Review and Measurements of PS Audio PerfectWave DirectStream DAC

[amirm]

That worries me.
I have come here because I wanted another tool in search of best value/performance gear.
Me personally doesn't know half what is written here. But I used his measurements to know how good a product is. Just like magazines use words to describe how good a product is.
But now I read on this other site that maybe the measurements are not done in the right way. This leads to a situation where I don't know what to believe anymore. Because I don't know all these technical stuff.
So it would be great if Amir would address these points.
Now it seems like there can be more different measurements taken and that the analyzer was not calibrated right. And we are again back at the start. Because then we have to choose whose measurements we believe. This is frustrating for someone with little knowledge.

I have addressed all of that extensively in the past. There are entire threads on it. See: https://www.audiosciencereview.com/...ts-of-schiit-yggdrasil-dac-inconsistent.3812/

Remember, Schiit advertises on their site and they have built a reputation of saying Schiit is the best audio gear there is. So of course they try go create some doubt. But it is without merit. I have also caught Marv in the past using USB cleaners to hide jitter issues with Schiit gear without being clear about that.

BTW, the link referenced was with my old analyzer. All the measurements you see now and across the last 175+ DACs are with my new, state-of-the-art APx555 analyzer I purchased in June of 2018: https://www.audiosciencereview.com/...s-a-new-audio-precision-analyzer-apx555.3442/

They are using toy-level analyzers compared to what I use.

My analyzer just came out of warranty. Yet is accuracy is the same as my previous 20+ year old AP analyzer. These are precision electronic components and simply do not degrade with respect to their application. I asked AP how often I need to calibrate the new one and their answer was basically never unless there were legal liability in some measurement.

If you disagree, tell me how often you calibrate your DAC! The analyzer is nothing but a DAC/signal generator and Analog to Digital Converter.

Note that there should be no dispute whatsoever regarding the test results of PS Audio PerfectWave DirectStream DAC. The reason was in the review. They publish the distortion spec on their website:

You see that line that says "THD&IM" that is less than 0.03%? This is what I got:

So my measurements actually show lower distortion than their own!

People did not pay attention because there was no comparative data. Here, we have that. We convert THD+N to dB which becomes SINAD. PS Audio's 0.03% THD+N would translate into SINAD of just 70 dB. That would have made the DS DAC the 7th worst DAC I have ever measured, behind a $20 phone dongle (VE)!!!

My measurements were also better than those conducted by stereophile.

And confirmed indirectly by showing distortions that even good transformers could create.

There just isn't an argument here regarding test results. I welcome PS Audio to show us Audio Precision analyzer test results that are different. As it is, we don't even know if they have a proper audio analyzer given the vague THD number above and no screenshots whatsoever.

Note that the pressure from my tests caused Schiit to purchase the same analyzer and start to build better products and release same kind of test documentation that I do. So whatever argument SBAF people had was just that: empty arguments.

I hope PS Audio does the same and doesn't just go by some unscientific impression of sound by a couple of people to build and sell $6000 DACs. You can't design seriously good audio gear without proper measurements. They are certainly big enough to be able to afford a proper analyzer.

If they never publish such measurements, then you know to stay away from them. That would be my recommendation.

 

[Krunok]

Most audio people are pretty ugly imo. Paul Barton, Bob Carver, this guy at PS Audio, the Schiit guys, Thom Holman, Henry Kloss. Siegfried Linkwitz was cute though, RIP.

Well, you can't expect all of them to have that cute Japanese look like Amir, right? And I didn't even mention classical beauty of Blumlein88..

 

[GrimSurfer]

It would be interesting to test PS Audio’s DAC with earlier software loads as a way of determining the magnitude of improvements these provide.

That said, I’m slightly uncomfortable with the notion of software driven audio components. Why?

1. It gives manufacturers an excuse to release the devices for sale before the problems are fully resolved;

2. It implies that future software patches will actually improve performance;

3. It introduces the very real possibility that the software updates will end when the next generation hw device is released; and

4. It represents another point of failure for the user to contend wit (especially when a sw load fails and locks out a user.

 

[AudioSceptic]

I have addressed all of that extensively in the past. There are entire threads on it. See: https://www.audiosciencereview.com/...ts-of-schiit-yggdrasil-dac-inconsistent.3812/

Remember, Schiit advertises on their site and they have built a reputation of saying Schiit is the best audio gear there is. So of course they try go create some doubt. But it is without merit. I have also caught Marv in the past using USB cleaners to hide jitter issues with Schiit gear without being clear about that.

BTW, the link referenced was with my old analyzer. All the measurements you see now and across the last 175+ DACs are with my new, state-of-the-art APx555 analyzer I purchased in June of 2018: https://www.audiosciencereview.com/...s-a-new-audio-precision-analyzer-apx555.3442/

They are using toy-level analyzers compared to what I use.

My analyzer just came out of warranty. Yet is accuracy is the same as my previous 20+ year old AP analyzer. These are precision electronic components and simply do not degrade with respect to their application. I asked AP how often I need to calibrate the new one and their answer was basically never unless there were legal liability in some measurement.

If you disagree, tell me how often you calibrate your DAC! The analyzer is nothing but a DAC/signal generator and Analog to Digital Converter.

Note that there should be no dispute whatsoever regarding the test results of PS Audio PerfectWave DirectStream DAC. The reason was in the review. They publish the distortion spec on their website:

You see that line that says "THD&IM" that is less than 0.03%? This is what I got:
View attachment 34380

So my measurements actually show lower distortion than their own!

People did not pay attention because there was no comparative data. Here, we have that. We convert THD+N to dB which becomes SINAD. PS Audio's 0.03% THD+N would translate into SINAD of just 70 dB. That would have made the DS DAC the 7th worst DAC I have ever measured, behind a $20 phone dongle (VE)!!!

My measurements were also better than those conducted by stereophile.

And confirmed indirectly by showing distortions that even good transformers could create.

There just isn't an argument here regarding test results. I welcome PS Audio to show us Audio Precision analyzer test results that are different. As it is, we don't even know if they have a proper audio analyzer given the vague THD number above and no screenshots whatsoever.

Note that the pressure from my tests caused Schiit to purchase the same analyzer and start to build better products and release same kind of test documentation that I do. So whatever argument SBAF people had was just that: empty arguments.

I hope PS Audio does the same and doesn't just go by some unscientific impression of sound by a couple of people to build and sell $6000 DACs. You can't design seriously good audio gear without proper measurements. They are certainly big enough to be able to afford a proper analyzer.

If they never publish such measurements, then you know to stay away from them. That would be my recommendation.

You use SINAD (THD + N). PS Audio quote THD + IM. Is it correct to compare them directly? Did you get an IM figure?

 

[Cortes]

I think they will learn as did Shiit, make sure they have measurements before releasing new products.

I hope this review will obligate them to finish the lifecycle of the DirectStream. They have beeng selling the same
obsolete hardware for many years, thanks to marketing generated by ther continous firmware releases. Every new release was received by the audiophile community as a new high end DAC, and received a proper cool name.

Indeed, I suppose that they use FPGA not for any musicality or engineering reason, that's because it's the perfect tool to have people purchasing year after year the same obsolete hardware. From a business point of view, a brilliant idea. Now they'll have to work more to meet marketing with good engineering.

 

[617]

I think we should get Audio Precision to make an AVR.

Seriously though that distortion spectra missed my attention before. Third dominated. Just ugly. PS Audio seems like a reputable company, but 6K for this DAC, performance issues aside, is pretty bad. The case and interface aren't that nice. But with the performance? Just terrible.

If I were PS audio I would buy an off the shelf topping, pot it in some kind of opaque epoxy so you can't see what it is, and put it inside a nicer case, like Goldmund and Lexicon did with dvd players.

 

[Blumlein 88]

Slightly OT, I hope you get an opportunity to test out the new EtherREGEN:
https://audiophilestyle.com/forums/...oduction/page/54/?tab=comments#comment-992008

Once again John is going to develop measurements that show the worth of it and write a white paper when he gets time. Of course it never happens. Funny how you design all this gear (and ethernet switch) and get greatly improved sound results without measuring the result or why it happens.

 

[mkawa]

It would be interesting to test PS Audio’s DAC with earlier software loads as a way of determining the magnitude of improvements these provide.

That said, I’m slightly uncomfortable with the notion of software driven audio components. Why?

1. It gives manufacturers an excuse to release the devices for sale before the problems are fully resolved;

2. It implies that future software patches will actually improve performance;

3. It introduces the very real possibility that the software updates will end when the next generation hw device is released; and

4. It represents another point of failure for the user to contend wit (especially when a sw load fails and locks out a user.

this is exactly how microsoft worked in the 90s. bank on your name for 3 major versions that are horrid, and then gradually get your act together until you bury your competitor...

 

[Blumlein 88]

You use SINAD (THD + N). PS Audio quote THD + IM. Is it correct to compare them directly? Did you get an IM figure?

Well what PS Audio has listed is THD and IM. Not quite the same as THD+IM (which isn't really a thing). What they are saying is THD will be less than .03% and IM will be less than .03%. Not some combination of the two.

 

[amirm]

You use SINAD (THD + N). PS Audio quote THD + IM. Is it correct to compare them directly? Did you get an IM figure?

They should not mix the two as they are different measurements. In addition, many people are confused about THD measurements and don't realize that analyzers include noise in there. So the correct designation is what I use: THD+N. In summary, I think they mean THD+N and just not using the right notation.

IM distortion is shown in much more detail in my measurements based on level:

Worst case SINAD there is 65 dB. But note that there are different IMD tests. Mine is SMPTE which uses a low frequency tone and brings out the type of issues we see here. They don't specify their so hard to tell what they measured. Or for that matter, at what output level.

And to be clear, they are saying THD and IMD independently are 0.03%. No one adds them together as I think you are saying.

 

[GrimSurfer]

I think we should get Audio Precision to make an AVR.

Seriously though that distortion spectra missed my attention before. Third dominated. Just ugly. PS Audio seems like a reputable company, but 6K for this DAC, performance issues aside, is pretty bad. The case and interface aren't that nice. But with the performance? Just terrible.

If I were PS audio I would buy an off the shelf topping, pot it in some kind of opaque epoxy so you can't see what it is, and put it inside a nicer case, like Goldmund and Lexicon did with dvd players.

Maybe the middle path would be best...

1. Develop their own products and seek opportunities to learn/improve.

2. Set a fair price to performance scheme.

3. Be transparent, open, and honest

4. Stay away from trite, folksy, or hyperbolic advertising.

 

[GrimSurfer]

this is exactly how microsoft worked in the 90s. bank on your name for 3 major versions that are horrid, and then gradually get your act together until you bury your competitor...

The market was wide open then.

The consumer audio business is saturated with overpriced crap. That’s a problem for both consumers and producers of said crap.

 

[AudioSceptic]

Once again John is going to develop measurements that show the worth of it and write a white paper when he gets time. Of course it never happens. Funny how you design all this gear (and ethernet switch) and get greatly improved sound results without measuring the result or why it happens.

This has always puzzled me. How do you know what components to use if you don't measure the results? Where do you even start, and how do you know if your production units perform the same as your prototype? Listen to every one?

 

[JJB70]

I still struggle to understand why anyone would pay 6k for a DAC and try and pretend it was about SQ. I can get why you might buy such a product if it offers drop dead gorgeous design or as a statement (even if it's not what I'd do) but in terms of SQ why would you spend that much on a part of the audio chain where fully transparent performance costs peanuts? If you want to colour the sound then use EQ or DSP.

And how can anyone claim that developing a product in a big box with transformers costing thousands of dollars is good engineering? Good engineering is making a $9 DAC dongle with superb performance.

 

[AudioSceptic]

Well what PS Audio has listed is THD and IM. Not quite the same as THD+IM (which isn't really a thing). What they are saying is THD will be less than .03% and IM will be less than .03%. Not some combination of the two.

You're right. I misread that: THD&IM. So each, separately, not together. That still excludes N, though. How do get THD + N? Is it simply the worst of the 2, or is there a formula for combining them?

 

[amirm]

Is it simply the worst of the 2, or is there a formula for combining them?

There is but again, I am confident they mean THD+N. Measuring them separately is much harder (requires spectrum analysis).

 

[AudioSceptic]

They should not mix the two as they are different measurements. In addition, many people are confused about THD measurements and don't realize that analyzers include noise in there. So the correct designation is what I use: THD+N. In summary, I think they mean THD+N and just not using the right notation.

IM distortion is shown in much more detail in my measurements based on level:

Worst case SINAD there is 65 dB. But note that there are different IMD tests. Mine is SMPTE which uses a low frequency tone and brings out the type of issues we see here. They don't specify their so hard to tell what they measured. Or for that matter, at what output level.

And to be clear, they are saying THD and IMD independently are 0.03%. No one adds them together as I think you are saying.

Thanks. I misread their spec.

 

[Rja4000]

There is but again, I am confident they mean THD+N. Measuring them separately is much harder (requires spectrum analysis).

Speaking about spectrum analysis: what FFT window size are you using?
32k for 48kHz and 128k for 192kHz?
And for the 1/3 octave?

 

[amirm]

Speaking about spectrum analysis: what FFT window size are you using?
32k for 48kHz and 128k for 192kHz?
And for the 1/3 octave?

Every test is different. The dashboard uses 32K points because it is fast. Stand-alone FFT/Multitone use 256K points.

Sample rate for dashboard is 44.1 kHz. Jitter and I think IMD are 48 kHz. Multitone is 192 kHz.

What is the 1/3 octave?

 

[amirm]

By the way, I wrote a much more detailed article on my measurements for Widescreen Review Magazine. I suspect it is published by now so when I get a chance, I will upload it here (have to reformat the pictures for web).