Review and Measurements of PS Audio PerfectWave DirectStream DAC

[LTig]

Conclusions
[..]
The sonic effects are there in my semi-formal tests. Perhaps the older audiophiles including the designer Ted Smith, have lost so much high frequency hearing that the harmonic distortion this DAC adds makes up for some of that and they think they are hearing more. If that is the case, just buy a Jensen transformer and put it on the output of your DAC and you would get the same effect!

I took part at a sighted listening test with/without transformer between CD player and amplifier. That was some 30 years ago at one of our local snake oil dealers. AFAIR the sound with transformer had more bass and therefore sounded fuller. Nothing what an EQ could not do as well - if you had one in the first place. So using a transformer as built in EQ is nothing new at all.

 

[maty]

...The French so-called authorities had already decided by then France would no longer have an industry, a process that's nearing completion. I can imagine the same decisions were made in Spain.

Spoiler: Spanish politics partits = sects -> fidelity before merit

I do not think it was a decision, we simply suffered a plague of mediocre politicians throughout Spain, a true reflection of the current widespread mediocrity. The best prepared people do not need the politics to live and are not willing to participate in organizations that function as sects and stop making a lot of money in private enterprise. Even new parties already work like old ones. The political class is a reflection of Spanish society. The same happens in other countries like...

 

[audioBliss]

Yet another one bites the dust. They can keep on fooling themselves but anyone who cares about Hi-Fi will not buy this DAC. Such a great time to be an audiophile where we can get data like this as well as go out and buy a reference DAC for a few hundred $ instead. LOL! I see all these companies as scammers and it's no use trying to explain to them or their followers as they clearly don't have the knowledge to understand. So far all these expensive DAC scammers have tried to convince their buyers otherwise but I think soon people will start to understand and then they will be forced to do better.

 

[prighello]

Our club recently did a semi blind dac shootout. The PSA Directstream did not get many votes. The Chord Qutest came in second. However, the Benchmark Dac3 which measures well also didn't score well.

I was also surprised by the low output of the Directstream dac which was evident during level matching with the other Dacs.

 

[Interference]

Right on… generalities are seldom helpful in comparison to case study specifics.

For more details on the internals of sigma delta modulation, the Matlab DS toolbox is a great resource
https://pdfs.semanticscholar.org/c45b/0a23369e43e68ccfdcb8b71f894952b8184d.pdf
or the Python port of same which is freeware, no Matlab license required
http://www.python-deltasigma.io/.

For Rob Watts’ approach at Chord, here are two good starting points:
http://tagnz.co.nz/wp-content/uploads/2015/01/Rob-Watts-DAC-technology-for-Asia-2014.pdf
https://www.moon-audio.com/mwdownloads/download/link/id/159/

For the PS Audio Direct Stream design, here’s a brief overview and you’ll find some lengthy interviews with Ted that provide more details:
https://www.psaudio.com/wp-content/uploads/2014/02/DirectStream-DAC-white-paper.pdf

For the Exogal Comet, this is a good introduction:
https://www.exogal.com/images/PDFs/EXOGAL_DAC_and_PowerDAC_White_Paper.pdf

For R-2R, here’s a recent overview and you’ll find lots of software simulations online which allow you to explore different design strategies and step through code with test data
https://pdfs.semanticscholar.org/c314/7d9cbc59f187dcdd99763262f4b3205c3d3b.pdf

FWIW both my DACs are sigma delta so not taking sides, just impressed with all the hard work that these companies and others are doing to raise the bar on playback quality.

- cheers

I appreciate the effort, but none of this seems to cover your original claim.

The marketing material from PS Audio is embarassing at best (calling it a "white paper" is a bit of an overstatement).

As for Chord, I will not spend time on an article with claims as superficial as "the more times you oversample, the smoother the waveform becomes", simply because it's very hard to discuss any concept if jargon is ill-defined to begin with.

The HifiCritic article is more interesting, but still quite elusive and not really pertinent. There may be benefits in a better PCM interpolator filter, but this has little to do with the design of the actual D/A converter (be it sigma delta or R2R) and the following analog reconstruction filter.

I will give a look at the Exogal documents, but I do not expect honest and accurate technical descriptions by manufacturers anymore, everyone is so focused on over-hyping their efforts.

P.S.: on the contrary, I am not impressed at all. Everybody tries to re-invent the wheel going to extreme lengths designing sophisticated solutions to problems that the industry has well understood and treated over the years. I do not see these one-engineer-show companies beating decades of know-how and millions USD spent in R&D by established chipmakers.

 

[TomJ]

I appreciate the effort, but none of this seems to cover your original claim.

The marketing material from PS Audio is embarassing at best (calling it a "white paper" is a bit of an overstatement).

As for Chord, I will not spend time on an article with claims as superficial as "the more times you oversample, the smoother the waveform becomes", simply because it's very hard to discuss any concept if jargon is ill-defined to begin with.

The HifiCritic article is more interesting, but still quite elusive and not really pertinent. There may be benefits in a better PCM interpolator filter, but this has little to do with the design of the actual D/A converter (be it sigma delta or R2R) and the following analog reconstruction filter.

I will give a look at the Exogal documents, but I do not expect honest and accurate technical descriptions by manufacturers anymore, everyone is so focused on over-hyping their efforts.

P.S.: on the contrary, I am not impressed at all. Everybody tries to re-invent the wheel going to extreme lengths designing sophisticated solutions to problems that the industry has well understood and treated over the years. I do not see these one-engineer-show companies beating decades of know-how and millions USD spent in R&D by established chipmakers.

haha, the Exogal article is even worse, if that's possible. And as someone will point out, "the Comet uses a sigma delta chip" (true, but only as an output stage, clever). that's why I suggested resources for using software simulations with test data to explore the details more closely, otherwise it's too easy to get tripped up in semantics.

 

[LTig]

To your first question "why FPGA", the advantage of an FPGA solution is that it has the potential to side-step the need for a reconstruction filter. [..]

An FPGA is nothing more than a piece of programmable hardware. It could be replaced by a DSP or a microcontroller, depending on the task. I cannot see how it would replace the reconstruction filter. And omitting the reconstruction filter makes things worse, as we have seen with some NOS-DACs without it.

By upsampling all incoming data to DSDx2 or more (this is probably what the code in the FPGA does) the construction filter could be a simple 1st order low pass filter - but it still is a reconstruction filter.

 

[maxxevv]

Our club recently did a semi blind dac shootout. The PSA Directstream did not get many votes. The Chord Qutest came in second. However, the Benchmark Dac3 which measures well also didn't score well.

I was also surprised by the low output of the Directstream dac which was evident during level matching with the other Dacs.

So which came out tops ?

I have come across similar comments on the Benchmark, but I always wondered why too, since its such a well measuring device.

 

[JBNY]

PS Audio recently dropped all its US dealers on short notice. Some dealers responded by unloading their DS inventory at $3,000 each (new, sealed box). Haha, no thanks. Likewise their S700 monoblock amps, where JA measured 877mV (?!) RF noise on the output jacks. Won't be buying those either.

You could always buy them at that price, you just needed a decent relationship with the dealer. They didn't unload them, just people started to buy them up before your only chance was to go direct through PS Audio for substantially more money.

 

[ahofer]

Spoiler: Album Player 2.111, a soft player from Russia (with own Kernel Streaming)

Some hours ago I made tests with SoX implementation of AP2 - Album Player soft player.
https://www.audiosciencereview.com/forum/index.php?threads/foobar-or-jriver.8364/post-233239

I could appreciate the differences between the three types of filters implemented. Many say they are incapable. Of course, the interface was that Russian program but the engine was JRiver MC.

...

And also when marking the Ultra High Quality and Steep filter options.

http://albumplayer.ru/english.html

Perhaps. Forgive me if my Bayesian priors make me skeptical of your ability to replicate under blind conditions - nothing personal. All I know is that most people have a very hard time distinguishing resolutions above 320k mp3 (hence my skepticism about any one individual), let alone reasonable filters.

For myself, I did a little better on orchestral music than other genres (in distinguishing lossless from 320 mp3).

 

[Interference]

haha, the Exogal article is even worse, if that's possible. And as someone will point out, "the Comet uses a sigma delta chip" (true, but only as an output stage, clever). that's why I suggested resources for using software simulations with test data to explore the details more closely, otherwise it's too easy to get tripped up in semantics.

OK. At this point I just think your original view is misguided The way I would put it is: oversampling itself is what makes the analog filter less critical, since it can be moved out of the audio band. This allows for simplified approaches like those passive approaches that are appealing to purist audiophiles, but have no actual benefit over the very precise and stable SCFs commonly used in DACs.

It's not a merit of this sophisticated DSP approaches to allow for simpler analog filters. And simpler analog filters have little to do with audio quality.

 

[LTig]

For the Exogal Comet, this is a good introduction:
https://www.exogal.com/images/PDFs/EXOGAL_DAC_and_PowerDAC_White_Paper.pdf

One more company that has not understood the sampling theorem in the first place ...

AFAIR there was a posting at ASR not too long ago where it became clear that interpolating according to some wishful thinking (linear, splines, whatever) is worse than simply using the proper reconstruction lowpass filter (EDIT: upsampling is interpolation followed by low pass filter).

EDIT: See Rick Lyons: Understanding Digital Signal Processing, 3rd edition, chapter 10.5.1 Time and Frequency Properties of Interpolation, 4th paragraph.

 

[Frank Dernie]

We'll see. Hope it is since Chord touts their measurements so much in their marketing. Chances are it might not even beat the current king Matrix X-Sabre MQA DAC since Chord is custom FPGA implemented. Based on empirical data, those who know how to truly maximize off-the-shelf DAC chips are the ones who measure extremely well. Any further complications/implementation just deviate from the absolute transparency. That's also why I'm very curious with the Linn Klimax DSM which probably costs about 1.5x DAVE but it uses an off-the-shelf AK4497EQ DAC chip

I seem to remember when I tried the Linn Klimax DS, which was current then, the only difference between it and the next model cheaper was the machined from solid case and a transformer coupled output, the pcb and power supply were identical.

 

[Ron Texas]

@amirm when I saw the broken pink panther in the first photo I knew what the result was going to be. With that kind of performance one might as well use the Realtek chip built into their motherboard.

File under Y, for why and Yaggi.

 

[davidvanderbilt]

Does this say anywhere what volume level he used for testing? I'm assuming 100 but just wanted to confirm.

 

[prighello]

So which came out tops ?

I have come across similar comments on the Benchmark, but I always wondered why too, since its such a well measuring device.

The winner was a surprise to all. The Black Ice Audio (Jolida) Glass FX DSD.

 

[AudioSceptic]

The winner was a surprise to all. The Black Ice Audio (Jolida) Glass FX DSD.

A tube/valve device, but at least the price is not outrageous at $800.

 

[maxxevv]

The winner was a surprise to all. The Black Ice Audio (Jolida) Glass FX DSD.

Never heard of them, but looking at their specs, seems reasonable if they use tubes. No outrageous claims that obviously stick out. (The DR figures perhaps. )

But would imagine it faring quite poorly on USB and closer to its specs using Optical or Coaxial due to the USB decoder.

 

[VintageFlanker]

Our club recently did a semi blind dac shootout. The PSA Directstream did not get many votes. The Chord Qutest came in second. However, the Benchmark Dac3 which measures well also didn't score well.

I was also surprised by the low output of the Directstream dac which was evident during level matching with the other Dacs.

Interesting.

What were the DACs? How did you deal to match levels considering higher output of the DAC 3 (Or Qutest in 3V mode)?

Also, what was the exact process of the semi-blind tests? Could be relevant to everybody here.

Did every listener give clearly usables stats? If the results were somewhat 28% 23%, 22%, 27%, it may be attributed to a randomness factor. Would you tell us more?

 

[AudioSceptic]

I seem to remember when I tried the Linn Klimax DS, which was current then, the only difference between it and the next model cheaper was the machined from solid case and a transformer coupled output, the pcb and power supply were identical.

Would you have considered the Klimax if the price was more reasonable? I mean, even at ½ the price it would still be expensive!