticalModuleDeluxe-Signature Rendu optical--Bricasti M3 DAC--DIY Purifi Amplifier-Focus Audio FS888-JL E 112 sub-Nordost Tyr USB, DIY EventHorizon AC cables, Iconoclast XLR & speaker cables, Synergistic Orange Fuses, Spacetime system clarifiers.                                                       

                                                                                           SONORE computer audio

Link to post
Share on other sites

Charles, regarding your ESS Sabre DAC upgrade. At what bit rate and sample rate does the ESS DAC convert PCM/DSD to analog? Also, which low pass filter (50kHz, 60kHz or 70kHz) are you recommending for DSD playback? No agenda here...I'm just curious.

 

Jesus R

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now


×
×
  • Create New...