Do Most DSD DACs Convert DSD To PCM?

Was reading about the BlueSound Node 2i. Owners have lamented its lack of DSD compatibility. BlueSound released an update that allows the Node 2i to convert DSD to FLAC, and then play the FLAC. BlueSound’s rationale is that nearly all DACs, except for a few Sony models, all convert DSD to PCM internally anyway.

Is that true? Do most DACs claiming DSD support actually internally convert the DSD to PCM before sending it off to the preamp?

I know this is the case for the Hugo TT2.

I believe that the DS uses DoP for all DSD playback. In DoP, DSD isn’t being converted to DSD, but it’s kind of using PCM to piggyback itself (unscathed) upon, the PCM carries the DSD information to the proper point. And the DSD info gets unloaded, totally unchanged, then processed to beautiful analog. It’s in some ways similar to the transmission of FM radio signals, where the audio took a ride on the carrier signal. @tedsmith and @Paul will probable laugh their heads off at my explanation, and hopefully give a proper explanation, but I bet they at least see I get the gist of it. And, in the case my old brain is getting it wrong, they will set me straight. But my brain remembers a video of Paul talking about DoP and the DSD. Hmmmm I can see it now….“Ted Smith and Paul McGowan star in the epic digital thriller, “DoP and the DSD”, playing at a DAC near you”. Sorry, I’ll take my twisted mind now, and go elsewhere.


Most DACs convert both PCM and DSD to a intermediate format, something like 5 bit PCM at a high sample rate and noise shaped. That isn’t PCM and it isn’t DSD. There’s a continuum of intermediate formats that can work. It’s much easier to build a 5 bit DAC than a 20 bit DAC. Then you can low pass filter the high rate analog signal to trim off the noise. There are complications if you have to do it in a single chip, but that’s for another time.

R-2R DACs can only deal with PCM so they have to convert DSD to PCM to play it (all that requires is a digital low pass filter.)

One bit DACs can only deal with DSD so they need to convert PCM to DSD. Ladder DACs and one bit DACs aren’t exactly typical.


@tedsmith that’s interesting, thanks. Are there DACs that keep DSD as DSD all the way through the conversion to analog process? Does the DirectStream DAC also convert DSD to high sample rate PCM as a matter of course?

And how does the intermediate internal (not DSD, not PCM) format compare to, for instance, converting DSD 64 to 44/88?

I’m curious if BlueSound’s contention that converting DSD to FLAC doesn’t matter because nearly all DACs convert DSD to a version of PCM anyway? Or if BlueSound is drawing inaccurate parallels?

Note in my first post I didn’t say that DSD and PCM were converted to 5 bit, high rate PCM. I said “something like 5 bit PCM at a high sample rate and noise shaped”. That noise shaping allows a better signal to noise ratio in the audio frequency band than you’d have if you just used 5 bit, high rate PCM. The DACs typically use a 5 bit (or whatever) sigma delta modulator to generate that intermediate format so one can just as reasonably argue that instead of DSD being converted to PCM that PCM is is being converted to something closer to DSD.

It may sound like just semantics, but if you are interested in the practical differences between PCM and DSD, noise shaping does matter and noise shaping PCM doesn’t give you PCM, it gives you noise shaped PCM. In that sense saying DSD and PCM get converted to PCM in most DACs simply isn’t true.

One can choose an intermediate format that can, in principle, be converted back to PCM or DSD losslessly (over the audio frequency band.) That isn’t true for 24/88.1k PCM vs DSD. You can arguably hear when DSD is converted to 24/352.8k PCM. But, I suspect that with the right up and down sampling filters DSD to 24/192 to DSD could be sonically transparent for most users.

“keep[ing] DSD as DSD all the way through the conversion to analog process” essentially means you are just using a low pass filter. The DS could do that, but I don’t want one mode that supports a volume and another that doesn’t.

The DS converts DSD and PCM to a superset of each: the sampling rate is at least the sampling rate of the input and the sample width greater than the sample width of the input. This intermediate format can be easily converted back losslessly to the original inputs. However with DSD input that upsampled format still has the noise shaping of the DSD and so isn’t strictly speaking PCM.

The overall point is that there’s a continuum of formats and that with proper handling many can act a good sonically transparent intermediate format between PCM and DSD.

IMO much of the character of various DAC chips comes from the quality of the upsampling and/or downsampling filters used not whether they use a particular format as a part of the process.


Thinking about that you SW design parts which are usually HW (and how much sense it makes if e.g. filters are that important), I wonder if in future there’s more to define by SW than today. It’s hard to imagine for a clueless like me what’s the criteria for a piece of HW to get definable by SW.

@tedsmith Thanks for the insight. Taking the DirectStream as an example. If the source material is DSD 64. Once it enters the DS and gets converted to the quasi PCM/ not quite DSD state prior to being converted back to DSD, what is the sample rate at that quasi PCM state point? Is it 24/88?

56.448MHz which is 20 times the rate of DSD64 (DSD1280 :slight_smile: )


I apologize in advance for the poor functioning of my wee pea sized brain :joy:. So that would be 24 bit/56.448 MHz instead of my hypothetical 24/88?

Which is then converted back to DSD64 X 20?

There is a rumor floating around some of the higher-end R2R DACs are actually implementing something like this to handle the DSD. They don’t use the ladder for DSD. This is why some claim “native DSD” but are still R2R. Or they are passing the converted “5 bits” through the ladder? It all gets well beyond me at that point. Signalyst

How fast are FPGAs getting with regard to reprogramming them? In my experience they are nowhere near real-time but would it provide any benefit to develop a “no volume” and “with volume” code. It gets loaded when that setting changed? That would achieve that 1-bit processing. The biggest advantage of DSD is the ability to use a simple low-pass filter but as far as I know no one does this because then, effectively, you have a “DSD only” DAC. And I’m not sure anyone wants that.

I mentioned a bit ago that an R-2R DAC cannot play DSD natively. The manufacture of the DAC under discussion is cagey in describing what their DAC actually does. It appears it converts DSD into something their DAC can play.

1 Like

All PCM and DSD inputs are upsampled to 30 bit samples at 56.448MHz. Then that’s multiplied by a 20 bit wide volume which gives 50 bits at 56.448MHz. Then that’s downconverted to quad rate DSD (1 bit at 11.2896MHz) which is then reclocked and filtered to give the output.


After the volume processing the DS is a one bit only DAC. And I do want it that way :slightly_smiling_face: My original prototype was just such a DSD only DAC. I added PCM to DSD later since I knew that most would like that better :slightly_smiling_face:


Provided their silicon can switch fast enough, they could simply map a DSD 1 to a large positive PCM value and a DSD 0 to the same magnitude but negative PCM value and send it through. That, roughly speaking, is what every single-bit converter does, even the DS DAC.

To be clear, that’s not “converting DSD to PCM”. It’s generating a positive pulse from a DSD 1 and a negative pulse from a DSD 0. After that there’s whatever analog low pass filtering the device has to offer.

1 Like

I am not following.

How would this datastream be decoded to an analog output using an R-2R ladder DAC?

As disturbing as it might be to some, apparently no DAC - R2R or otherwise - can play DSD files without doing some sort of pseudo conversion to PCM/ quasi DSD along the way. Whatever that transitional form precisely is, it’s transforming the original DSD file into something else before being converted back to DSD. From what I gather, this necessary internal conversion to a PCM/DSD hybrid variant does not degrade sound quality if done properly, but it’s a conversion away from the original DSD nonetheless. A universally applied conversion in the world of consumer DACs.

None of this bothers me personally as I scoff at the notion of maintaining original file purity as being largely a meaningless delusion to begin with - especially as it relates to the enjoyment of music.

Remember I only made the observation that no R-2R DAC can play DSD natively to counter the manufacturer’s erroneous claim their R-2R DAC could play native DSD. This is untrue.

I have no objection to intermediate formats.

This is the core voodoo that blew my mind when I first started learning about the way Ted built the DS DAC. It comes down to deeply understanding what digital sampling is and how single-bit sampling straddles the line between digital and analog.

First, DSD isn’t so much digital as it is binary, ie having two states. Numerically we represent those as 0 and 1 but those are just abstractions corresponding to two different energy levels of an analog waveform. For audio we generally find it most practical to map those onto a particular voltage but with differing polarities. So a 1 might be +2V and a 0 might be -2V. This means that you can create a DSD-to-analog converter circuit with nothing more than a switch that gives you one polarity for a 1 and the opposite polarity for a 0. That’s what you’ll find in the DS DAC albeit with a different voltage.

Making that useful for hi fidelity audio requires more work though. You need a sampling frequency that’s high enough to capture the audible frequency band, and you need some way of sidestepping the enormous quantization noise. So DSD samples at 2.8224MHz and then applies noise shaping to give us high SNR in the audio band and a whole lot of ultrasonic noise. But it’s still just switching between two voltage levels.

An R-2R ladder DAC is entirely capable of switching between two voltage levels. You can take a DSD 1, map that to some chosen multi-bit value and set your DAC circuit to output a positive voltage. And you can take a DSD 0 and map that to the same voltage but in the negative. If you can do that fast enough in your decoder and ladder, you will output a DSD signal.

In both cases you need an analog low-pass filter after that to absorb most of the ultrasonics. What flows through is your DSD-encoded audio signal.

1 Like

? No one said that. Your earlier questions talked about “nearly all DACs”, that’s by no means the same as “all DACs”

Sure there are pure DSD DACs, it’s a low pass filter, not hard to to in principle, but great sound quality takes a little work.

You can build ladder DACs to do pure PCM, tho it’s much easier with higher sample rates so your antiimaging filter can be done practically in analog. Once again great sound quality takes some work.

Most chip DACs take an intermediate path for economy’s sake.

Elk’s mistaken, if the R-2R dac can run reliably at the DSD sample rate, then only one input wire has to change with each DSD bit and all the others are left alone. You pick which wire to wiggle based on the output level you want. (There are more complicated ways as well, but they don’t have to require, for example, math.)

[dvorak expressed parts of this much better than I did.]