PS Audio DirectStream DAC

[Booster MPS]

It is a big letdown because they were so close to getting it right. They chose an FPGA over an off-the-shelf chip, they had total freedom to do a proper direct stream implementation, they even named the dac 'DirectStream', and still failed to deliver direct stream playback.

 

So no need to listen to it and review it then. Complete and utter failure.

[matthias]

So no need to listen to it and review it then. Complete and utter failure.

 

No failure for PCM.

But the irony is that this Directstream DAC has been created for the best playback of PCM, not DSD.

PCM is converted lossless to DSD and in a very simple way to analogue.

But for DSD you have one additional step: The conversion to PCM.

Even Gus Skinas admits in the PS Audio video that PCM sounds better than DSD with this DAC.

 

Hiro is right:

 

They failed to deliver direct stream playback.

 

Matt

[john57]

Yes the music that I was listening to at PS Audio with the Directstream was PCM based as I understood it at the time for the demo. My take with the Gus Skinas statement was that PCM playback on the DirectStream is as good or better than some of the DSD playback he has heard before. He did not say that the PCM playback was better than the DSD playback on the DirectStream as far I can tell.

[Mike Mcsweeney]

Even Gus Skinas admits in the PS Audio video that PCM sounds better than DSD with this DAC.

 

 

 

Matt

 

watch the video again...you misunderstood him.

[Hiro]

No failure for PCM.

But the irony is that this Directstream DAC has been created for the best playback of PCM, not DSD.

PCM is converted lossless to DSD and in a very simple way to analogue.

But for DSD you have one additional step: The conversion to PCM.

 

Exactly. I was only referring to the converter's "DirectStream" playback, which in actuality the DAC can not deliver, as it subjects DSD to two conversions internally: DSD>PCM , PCM >DSD.

[k6davis]

No failure for PCM.

But the irony is that this Directstream DAC has been created for the best playback of PCM, not DSD.

PCM is converted lossless to DSD and in a very simple way to analogue.

But for DSD you have one additional step: The conversion to PCM.

Even Gus Skinas admits in the PS Audio video that PCM sounds better than DSD with this DAC.

 

Hiro is right:

 

They failed to deliver direct stream playback.

 

Matt

I really think people make too much of these conversions. In the quest for purity, it's easy to dislike them in principle, but we're talking about extremely high resolutions here, not something like mp3. Of course, I think they should be done properly and only if there is an additional benefit to doing them. But the mere existence of a conversion in the chain doesn't automatically mean disaster.

 

For example, JRiver will bitstream DSD perfectly untouched to your DAC. But if you choose to use volume leveling, or upsampling to DSD128, or other DSP, it converts the DSD to PCM and back to DSD (gasp!!!). The GUI is very upfront about all of this, BTW. These conversions are operating at 64bit-352.8 kHz for DSD64 and 64bit-705.6 kHz for DSD128. Those PCM bitrates and resolutions are not only extremely high, they also match the resolution of the source DSD input exactly. That should be sonically transparent - offensive only to belief systems and not to ears.

 

Besides, and this is the most important point ... PS Audio is saying (and I agree) that the harm of PCM is not inherent in the PCM data. The problem is in PCM decoding. Others have made this point as well. It's the marketing basis for this new DAC. So even a signal that has gone through the dreaded DSD-PCM-DSD process I mentioned above still has the benefit of DSD decoding - with no loss of resolution.

 

I won't change any minds about this, but I can imagine that the benefits of feeding a typical DSD DAC an upconverted DSD128 signal, for example, could easily outweigh the liability of the conversion step.

 

I understand the argument that HQPlayer can perform volume control on DSD directly and I suppose that could be preferable, but I still think that something like a 64bit-352.8 kHz step in the middle of chain gets more disdain than it deserves.

 

TL-DR (too long, didn't read version):

 

While "perfect" is preferable, I doubt a temporary conversion to 64-352.8 PCM (which no DAC can even play) is audible - let alone horrible.

[matthias]

watch the video again...you misunderstood him.

 

Okay guys,

 

Gus Skinas:

PCM, it was so good, it was so close to what I was listening to in DSD, if not better in some ways.

 

Matt

[Hiro]

Besides, and this is the most important point ... PS Audio is saying (and I agree) that the harm of PCM is not inherent in the PCM data. The problem is in PCM decoding.

 

IIRC Ted Smith said in his presentation that no upsampling process is perfect and that 16/44 PCM will never sound as good as a studio master, although upconverting it to 128x DSD helps get the most from the PCM recording.

[k6davis]

IIRC Ted Smith said in his presentation that no upsampling process is perfect and that 16/44 PCM will never sound as good as a studio master, although upconverting it to 128x DSD helps get the most from the PCM recording.

Agreed.

 

My understanding is that the reason for the improvement is that the process of decoding DSD is superior to the process of decoding PCM.

[john57]

I only started to learn about DSD a few weeks ago. I also found that pure DSD is a very limited format as far working with it. Here is what I could understand so far.

Volume attenuation will not be made to the DSD data stream as Ted stated. "You cannot perform math on a 1-bit stream; there is no room, you need multi-bits to perform any math. "

 

This sounds to me if you are going to do any mastering of the DSD master tape you are going to have PCM "container" on the DSD file.

 

The DirectStream accepts DSD directly over the I2S inputs and DoP (DSD over PCM) on all inputs.

 

DoP is a lossless packaging of the raw DSD bits into a 24 bit 176.4kHz stream.

 

All inputs (whether PCM, DSD or DoP) are upsampled to 30 bits @ 10 x the normal DSD rate (28.224MHz).

 

 

Then the DirectStream downsample to 2x DSD and the DSD goes directly to the LP filter.

 

Just now Ted Smith said this on the PS forum:

 

"I have no reason to believe that taking DSD or double rate DSD to 10 x DSD rate PCM and then back to double rate DSD looses anything and indeed have based my design on that assumption."

[john57]

Okay guys,

 

Gus Skinas:

PCM, it was so good, it was so close to what I was listening to in DSD, if not better in some ways.

 

Matt

 

Yes he said that. Just means that the DirectStream is so good that it sound better than the equipment he was using for DSD playback in some cases. Quite a statement I would say.

[Hiro]

Just now Ted Smith said this on the PS forum:

 

"I have no reason to believe that taking DSD or double rate DSD to 10 x DSD rate PCM and then back to double rate DSD looses anything and indeed have based my design on that assumption."

 

Since Ted already admitted that no upsampling process is perfect I don't see how subjecting the 128x DSD signal to upsampling and then downsampling it back to 128x DSD can be beneficial to the signal.

[matthias]

I really think people make too much of these conversions. In the quest for purity, it's easy to dislike them in principle, but we're talking about extremely high resolutions here, not something like mp3. Of course, I think they should be done properly and only if there is an additional benefit to doing them. But the mere existence of a conversion in the chain doesn't automatically mean disaster.

 

I understand your POV, but we are talking not about some disaster but about the highest level of music reproduction today.

For PS Audio the best way to playback PCM is to convert it to DSD to simplify the conversion to analogue. That means for PS Audio the combined conversion to DSD and analogue is better than a conversion from PCM to analogue. They never stated that a conversion from DSD to PCM has any benefit. At the starting point we have DSD and PCM files. The PCM files are converted to DSD by the Directstream DAC. Why should we convert DSD to PCM? For volume control? HQplayer is able to do volume control in the DSD domain. Both Paul McGowan and Ted Smith prefer to have a preamp in their audio system. So we do not need a volume control. What would have been the best way? Surely not to convert DSD to PCM. The best way would have been to convert PCM and DSD to DSD640 or similar, to get rid of the volume control or to perform volume control in the DSD domain like HQPlayer. The real benefit of the Directstream DAC is their conversion from DSD128 to analogue. I like this approach very much. That would be my DAC of choice.

 

KR

Matt

[Ted Smith]

Howdy All

 

I'm the Ted Smith mentioned a few times in this thread. I know ted_b thru the AudioAsylum, but I'm new here. I thought I'd volunteer to answer any questions you'all have directly here if that's in the spirit of the forum.

 

-Ted

[matthias]

Howdy All

 

I'm the Ted Smith mentioned a few times in this thread. I know ted_b thru the AudioAsylum, but I'm new here. I thought I'd volunteer to answer any questions you'all have directly here if that's in the spirit of the forum.

 

-Ted

 

You are very welcome here, Ted, it is a pleasure to have you here.

My question: Why does the Directstream convert DSD to PCM 30 bit / 28 MHz ?

 

Thanks

Matt

[Hiro]

Howdy All

 

I'm the Ted Smith mentioned a few times in this thread. I know ted_b thru the AudioAsylum, but I'm new here. I thought I'd volunteer to answer any questions you'all have directly here if that's in the spirit of the forum.

 

-Ted

 

Hi Ted,

 

Is there any chance we will see a more affordable version of the DirectStream DAC released down the line that will be capable of true direct stream playback of DSD files?

[ted_b]

Welcome Ted! (btw, does the name "Ted" mean you like DSD! Seems to. )

 

I think you'll find us fair and thoughtful. We can't ensure that there are no trolls, but we do a decent job of starving them.

[Ted Smith]

Howdy

 

I'll take a stab at some of the comments in reverse order I see them in this thread (ah I see some have already been directly asked again I'll post what I've typed so far.)

 

No upsampling is perfect, but synchronously upsampling keeping at least the same number of bits, etc and going to a higher sample rate is fairly benign. Upsampling all inputs to the same rate so everything can be handled the same way has obvious benefits. Blindly accepting any input DSD stream, single or double rate and sending to my output filter has some (perhaps unlikely) risks. One potential "bug" in the stream: is it DC balanced?, in particular is it all ones or zeros. A SACD DSD stream shouldn't have less than four or more than 24 ones per 28 samples (I'm being sloppy, it might be less than 5, more than 23, etc.) I have the same 0dBFS as SACD (-6dB from all zeros or all ones) and need approximately the same DC balance to get thru the isolation caps on the way to the reclocking flipflop... Also it's not good to saturate the output transformer (tho it is quite oversized for this use)

 

Similarly an arbitrary input DSD or double rate DSD may have an arbitrary "DSD noise hump". My modulator has a fairly small hump of about 15dB out at 60kHz. This is significantly more benign than SACD's allowed 80dB hump. Keeping the ultrasonic noise shaping noise at as high of a frequency as possible helps everything down stream, not just my output filter but the preamp, ...

 

Tho technically you do interesting mixing, filtering, etc. in the sigma delta modulator and so technically always use 1 bit between every block, there's no downside at all to letting the single bit get wider to accommodate the operations (e.g. two bits for a sum) and then do the remodulation from the wider word. I've chosen a 30 bits since it's the natural output size of my upsampler and 10 x DSD rate is the LCM of 192000 and 176400.

 

Tho people at times think that single bit is THE defining feature of DSD, it's really it's high rate and noise shaping. I do maintain that the simplest signal to convert from digital to analog is a one bit signal (i.e. all you need is a low pass filter) But there's no reason to assume that you loose all DSD goodness if you allow the bit width to grow. The key point is to not loose info by capriciously lowering the sample rate or loose info by truncating the size with an appropriate re sigma delta modulation (analogous to using dither when truncating width in PCM.)

 

-Ted

[Ted Smith]

Is there any chance we will see a more affordable version of the DirectStream DAC released down the line that will be capable of true direct stream playback of DSD files?

 

As I intimated this version already supports DSD well. I have believed in DSD over low rate PCM for years and think all music deserves the best we can do in algos, etc. Or perhaps I should say I lament DSP written by people that obviously have never taken a numerical analysis/methods class.

 

I can't speak to products that PS Audio may or may not choose to develop, but they do recognize the hole in their product line left by having the DirectStream having a higher price than it's predecessor. They also firmly believe in this technology.

[Ted Smith]

Yes Gus was amazed that PCM could sound as good as it did on my prototype. PS Audio is trying to make the point that there's a lot more information in recordings than often reproduction equipment delivers.

 

We didn't take the time to hook up DSD at Gus's. He would have heard what he expected in that case, DSD still has the edge over lower rate PCM.

 

I'm not trying to put words in his mouth, but he wasn't repudiating DSD in any way.

[Ted Smith]

No failure for PCM.

But the irony is that this Directstream DAC has been created for the best playback of PCM, not DSD.

PCM is converted lossless to DSD and in a very simple way to analogue.

But for DSD you have one additional step: The conversion to PCM.

Even Gus Skinas admits in the PS Audio video that PCM sounds better than DSD with this DAC.

 

Hiro is right:

 

They failed to deliver direct stream playback.

 

Matt

 

Howdy

 

I would disagree, I put PCM connector on my DSD DAC board just because it seemed silly not to when I added an FPGA. Since I'm primarily a software guy I assumed that after I got the DSD DAC working I'd be able to shoehorn in PCM upsampling. Upsampling proper is not magic. Perhaps choosing good sounding upsampling filters is.

 

There's no need to get religious about DSD vs PCM. In my mind the critical thing is that the sample rate is high enough to not loose information and that you not drop bits on the floor when you do any kind of processing. Single bit is important exactly when you go to analog in that it allows the whole converter to be just a low pass filter.

 

Editing, PCM or DSD is usually done using a wider internal sample size and often at a higher sample rate. After that it's shoehorned with dither and/or sample rate conversion to fit on the distribution media. But any further edits are best done with the material before it was compromised for distribution. In DSD land single rate DSD isn't the best editing format. Going to double rate allows many many remodulations before the noise grows too high and encroaches on the audible bandwidth. Keeping 8 bits or more at double rate during editing helps a lot too.

 

-Ted

[gabeg]

Hi Ted -

 

I know one of the big criticisms of dsd has been the ultrasonic noise. Are you saying that you've managed to reduce that noise to a level that's really no more than pcm might experience? Also, is your process in theory similar to what NAD does with their M51 dac? Finally, does you dac allow for direct connection to a power amp (or in my case an active crossover)? I'm talking output impedence, etc.

 

 

 

Similarly an arbitrary input DSD or double rate DSD may have an arbitrary "DSD noise hump". My modulator has a fairly small hump of about 15dB out at 60kHz. This is significantly more benign than SACD's allowed 80dB hump. Keeping the ultrasonic noise shaping noise at as high of a frequency as possible helps everything down stream, not just my output filter but the preamp, ...

 

[Ted Smith]

It will be interesting to listen to the final product. The fact that it switched PS Audio from DSD critics to believers suggests it will be one to evaluate.

 

I was also amused to hear that while Ted was inspired to design it after hearing a Sony Multichannel SACD demo, the unit is only 2 channel! I guess he was impressed, but not that impressed....

 

I know you are kidding but I have a MC system at my house and really enjoy it compared to two channel. I have literally thousands of MC SACDs and enjoy most of them.

 

My MC transport has given up the ghost and my grandkit poked his finger thru one of my rear beryllium tweeters so till I get some breathing room I'm stuck with two channel

[matthias]

Ted,

 

what is the rationale for the upsampling to PCM 30 bit / 28 MHz at all ? Volume control?

 

Why not direct to DSD128?

 

Thanks

 

Matt

[Ted Smith]

Hi Ted -

I know one of the big criticisms of dsd has been the ultrasonic noise. Are you saying that you've managed to reduce that noise to a level that's really no more than pcm might experience? Also, is your process in theory similar to what NAD does with their M51 dac? Finally, does you dac allow for direct connection to a power amp (or in my case an active crossover)? I'm talking output impedence, etc.

Howdy

 

The output of the DAC is transformer coupled and has plenty of available current behind it. To implement an optional 20dB attenuator we use a relay to put in 15 ohm resistor shunts to ground. Paul at PS Audio historically has preferred no preamp and until very recently always connected it directly to his power amp.

 

Offhand I don't know what NAD does.

 

I keep the DSD hump down by using double rate DSD to give me room to relax the noise shaping. Since I'm "the final stop" I don't need to have a S/N > 160dB (or whatever) over the audio band. I lessen the noise shaping (and the height of the hump) by trading off some theoretical headroom in the audio band than I'll never reach in my hardware. I also use the higher sample rate to allow my passive low pass filter to not be as steep as it otherwise might have need to be.

 

-Ted