Edit, May 2011 :I just changed the title for its "24/192" part into "24/384", because this is of course what the Phasure NOS1 DAC turned out to be.
I also removed the question mark about 24/192 being the first NOS filterless DAC, because by now we really know it didn't exist. As it still doesn't if we count out the NOS1 (which of course is filterless at all sample rates).
Also, the topic has been moved from the Chatter board to the Phasure NOS1 DAC board.
For those who are new, well, you can try to read it all through, because it really shows the history of how someone like me started to produce D/A converters, which after over two years ended up in the very best DAC existing at this time. Not because I say so, but because by now it replaced all the great names people owned, them people usually gaining a fair net amount of money after putting their old DACs to eBay.
You may ask for your DAC to be potentially replaced whether it has (you could do it in this topic), and or I'll wait a few days for the now NOS1 owner to respond himself but will answer myself if nothing pops up, or I will respond right away if I know it's a "new" one.
Anway, of course all ended up in a DAC which uses completely uncommon technology (interface, less is more approach with hardly anything in the signal path, no switches, NON-oversampling which measures as good as OS but showing the *real* figures, in-software filtering (e.g. by means of XXHighEnd), really zero in-DAC filtering, special differential setup of the 8x PCM1704U-K chips with better figures than even the chips themselves, super low noise and jitter levels) ... all from an external DAC.
But maybe it is more efficient to start reading somewhere at the back of this topic.
Lastly for here in this edited first post, maybe it's nice so see some new posts in this topic, because it has been a great time throughout creating it with many of us, still having the nice memories about it. Or maybe I will add some newbies to the topic myself later ...
Regards,
Peter
Guys,
I must be honest with you. I have been working on other things than software lately ...
See below, which as far as I know is the first 24 bit 192KHz DAC explicitly assembled for non oversampling and filterless mode.
36 hours ago the champagne was opened because it produced sound for the first time.
Some technical insight for those interested :
- Balanced passive I/V conversion by means of two balanced DACs (PCM1704) per channel as the base;
- Balanced or RCA out;
- Output 1.5 VRMS (2 VRMS possible);
- Possibility for oversampling to 211KHz with 1.2ps RMS total jitter;
- Possibility for filtering (2 pole Bessel 100KHz);
- Drives long interlinks sufficiently to avoid a pre-amp or other means of buffer;
- S/PDIF connected, and due : directly Firewire connected without soundcard into I2S;
- Terminals for an external clock;
- The most stable shunt regulated power supply imagineable, both channels completely separated throuhgout;
- Everything operating in pure Class A;
- Housing : WxDxH : 43 x 26 x 8.5 cm = 17.2 x 10.4 x 3.4 " worth of 15Kgs weight;
- The various elements not designed by me, but chosen as *the* combination for the best DAC ever (so 1 % credit for me really
);
- Name I dedicated to this combination : NOS1.
You may ask yourself : Can I build this too ?
If you are like me, forget it. I drove various people to complete madness and received some burnouts myself.
Anyway, it took me over a year thinking about it and "designing" various concepts, 7 weeks throughput of ordering modules and parts, three weeks of understanding, building and getting more parts, and occupuying the dinner table. The (programming of the) Firewire connection still to do.
BackgroundI didn't start this project well over a year ago to perceive better sound, and I was very happy with my current NOS 18 bit 96KHz DAC (the TwinDAC+). However, since the hires material is on the horizon closer and closer, including the *fact* that a pre-amp should be avoided encouraging for digital volume control - that by itself taking out bits, I got obessed of wanting more. Technically more.
Being a convict of non oversampling filterless DACs, I ran into a problem : 24 bit 192KHz DACs do not exist.
Many months of thinking and designing brought me the concept of a 32 bit 384KHz DAC completely software realtime controlled (yep, the firmware part), of which the hardware arrived May 2008. But after I squeezed out the first sines from it, I coulnd't find enough time for it.
While I started this off thinking 32 bits would be needed for a proper digital volume control, underway I created the digital volume for XXHighEnd which very sufficiently operates at 18 bits or even 16 when the output voltage of the DAC isn't too high and the gain of the amplifier isn't too much. This brought me to again searching the internet and it came to me that the PCM1704 should theoretically be able to do the job, but nobody just made something out of it. Or ?
For over a month I had bookmarked a company which expressed specs that looked promising but kind of vague in the mean time. I kept on coming back to this because there wasn't anything else and in the end I started to send an email to that company ...
Ok Peter, why don't you mention the name of that company ?This is a strange story, and possibly I can't even justify it 100%, but I have the feeling that right now this company shouldn't receive dozens of emails from you guys asking away because they just can't bear it. Maybe later, but not right now. This is all my fault and prices would double instantly ...
People who know me, will know that the occasions I asked someone for help in my professional life can be counted on one hand. Until one month ago that was. Right now, some email boxes have been overflown, just because I started some DIY project with modules provided as kits and a withgoing FAQ that says all is for skilled people only. Hmm ...
Add to that this isn't about a "DAC kit" as such, but a couple of modules which makes it a DAC. Add to *that* again that I asked for modifications which were politely met, but that these functional modifications ended up in technical stuff beyond my knowledge, and no existing manual could explain to me how to deal with things. Just being honest here ...
So even if you know what company this is about, please don't go there right now, and if you indeed know, you also know this is related to respect. I'm sure you will understand in a later stage ...
First listening impressions (with 0.9w-3 and which is theoratically not the best (thing to do))
Since rather many combinations of settings exist, all with their own merits, right now it is rather hard to determine what is the best setting. But no matter what setting I tried ... it is one
big step ahead in audio playback. To mention a few things I definitely recognize (just 16/44K1) :
- Enourmeously increased resolution.
This is unrelated to the higher samplerate (because I'm just not using that right now) and also this is unrelated to the nos/filterless principle;
It is related to the enormeous stability of the PSU, which, mind you, must deal with such small voltages (.0001 etc. mV). And, knowing that creating 1V of output here easily degrades the .0001mV output there, you will know what this is actually about, hence what I was after.
Also, of course I know how accurately things were trimmed by me, but more importantly : how accurately they *can* be trimmed by design.
Cymbals now receive an additional dimension, which I generally express as : now you can see the size of the cymbals much better. Also, compared to reality (live cymbals) it comes creapily close now.
- A bass response of which I didn't think it was possible.
For the first time in my life I "heard" a powerful bass response as how we all perceive the large heavy amplifier being able to produce good bass response. We all know that the latter is non-sense, since the class D and other 100g chipamp stuff can do it just the same. What I did not know is how a very good PSU for a DAC incurs for the same but in exponential form. But it can well be that it is the other way around from what I sad right above : the small voltage spikes needed for the higher frequencies won't tear down the more current eating bass waves (never mind this all occurs at micro levels). That the I/V (current to voltage) conversion occurs in a high current domain is another thing that for sure contributes.
As everyone I use a bunch of test tracks and throughout the subwoofers speak where they did not before. This is the most occurring, because - apart from XXHighEnd anomalies as some may recall - usually this comes together with a more profound "just low bass" region (say 40-200 Hz) while right now there is no spur of this. Oh, the bass is more profound allright, but it doesn't incur for any colouration or the tempt to change the cross over. It is just all good good good.
- Compared to hires it is now an apples with apples comparison.
Since this DAC allows for NOS/Filterless at the higher resolutions, I now can compare 16/44K1 with 24/96 and 24/192 without having to hop over to a oversampling sigma delta DAC.
I sure did not recognize any different nature at the higher resolutions (tried 192 as well), but I also did not perceive it as better. This latter is not meant to be related to the DAC but merely is a confirmation of what I already though before but couldn't prove (apples and oranges thing) : redbook (hence 16/44K1) seems to suffice and any higher samplerate plus bitdepth doesn't bring more to our brains. Something for another discussion !
- There's an unary "experience" listening to my speakers now, and I don't know how to put it into words yet.
It is not "black" as we often hear or experience, but in that area. An "emptyness" maybe. A being swallowed by the music, you being in that black hole. It's not a negative, but a strange one. Again things are happening in mid air (I talk about this more often), but now they more happen in 3D while not being able to point at it.
It sprung to my mind that bass is more directive as ever (I use subwoofers in stereo setup) and besides that the veils from drums suck on you. The latter is again not a negative, but at trying to express what I feel, this has to be in the equation somehow. Hmm ... it could well be that absolute phase is wrong (just change it in XXHighEnd of course), which btw *is* very well possible because of the various taps which can be used, and each "stage" inverts the absolute phase.
Anyway, it looks that the "boost" of music is more profound now, where boost becomes inverted boost (suck) when indeed the absolute phase is wrong.
- Very good micro detail at the larger instruments.
This may sound strange and also looks similar to the beforementioned higher resolution, but I am fairly sure this is about the inheritent speed of the DAC (about slew rates and such, possibly even feedback 
).
Under "increased resolution" I talked about the other dimension in cymbals, and here I talk about yet another dimension : the attack of cymbals. Ok, maybe you must have the experience of holding a drum stick a couple of times, but when a drum stick hits a cymbal this doesn't need to be one surface touch. I mean : when a ride cymbal is "ridden" this *is* one touch only, but when a smash cymbal is hit, before the stick goes away the cymbal hits the stick again. It is this which just is audible now.
Brushes, even more delicate and far worse to express properly as I found lately, can't avoid the "feedback" of a cymbal and snare (or tom). That too is audible now, though not exactly in the leage of the "larger instruments". I mean this opposite to the more commonly known detail like breath in a flute or sax. So it is not about this.
Similar IMO is the wood of a bass becoming audible. I hear you say "man, I heard this long time". But no, I don't think you did. I know my abouts, and this is in another leage. This is about jitter (in good combination with the other virtues of this DAC);
I don't know whether it is a world record, but the overall jitter of 1.2ps seems to incur for what I just described. Why do I think that ?
Because this low jitter only can be achieved in oversampling mode of the DAC and it is there where this is happening. This is *not* because of the oversampling itself hence impeeding higher "resolution", because at least I claim to know that oversampling only degrades. And for sure it won't "create" the beauties I hear here. If that would be so, XXHighEnd could do it just the same, and it can't.
All 'n all this subject is not really honest, because it uses oversampling mode which net I don't like. It colours the music, and all starts to sound similar because of it. Nothing different from what I was used to.
But for the more or less solution, the direct Firewire connection comes into play, because that will incur for less jitter because of the I2S connection beind it. And to get the real merits of this : at non oversampling mode the DAC depends on the clock of the incoming data. At oversampling mode this is just unrelated ... 
Might you want to know : none of the modules I used, emerged from more or less known DIY projects as known from e.g. DIYAudio. I can tell you that each single part was designed from highly qualified engineers and that the designs themselves testify high quality engineering. Not that "DIY engineering" wouldn't be any good or the engineers being out there are not trustworthy for their profession, but what I have here contains all the consistent knowledge of the most qualified people imagineable, for whom I make a deep bow.
This is not a matter of simple is less, and looking at the picture below I can tell you that the DAC (ok, SMD) comprises 1/25th of the total PCB surface. 24/25 comprises of PSU stuff and getting the output current from the DAC into the output terminals.
Oh, I didn't count all the switches needed for the various input and output settings.
Show Posts
better don't do it !
