XXHighEnd

In my quest to find an affordable NOS dac that would let me experience the virtues of QAP, I've been doing quite a bit of reading. For me this is dangerous cause I'm no scientist, engineer, computer programmer or even a true audiophool who is experienced in identifying the "best" sound. I do love music and, as I've proceeded past my zenith, I've come to appreciate more kinds of music, especially classical, than ever before.

XXhighend has been a revelation for me regarding what good sound "is", and I have totally accepted that a NOS dac is the only way to get to the very best that xxhighend has to offer. However, the other day I ran across this thread at Head-fi: http://www.head-fi.org/forums/f7/nos-dac-marketing-bs-438220/index4.html  and now, after reading Dan Lavry's responses in the thread I am totally confused........

I'm sure that the upsampling dacs he talks about are multi-bit and not just delta-sigma. What's the difference and if they are so good why wouldn't Peter be develpoing a multibit dac instead of a NOS.

HELP.......(in laymens terms if possible.....)

Haha, I have been following that thread closely when it was live, and apart from the "shouting" mr. Lavry likes to exhibit - he makes a few mistakes (says me !). But mind you, what he states is nothing far from accepted knowledge, so it is just me who (was ignorant and) changes things a bit. Plus to keep in mind : numerous people like NOS better ... why ?

In brief :
Those who believe normal filtering has been proved to be scientifically correct, take the ringing which goes along with that for granted. "It can't go otherwise anyway". Mind this latter ...
So, this is not what I do at applying the (sure necessary !) filtering, and while thus the filtering is applied (my own way) it doesn't ring a single sample.

Edit : Besides this, normal filtering - and the measuring which would prove how good it is - takes a same sequence of one frequency (or two) in order to measure the result. The result shows good just because of the repetitive character. This is not how "music" behaves, and the way music behaves can NOT be tackled by that common filtering (let Lavry prove that it works there too :)).

The above may not me totally clear because it is not 100% related to that thread, and I guess it also needs to read his "paper" on this (57 pages IIRC) in order to understand his (and everybopy's) beliefs.

But also don't forget this :
The beginning of that thread (when Lavry chimes in) is about "are you 100% sure your DAC isn't oversampling ?" ... and of course while my Phasure NOS1 100% sure is not, it is now the software doing it. Without that (or any other means of filtering) harmonic distortion is way high.

All 'n all -even without my filtering- I *still* would definitely vote for NOS/Filterless - even at knowing how bad it measures. It sure doesn't sound as good compared to what we have today (Arc Prediction) but it still sounds way better than heavy oversampling and the filtering going along with that.
Of course these are my words, but I don't think they are subjective and I hope you trust me a bit on this (by now). Oh yes, it may carry a more black background, but it just doesn't work for musicality and instruments can't be recognized. NOS/(really)Filterless does just that, and I guess this completely overrules the heavy HD coming from it (without filter).

I'm afraid I can't be much clear when I am this brief, but I hope it does a little good that I knew about that thread, which btw has been discussed in other places as well.

Peter

Firstly, you need to understand that all NOS DACs are multibit. But not all multibit DACs are NOS.

Oversampling was a method employed en-mass in the late 80s and early 90s (no delta-sigmas, only multibits then) to help the filter do its job more effectively. And it works... if measurements are what turn you on. (Dan Lavry is obviously a fan of measurements, which I think is a good thing.)

However, a multibit DAC does not need oversampling... But IMHO, it only makes sense to go non-oversampling if the sample rate of the material is high enough (in my belief, at least 176.4KHz, which is 4 x the sample rate of redbook CD, or 4fs).

Now, there isn't much 4fs material. But that's OK, because with Quad Arc Prediction engaged, all of your ripped CDs become 4fs (and 24bit)! Now all you need is a NOS DAC on which to play them.

And there's the rub - there aren't many 24/176.4K NOS DACs available. But Peter's NOS1 (I believe) will be a NOS 24/352.8K DAC. To use it effectively, Peter will introduce 8 x Arc Prediction for you to use with your CD rips and 16/44.1K downloads.

In short, Peter is creating what I believe is the world's first NOS 24/352.8K DAC.

Mani.

PS I've never even met Peter, so I hope most of what I've written is correct!

Thank you Mani. You explained that a whole lot better to the point than I did. I also skipped the multi bit thing. Thanks.


PS: The quad upsampling measures as good as the normal (say, D.Lavry method); Double does not (it's just not sufficiently enough).
PPS: Yes, the NOS1 will do 24/384 but it's theory so far (read : I don't have that myself yet -> waiting for the new PCBs needed for getting 384 out of the PC ...).

Given how great xx is I have no reason to doubt you guys. Let me ask this. If one's dac budget cannot exceede $1,500 US what would you recommend: an upsampling dac like the Lavry DA 11 or benchmark, or one of the asian NOS like the valab or the MHDT Havana? Just looking for opinions. Thanks.

Sorry, I have no idea what to go for around the $1500 mark.


I'm not so sure that I would... if it's just 16/44.1 capable.

You know, there are some 8x oversampling DACs with PCM63K or PCM1704U-K chips that sound great. Quite honestly, if it weren't for computer audio, I'd be quite happy to stick with my made-in-1995 Rotel RHCD-10 CD player...

Mani.

I think this is worth stating again:

This is what I find so exciting. The best sound and measured performance is with oversampling... BUT done right. And this is exactly what QAP gives (and OAP will give) us.

One of the biggest issues with 'old' DACs like mine is that they just didn't have the computing power back then to do oversampling/filtering as well as we can now. But if this is now done in the software, and the inferior oversampling/filtering in the DAC can be defeated, then such DACs are given a new lease of life.

In a way, this is a bit like the original McLaren F1 road car. Put it on a track against a 'modern' supercar, and it will struggle (see Martin Brundle's video from a few years ago). However, replace its brakes with modern ceramic ones... and I know which car I'd take.

Mani.

That's all it would take, Brakes? Wow.

Ho ho ho ! (yea, a bit late) ... Never touch the car of an englishman !
:)

Peter, when is your NOS1 DAC going to be available for sale?  I hope you are on your way to CES2010 to unveil this new beast....   :)

Mani/Peter, given that the computing power now exists today to do the oversampling in hardware integrated with the DAC why would one expect to achieve better sound quality by doing it outboard of the DAC and using a NOS DAC? 

Well, that's the problem with analogies - you open yourself up to being shot down :( And with our decimated car industry, maybe I should keep the car references down to a minimum. :( :(

Back OT, when is the NOS1 going to be available? I think you said it won't be until February at the earliest, right? I can't wait to take this 'Dutch Koenigsegg' on with my 'ceramiced-McLaren'.

Mani.

Well, there are a lot of components that need to be gotten right - the power supply, the oversampling/filtering, the use of true multibit DAC chips, a bomb-proof output stage, etc, etc. There is no 'integrated' DAC that I know of that hits the mark, regardless of price. Certainly, no AD/DA converter like the Model Two will ever be built again - it'd be way, way too expensive.

Mani.

It would be nice if you could hear the ULN-8 in your system and compare it head to head with your Model 2.  You'd need a Mac to audition it, however.  I know one guy, however, who has a couple of Model 2's and he's heard the ULN-8 and he just says it sounds different, not better.  He also said he has yet to hear a DAC with an async interface that sounded good.  Note, supposedly the ULN-8's firewire interface is not async according to the primary USA dealer.  It will be interesting to see what comes out of CES2010.  Did you see the teaser from Chris on the Computer Audiophile forum?

Interesting. But I bet he hasn't heard the Model Two with QAP. I really can't stress enough how much better it sounds when set to 176.4K and fed from XXHE with QAP.

The ULN-8 uses delta-sigma chips [EDIT: the AK4395 with 128x oversampling]... that's a bit like buying a front-engined/front-wheel-drive 'sports car'... which is an oxymoron IMHO. Yes, they can be very good, but never ever as satisfying to drive as mid-engined/rear-wheel-drive cars (or indeed the Teutonic engines-hanging-out-the-back/rear-wheel-drive cars :prankster:).


Yes, totally nasty on Chris's part. I'm dumbfounded as to what it could be...

Mani.

Flexibility and asynchronous power. So, the second version of Arc Prediction is on its way, and if you would have had the DAC and it would have been internal, no way upgrades would exist for that, or it would be relatively (very) difficult (which is what I planned originally but it takes additional hardware and complexities).
The power emerges from the fact it can be done just semi real time (as it is done now, in the preprocessing stage).

February is the plan, but last week I learned that the transformers I need :whistle: will be produced in February ... :fool:
But at least something improved again (which I didn't tell about yet).

Thank you for repeating what i was trying to explain on computer asylum from about one year ago :)
The problem was that until now it lacked a software able to do that seamlessly and in realtime.

I think it is the second DXD filterless DAC, the DAD can work filterless only at DXD sampling rate (which now i dont remember if it is 352.8 or 384k).
It is for sure the first filterless DAC at 24/176,4k

I know nothing about the AX24, other than it uses multibit delta-sigma chips (as does dCS). I didn't realise that you could use these without oversampling/filtering...

Mani.

Which - if it is delta-sigma - I won't believe much ... :)