[MLGrado]

not the ZEN DAC, but I had to also take advantage of the new product updates to remove the ZEN Phono from my heavily modified Rega RP3 setup and try out the iFi ZEN Phono V3... cannot wait....

 

[sunjam]

That thread you linked gives me a headache. lol. It inspired me to start writing an easy, fairly concise summary of the 'pure DSD' Dacs, how they ALL work around Thermometer/unary coding/dynamic element matching on my blog at euphonicreview.com. This has just been an idea in my mind but looks like this post may be a start.

The Signalyst DSC is an excellent proof of concept of 'true' 'pure' 'native' 'insert adjective of choice here' DSD DAC, but it's the same technique used (of course with minor variations with different makers) by Burr-Brown, T-A, any AKM DAC that actually has the bypass option available as a choice.

ROHM (a new player on the hi-fi scene as best I can tell. Not new to the IC market. My SMSL D300 is powered by their top chip, and I love it.) They offer no choice how you convert DSD. If you put in PCM, it is converted just like PCM typically is in a Delta Sigma DAC. If you send DSD, the ONLY way it will process it is native DSD via its analog FIR filter. There is no possibility for any DSD DSP. (I love it lol. To be fair, when it comes the DSD conversion Burr-Brown/TI chips work in similar manner)

Holo Audio is essentially the same thing although they use marketing jargon and call their DSD conversion a 'ladder' DAC. In TRUTH, ALL DSD FIR DACs that use resistors have a so called 'ladder' of usually equal weight elements. What confuses/interests me about Holo is how they use a redundant 'ladder' for each filter. I have some educated guesses how that might work, and at the same time I have actually no idea! But I do know that depending on the version, its either an 8tap (9level) or a 16 tap (17 level) FIR moving average filter.

Denafrips uses marketing to muck things up too.... calling it a 6 bit DSD converter?? No its not a mulit-bit delta sigma converter. Its a PURE DSD Direct analog FIR filter with 32taps/33 levels. By MY math, it could technically convert, if you used parallel in parallel out registers just barely over 5 bits.. by 1 tiny level. I guess they like to round up... WAY up lol. Or maybe there is another explanation. I have asked many times to no avail. But as it actually is, its a 1-bit converter, with serial in parallel out shift registers just like the pure DSD DACS mentioned above.

And yes, there are others out there. All the DACs of the above types that I have had the change to evaluate, which is most, sound truly OUTSTANDING with HQPlayer. It's the best of both worlds. The simplest, cleanest Digital to Analog Conversion out there, with the best digital DSP you could have, IMO. This is especially true with the lesser expensive DACS. Realistically, no one in their right mind would have a PC over 1000 bucks, a nearly 400 dollar piece of software to maximize out a 250 dollar DAC lol. BUT, IF YOU DO HAPPEN TO HAVE THE CHANCE TO TRY IT, you would be amazed at the sound you would get. One would swear they were listening to a multi-thousand dollar DAC in a blind-test. And yeah, in a sense they would be lol. Diminishing returns would be found as you get better and better DAC's though. I have not tried it yet, but I don't expect HQPlayer running my iFi IDSD PRO at DSD512 or DSD1024 is going to sound that much better than the already outstanding DSP on FPGA custom made by iFi. Still, these kind of pure DSD DACs can be and are audibly some of the most impressive, especially at the budget level when given the source to shine.


As far as DSD DACS that don't belong with those above.


ESS--YES the actual DAC section is an unary coded 64 element trimmed resistor FIR filter per channel (what many still consider 1-bit conversion, because unary code can be seen as multiple 1-bit streams), but the hyperstream DSM works at 6 binary bits. Getting ahead of ourselves though. back to where things begin in the DAC, not end...

Way back Before all that, ALL DSD first goes to a specific DSP that converts DSD to a multi-bit intermediate via FIR filter.(* see ESS diagram attached)

However many bits comes out of the FIR filter, is multiplied by a 32 binary bit gain control. It would be easy to think if one doesn't use the volume control, the entire DSP of filter and gain control is bypassed. This is incorrect. The FIR filter is the first necessary step to go from a bitstream to something multi-bit. You can go all the way back to Sony's white papers on DSD-Wide, and see this is EXACTLY how they accomplish the same thing. Nothing new under the sun!

Of course the FIR filter will leave redundant samples as in ANY type of filter, although they do NOT remove the redundancies, keeping the sample rate at the same DSD rate that came in. Thats how DSD can go through a ASRC for jitter control as well. Oh, and before the ASRC, there is another more specific IIR filtering stage for noise control before the noise shaper.

Yes, filtering and 'redundancy' in samples happens in a Pure DSD DAC too, but at the DAC side it only happens ONCE (twice if you count the completely different analog I-V or RC filter that will follow). But it happens in the actual digital to analog conversion itself (the analog FIR output filter) and doesn't have the massive amounts of DSP and remodulations attached to it.

Any AKM NOT specifically in DSD DIRECT or BYPASS mode. See ESS above for how it works lol. In the most basic sense, that is. It has its own nuances and functions that lead to the same thing.

Chord... his FPGA DAC isn't Pure DSD in the sense we have defined already with the Signalyst DSC DAC as the example par excellence. Its more like the ESS types, or actually a lot like the PSAudio DirectStream, except the output stage of the PSAUDIO Direct Stream IS a true 128x DSD 1-bit modulator with analog filter.


And there are many others.. and I don't want to knock these things. I am just saying they are not pure native DSD conversion in its simplest form.

Some of them may have BETTER sounding results with their ingenious DSP. Like Mola-Mola which converts all to a PCM rate over 3 megahertz I think for processing, then the final DAC is a true 1 bit Pulse Width Modulator with noise shaping converted via an analog filter with further smoothing accomplished with its I-V converter. I would LOVE to hear one of these. The implementation sounds amazing.

Another GREAT one is DCS and their 'ring' DAC.

Its actually a lot like the ESS chip in how it works. ALL data PCM and DSD is processed by a 5 binary bit delta sigma modulator, that then is (still) on FPGA converted from binary to unary code, uses dynamic element matching to something like 48 elements? More elements than necessary, but all the better for linearity. (yeah, I forgot to mention that ESS in their output stage uses similar dynamic element matching and calls it a revolver DAC)


Now for those who will misunderstand me and get me wrong... I consider all these examples to be actual, real DSD DACS. With different ideas on how to best do conversion.


They are NOT in the category of what I would consider NON-DSD DACS that still accept DSD. Now THAT third if you will category of DSD DAC are non-native PERIOD.

DACS like this tend to be R2R DACs that do a full filtering and removal of all redundant samples, to something like 24bit 176.4khz for direct non-oversampling conversion with their R2R ladder. In other words decimating DACs.

(well, this is what most people think of when they hear decimation, although I could have used that word earlier in this post and still have been correct, but people would not have understood what I indeed meant and a fanboy argument would persist.)

What kind of DACs do this? I know there a few but none are coming immediately to mind that accept DSD, display DSD, yet convert immediately to a standard PCM rate and never look back.


Anyway... there is a start to the next part of my blog series... sorry I fried my brain a bit.. .which was my original complaint about this subject in the first place lol!!!

DSD Direct Discussion Thread

"That thread you linked gives me a headache. lol" <== Yes, I felt the same too when I joined the thread around late March initially. I think Roon's discussion forum system is pretty bad for people to read / follow old posts as the posts are badly organized/structured.

However, I felt better once you joined the discussion for a few days. I could follow new posts easily as it appeared. I asked some stupid questions to start with and the people there is so nice in explaining a lot of things to me, especially, Jussi Laako, the developer of HQPlayer. I learned a lot related DAC internal, DAC details, digital filtering, over- / up- sampling, etc. from that thread (as I ran off-topic from time to time in order to learn more). I need to go back to that thread from time to time in order to keep track of posts there, otherwise, I would be at lost again.

In my mind, Jussi is one of the best DAC experts in the world as he did a lot of tests / measurements of different DACs. The DAC you mentioned earlier, The Signalyst DSC, was desinged by him.

In fact, I was looking at a DSD Direct DAC iinitially for DSD1024/DSD512 (e.g. E30 II, and DIY DAC based on 4499EQ) . I ended up got the iFi Zen DAC v2 based on his (and other pro uses') recommendation. It was one of the great buy for me (the other one I did was the purchase of Fosi V3 class D amp).

I will always go back to him for anything related to DAC (and even stuffs related to the Sampling thoery used in Digita Audio). He is always there in the thread I mentioned earlier. No sure if he could help you for your writing but he is always there for me for any DAC related questions.

(BTW, here may be a nice point to enter if you are interested to try again with the thread regarding "questionable AKM DSD Direct mode")

Native DSD = DSD Direct?
It looks to me there are a lot of confusion regarding the term Native DSD. I suggest in your official writing, you could consider to help people to clearify the term "Native DSD" when they see it on any marketing materisls.

Here is my understanding of these terms are (in the companies’ marketing sense) (please let me know if these are not correct)

1. Native DSD support: means DAC can take DSD data stream natively (i.e. raw DSD data stream) for processing. It has nothing to do with how the DAC handles the DSD data stream internally. This usually requires ASIO driver for Windows and Mac does not support it
2. DoP (aka DSD over PCM) support: means DSD data stream is packed in PCM packets for sending to the DAC (as all the DACs support PCM flow). As it requires larger PCM packets for sending the DSD data, it usually supports lower DSD bitrate under DoP. i.e. DACs support DSD512 (DSD Native) may only support DSD256 (DoP). Moreover, I think most people / companies consider DoP as not “DSD Native” (even it can transfer bit perfect DSD data to the DAC). Again, it has nothing to do with how the DAC handles the DSD data stream internally in the D-to-A process. Windows can support it without any driver installed.
3. DSD Direct, Pure DSD (or something like that): means the DAC would by pass all the DAC’s internal oversampling / Delta-Sigma modulation / etc and would only use very simple / direct way to convert the 1-bit DSD datastream to final analog signal. It could happen or not with the “Native DSD” or “DoP” flow (i.e. non “Native DSD” flow) depends on if the DAC support this DSD Direct option.

Some chips do not support DSD Direct option at all (e.g. ESS)
Some chips only support DSD Direct (e.g. BB’s DSD1793)
Some chips make it optional (e.g. AKM). It depends on the indivdual maker / DAC design

Mis-leading information regarding DACs

During my journey in finding the best DAC for my HQPlayer / DSD upsampling setup, I came across a lot of pseudo science claims from various so-called "Audio Science" discussion forum. The worst one is, no doubt, ASR.

One of these pseudo science claim is:

"if different DACs are measured audibly transparent, it would sound the same"
"if you hear any difference, it is either your brain is fooling you or you are using a poor reconstructed filter that cause it"

Here, they have a very different definition of "audibly transparent". Not the one commonly used in the Audiophiles world.

In my eyes, this pseudo science claim is killing the DAC market for the Hi Fi industry. In the end, if the general public truely believe what they claim, the market would end up with all copy-cat mass produced standard chip-based DACs.

Mis-leading information regarding "Hi-Res is useless"
Another pseduo science claim is that "Hi-Res is useless". The people in the "science based' discussion forum believe this claim in a "religious way" IMO rather than based on factual scientific research.

On these forums, you would hear from the senior members, who seems know audio sciecne well, attempt to brainwash you with comments like "you SHOULD not hear any difference between Hi-Res and CD, if you do it means you don't trust science" or
"science proved that you cannot hear it" or
"science proved that Hi-Res is useless"...

LOL... I attempted to argue with them with scientific facts... In the end, they just kicked me out and deleted all my postings there. LOL....
In my eyes, they are using this pseduo science to kill the Hi-Res market. I understand some people may not be able to hear the difference but why they HAVE TO kill the Hi-Res market and stop other people from enjoying better music? Really, I don't get it.

Worth to mention these mis-leading info while you are talking about DACs?
To me, I feel the obligation to debunk these pseduo science claim (as my background seems to be able to help).

I decided to start my own blog (on my signature) and try my best to debunk all these stupid pseudo science claim. "With great power comes great responsibility?" LOL... I am just a ordinary guy

I attempted to join/create various discussions on supposed-to-be "audio science" based discssion forum in order to do so....

I did that on ASR. Failed.... I was banned totally and they removed all my postings there
I did that on Roon forum. Failed..... I was banned temporarily for a month. My posting there was hidden but still accessible there via direct link.
I am doing it here, on Head-Fi...... I encounter a lot of opposing pressure but the good thing is that the thread is still active. Feel free to join and comment there.

Not sure if you are interested in "got involved" in these topics, i.e. "different DACs sounds the same", "Hi-Res is useless". IMO, a few words on your official writing would have a great help on the "dying?" DAC and Hi-Res market.

Worth to mentioned while you are talking about DACs?

"Realistically, no one in their right mind would have a PC over 1000 bucks, a nearly 400 dollar piece of software to maximize out a 250 dollar DAC lol"

In my mind, I would re-writie it like this "Realistically, no one in their right mind would have a PC over 1000 bucks, a nearly 400 dollar piece of software to maximize out a 250 dollar DAC. They could save the money for better amp and speakers as they don't have to worry about the source anymore. lol"

(Sorry for my long post. It may be a bit off topic but my expereince and recent journey in the audio science world help me to make a wise decision IMO, i.e. I end up got the iFi Zen DAC v2 for my upsampled DSD music. I may upgrade to v3 later. Cheers)