PS Audio DirectStream DAC

[christopher3393]

Looks like this will be officially announced March 1 and available in April at around $6K:

 

Mono and Stereo High-End Audio Magazine: PS Audio DirectStream DAC

[firedog]

Will be very interested in hearing reviews of this based on listening. In concept very intriguing.

[Savage]

Another article here - To DSD or Not to DSD, That Is The Question: PS Audio PerfectWave DirectStream DAC v Berkeley Audio Design Alpha DAC Reference Series | AudioStream

[Matias]

They removed the light blue blocks of power and the active output drivers, so that surely simplifies the diagram at least.

[john57]

I have already heard the thing quite a improvement to the sound over the PWD.

[k6davis]

I have been a big fan of DSD since the SACD days.

 

My (very limited) understanding is that the decoding of DSD has theoretical advantages over PCM because DSD much more closely resembles an analog waveform and therefore allows for a simpler conversion to analog.

 

I've been upconverting all of my music to DSD128 since JRiver 19 made the option available and have been very pleased with the results. This DAC promises to take that concept to a much higher level.

 

If this DAC lives up to its promise, I suspect that other DACs will follow that convert all input to DSD and eliminate PCM processing altogether.

[christopher3393]

A little bit of additional info from John Darko, who will have a review unit soon:

 

PS Audio to 'rescue' PCM with DirectStream PerfectWave DAC | Digital Audio Review by John Darko

 

Includes a link to this "white paper" from PS Audio, dated today:

 

http://www.psaudio.com/wp-content/uploads/2014/02/DirectStream-DAC-white-paper.pdf

 

The Direct Stream will employ a Xilinx Spartan 6 FPGA (Field Programmable Gate Array) instead of a DAC chip.

[Regnad]

Could someone please explain how a "digital" FPGA can output an analog signal? Aren't off-chip electronics needed, filtering, low pass, etc. Thanks.

[wgscott]

They removed the light blue blocks of power and the active output drivers, so that surely simplifies the diagram at least.

 

The need for electricity is part of the great "30 year PCM cover-up" conspiracy.

[barrows]

Could someone please explain how a "digital" FPGA can output an analog signal? Aren't off-chip electronics needed, filtering, low pass, etc. Thanks.

 

I actually believe the FPGA in this case outputs a single bit data stream, which is then converted to analog via a relatively simple, discrete, low pass filter stage. Nice thing is, they convert all incoming data to ~32 bits internally first, and then they can apply a no loss digital volume control (even for DSD input). This is kind of similar to what goes on inside the ESS chip, except that ESS converts a multi bit stream to analog, and PS Audio converts all the way to single bit, allowing for a very simple D-A conversion stage. They are going to have to be very careful with noise, as the processing they run seems fairly complex, but if they can keep the noise out of the clock and analog output, it could be really, really good.

[kana813]

[Regnad]

... which is then converted to analog via a relatively simple, discrete, low pass filter stage.

 

If so the FPGA is a really a DSP, not a complete DAC, so the claim is a bit mis-leading to me. The "conversion" is still based on external circuits and greatly affected by their design and quality.

[One and a half]

If so the FPGA is a really a DSP, not a complete DAC, so the claim is a bit mis-leading to me. The "conversion" is still based on external circuits and greatly affected by their design and quality.

 

An FPGA is only that a Fully Programmable Gate Array. Since there are no specific DSD chips available these days, DAC designers need to roll their own chips by programming the FPGA. It's not strictly a DSP but can be programmed in if the need or want is there.

[ted_b]

 

I guess I'm just not getting it. I mean, there are now 100+ DACs that do DSD (in one way or another), not to mention that PSAsudio was one of those DAC companies that had always said "DSD is not where we are going". I asked several times! And the audience in this Youtube video, the Colorado Audio Society, is not your average newbies. And there are plenty of resamplers (PCM-to_DSD) out there, like JRIver, HQPlayer, Audiogate, etc etc. Why is this being touted as historical??

 

BTW, Chord has several DACs, including the Qute HD, Qute EX and the new Hugo, that are all FPGA-based. FPGA is a wonderful platform for "DAC chip" innovation.

 

Edit: I reread my post, and I;m really not trying to be a jerk here. I REALLY think I am missing what is the breakthrough idea on this DAC. I'd love some feedback. I respect Ted Smith and many of the folks on different forums who are posting things like "intriguing". I am not seeing something.

[k6davis]

I guess I'm just not getting it. I mean, there are now 100+ DACs that do DSD (in one way or another), not to mention that PSAsudio was one of those DAC companies that had always said "DSD is not where we are going". I asked several times! And the audience in this Youtube video, the Colorado Audio Society, is not your average newbies. And there are plenty of resamplers (PCM-to_DSD) out there, like JRIver, HQPlayer, Audiogate, etc etc. Why is this being touted as historical??

 

BTW, Chord has several DACs, including the Qute HD, Qute EX and the new Hugo, that are all FPGA-based. FPGA is a wonderful platform for "DAC chip" innovation.

 

I'll take a stab at why it's historical. I admit that my technical knowledge is rather shallow, so anyone please feel free to correct me.

 

It appears that the power of their FPGA allows them to upsample everything to 10x DSD or DSD6400 (!!!). AFAIK, that's never been done before. While it's certainly an impressive technical achievement, it'll be interesting to see if it actually leads to better sound.

 

The other thing that's new is that they claim to not use a DAC chip, which is only possible with DSD because 1-bit DSD is a much simpler and more direct representation of an analog waveform than multi-bit PCM. In a sense, unlike PCM, DSD is both digital and analog. Said another way, the DSD algorithm is so close to analog that it doesn't need complex/traditional decoding.

 

My (kindergarten) understand of the DSD datastream is it that it simply offers an instruction of whether the next sample in the analog waveform is of lower or higher amplitude than the last - at an extremely high frequency. As PS Audio states in their announcement of this DAC, it's a far superior method of digitizing sound but the limitations of consumer level computing horsepower back in the 80's precluded the use of a technology like DSD.

 

I have heard the LampizatOr DSD (only) DAC, which is actually the very first DAC I've heard of that decodes with no DAC chip, and it sounds amazing. I hope that a reviewer will be able to compare at some point.

 

So, to summarize, the new things I see that the PS Audio brings to the table are:

 

• Fastest sampling rate I've ever heard of
  
• Conversion without (what we think of as) a DAC chip. There's no Burr Brown, ESS Sabre, Wolfson, T.I., etc.
  

Whether or not these attributes are good things and lead to better sound is up for debate, but it think they are what makes this new DAC unique.

[ted_b]

K6davis,

Yes, maybe 10x is new, but as I pointed out in my question FPGAs have been used by companies like Chord to produce wonderful music for awhile now, with their latest FPGA architecture (26k taps, same Xilinx Spartan 6 FPGA as PS Audio is stating, frankly)) being employed in their reference $2k Hugo DAC/hp amp. These FPGA DACs do PCM and DSD (I have a big review of one of them here)..

[barrows]

I guess I'm just not getting it. I mean, there are now 100+ DACs that do DSD (in one way or another), not to mention that PSAsudio was one of those DAC companies that had always said "DSD is not where we are going". I asked several times! And the audience in this Youtube video, the Colorado Audio Society, is not your average newbies. And there are plenty of resamplers (PCM-to_DSD) out there, like JRIver, HQPlayer, Audiogate, etc etc. Why is this being touted as historical??

 

BTW, Chord has several DACs, including the Qute HD, Qute EX and the new Hugo, that are all FPGA-based. FPGA is a wonderful platform for "DAC chip" innovation.

 

Edit: I reread my post, and I;m really not trying to be a jerk here. I REALLY think I am missing what is the breakthrough idea on this DAC. I'd love some feedback. I respect Ted Smith and many of the folks on different forums who are posting things like "intriguing". I am not seeing something.

 

TedB. I do not think you are really missing anything here, but perhaps we should all be a little patient and get a chance to listen. Certainly, using an FPGA to run ones' own custom processing algorithms is nothing new here. But the idea of using an FPGA is not a big deal, Ayre has been running there proprietary OSF in an FPGA for years, and dCS and Chord have been using FPGAs for implementing their own proprietary oversampling and conversion solutions. This is PS Audio, and some of the hyperbole is going to be marketing speak, as it is with many other companies.

That being said, I suspect that the "meat" of this DAC is likely in the proprietary algorithms which Ted Smith has been developing for years, which oversample to 10 times DSD, and then output a 1 bit stream to a discrete low pass filter for conversion. We know there are only a very few DACs which actually convert DSD as a single bit stream (like Meitner/EMM, Playback...), but none of those DACs use the same algorithms as what PS uses. So perhaps PS' approach is better, perhaps not, their claim that there conversion from PCM-DSD is so good is very confident at least... we shall see how good it really sounds.

To me this sounds like a combination of the math centric processes of dCS, Meitner, Playback, and the simplistic DSD only, discrete conversion process offered by LampiZator.

Maybe the results will be spectacular, we shall see.

[barrows]

If so the FPGA is a really a DSP, not a complete DAC, so the claim is a bit mis-leading to me. The "conversion" is still based on external circuits and greatly affected by their design and quality.

 

I think that the point here is that conversion of a single bit data stream to analog, at such a high sample rate, is child's play which can be accomplished by a relatively simple low pass filter with high quality components. It is the math done in the FPGA which makes such simple conversion possible, so what is going on, and the algorithms, in the FPGA are where the majority of the uniqueness of this DAC lie. The FPGA is just a blank slate, it is the code it runs which is going to matter, and, if we believe the claims, Ted Smith has been developing these algorithms for a long time to get to this point (long before he started working with PS Audio).

One of the points which PS seems to be making is that the separation of digital processing (in the FPGA) and the conversion to analog, is key to the overall sound quality, keeping digital processing noise out of the analog stream, by not converting to analog in a silicone chip which is running high speed processing.

[k6davis]

Thanks to you both, ted_b & barrows.

 

I hadn't looked closely at the Chord Hugo, which I thought used some kind of traditional DAC chip, but I was wrong.

 

The proof of the merits of PS Audio's approach to this concept will be in the listening, but I'm definitely getting the impression that these won't be the last DAC's to eliminate the traditional PCM conversion process.

 

I know the excellent sound of the LampizatOr DSD DAC very well. The Hugo has gotten raves. If the PS Audio is anywhere near as good as they are claiming, then this approach will likely gain momentum. Prices for future implementations of this concept could fall quickly, especially if the bulk of the manufacturing expense lies in the new, powerful FPGA chip.

 

What's really interesting is that Berkeley has gone in the opposite direction, and at top speed. They're about to issue a PCM only DAC at $15k (!!!!). It would require you to downsample any DSD content to 24-176. These two DACs will hit the market at approximately the same time. I eagerly await the inevitable reviews comparing these two!

[ted_b]

TedB. I do not think you are really missing anything here, but perhaps we should all be a little patient and get a chance to listen......

That being said, I suspect that the "meat" of this DAC is likely in the proprietary algorithms which Ted Smith has been developing for years......

Maybe the results will be spectacular, we shall see.

 

Barrows, thanks. You are right (frankly I was waiting for your comments ). The combination of FPGA prowess and Ted Smith's 10 years of mad science may be exactly what is needed here.

 

What I find intriguing is the use of FPGA for DSD, cuz Chord's use of it has created incredblly real PCM music...but for DSD it's still not at the level of DSD that I'm used to with Meitner. Maybe DSD focus is what is needed, dunno.

[k6davis]

I think that the point here is that conversion of a single bit data stream to analog, at such a high sample rate, is child's play which can be accomplished by a relatively simple low pass filter with high quality components. It is the math done in the FPGA which makes such simple conversion possible, so what is going on, and the algorithms, in the FPGA are where the majority of the uniqueness of this DAC lie. The FPGA is just a blank slate, it is the code it runs which is going to matter, and, if we believe the claims, Ted Smith has been developing these algorithms for a long time to get to this point (long before he started working with PS Audio).

One of the points which PS seems to be making is that the separation of digital processing (in the FPGA) and the conversion to analog, is key to the overall sound quality, keeping digital processing noise out of the analog stream, by not converting to analog in a silicone chip which is running high speed processing.

 

Well said barrows, and really interesting.

 

I use JRiver to upsample everything to DSD128, which they offered as superior to the upsampling in a DAC chip like the ESS Sabre due to the far more powerful processor in my my computer. The results of sending that signal to the LampizatOr DSD DAC are impressive to say the least.

 

From all indications, PS Audio's upsampling implementation with the FPGA should far exceed that of JRiver's with an Intel chip. What's unknown is how much sonic improvement it yields. I don't discount it - for all I know, it could be tremendous.

 

For me personally, I don't know if the unique attributes of the LampizatOr (tubes in the power supply and output stage, balanced Duelund caps, etc.) could give it an advantage that outweighs the powerful DSP of the PS Audio unit.

 

As I'm still evaluating the LampizatOr, I'm glad to hear that PS Audio has units out for review.

[firedog]

For me the 2 interesting questions are:

does it sound better than upampling to DSD with a PC on the fly, or even compared to upsampling PCM permanently to DSD and playing those files back?

 

How good is the preamp quality?

 

If you can buy this unit and truly not need a pre, it will make the price much more attractive. The price is above my level (especially since it will be much more where I live), but I would consider it if it means I could sell my DAC and not buy a pre.

[Regnad]

An FPGA is only that a Fully Programmable Gate Array.

 

No, it's Field Programmable Gate Array. Which is good because updates can be done by the end user.

[k6davis]

For me the 2 interesting questions are:

does it sound better than upampling to DSD with a PC on the fly, or even compared to upsampling PCM permanently to DSD and playing those files back?

 

How good is the preamp quality?

 

If you can buy this unit and truly not need a pre, it will make the price much more attractive. The price is above my level (especially since it will be much more where I live), but I would consider it if it means I could sell my DAC and not buy a pre.

 

I was thinking that this unit didn't have volume control. The word "volume" isn't mentioned in the lengthy press release at the monoandstereo.com link.

 

Hmmm... If the sound is anywhere near as good as they say and it's a decent pre (great volume control, multiple digital inputs & remote control), then this unit becomes a lot more tempting.

[barrows]

I was thinking that this unit didn't have volume control. The word "volume" isn't mentioned in the lengthy press release at the monoandstereo.com link.

 

Hmmm... If the sound is anywhere near as good as they say and it's a decent pre (great volume control, multiple digital inputs & remote control), then this unit becomes a lot more tempting.

 

Yes, it includes a digital volume control implemented in the FPGA.