DAC designs

[iansr]

Ian: I think it would be great if you could bring at least the dac to Owlston, I would certainly like another chance to hear it (and also have another look inside if possible to see just whats in there).

.

I'd be happy to do that but unfortunately it seems I'm going to Owsten not Owlston

[Andrew]

Andrew: my gut feeling at the moment is that current DACs may be going in the wrong direction, with our nyquest hats on, we both know that vinyl is actually less than 14 bit resolution, has some major HF issues, and has mono bass, but that doesn't seem to stop it sounding very very good.

OK you got me here, I've always thought it was the sample freq that let CD down, not the bit depth, is that what I think your saying?

My naive view of the problem is that the harmonic info above Nyquist is lost, first by CD's 44.1kHz limit and then and/or chopped further by the brick wall filter required to get rid of the noise above 22kH. Does vinyl have much above 22kHz or is it phase related, I dunno?

Probably a case of double jeopardy, we hear both the lack of harmonics that fold down into the audio band and we hear the brick wall.

As you say we, (in the loose "collective" sense) appear to be quite happy with 14bit equivalent resolution, I will believe you on this but it sounds about right. The question is does the vinyl have a higher bandwidth? And if its not that what makes it sound better? Lack of jitter is one possibility, I suppose?

I could be talking so much tosh, of course, but what your post seems to say to me is work out what it is about vinyl that makes us happy and build a DAC that has the same characteristics.

-- Andrew

[Nick]

I could be talking so much tosh, of course, but what your post seems to say to me is work out what it is about vinyl that makes us happy and build a DAC that has the same characteristics.

That wasn't quiet the point I was getting at, more that if we applied the same metrics to vinyl that we do to a DAC, then we would assume that vinyl would sound very bad, but it doesn't, so I am saying that maybe the way we consider CD is equally invalid.

I would guess that a new record does have content above 22k, but I doubt some of my treasured records that have been played over and again have much signal related information up there.

And I know the small effect of adding a super tweeter to my OB's has been to add something, but not the difference between CD and vinyl.

I guess what I was getting at (it was late and I was sober, so don;t expect it to be entirly sensible) was maybe adding oversampling to allow better filtering is missing the goal, maybe we don't need ruler flat response up to 22k, maybe we are replacing damage in the frequency domain, with damage in the time domain. Maybe the valve amplifier I use, and the speakers I use, and my ears provide enough filtering. I can see it might be different if we are using SS amps with a DC to light bandwidth, and the metal tweeter on the end may dislike all those steps on the signal output.

I know whan I look at RIAA response in spice, I play as much attention (esp the 3.18us corner) to the phase response as the frequency one, maybe the same is worth looking at with our CDPs

But, it was just one of those open thoughts.

[Andrew]

Well, yes I agree, I think, phase occurred to me, which was why I mentioned it, but how does this all help us, are we back to the stacked TDA1543 et al?

My question is, I suppose, do we really know where the problem lies?

And, if your hypothesis is correct, do we definitively know that NOS-style solutions are the cleanest in both the combination of time and freq domain?

Here's my conundrum re the mag and all this, the NOS DAC listens well according to the reviewer (you have heard similar also ) and yet it measured 'out of the norm'. I have no doubt Noel's measurements were correct, which can only means our frame of reference is incorrect. Or what we view as the measured issues with NOS aren't really problems at all.

To paraphrase another from our midst - measurements are seldom wrong but are often mis-interpreted since we do the interpretation in the context of our current understanding.

So why the discrepancy....?

Right I need a beer, let's find a bar and talk about regulators, that's much easier.

-- Andrew

[pre65]

Hi-with regards to transports,i believe one of our members has a CDM2 in his system,was it David C ?

Perhaps he could let me know what is involved to build one please ?

[Nick]

Yes, Andrew, I think we are in the same place on this now, I agree that regulators are simpiler

I did think it was interesting in the mag review of the NOS dac just how at odds the measurements and listening was, and I admit with my SET hat on, that distinction does get my interest. And I freely admit, it was a shock just how much better (IMHO) the 1541 based dac was.

My question is, I suppose, do we really know where the problem lies?

No, I know for certain that I don't, but I just have a feeling that the problem that many of the current designs are fixing isn't the one that needs fixing the most (hope that makes sense).

[SimonC]

No, I know for certain that I don't, but I just have a feeling that the problem that many of the current designs are fixing isn't the one that needs fixing the most (hope that makes sense).

But it is one that is easily measurable (and marketable).

I've got a vague memory of Audionote producing a DAC without the brickwall filter on the grounds that it should sound better without, even though it would measure like a POS. I'll have a search around and see if I can find their reasoning as to why and the results.

[Edit] Found it, they are just using transformer coupling to do the I/V ( http://www.audionotekits.com/agrovedac.html)

Simon C

[Tony Moore]

I have to say that I was also blown away with the sound of the 1541 when I first heard the Arcam Alpha 5+ that had been modded by Leo to be NOS. (ie bypassed the digital filter chip)

I'd taken my Hawk DAC and Kit6550 over to Nottingham, initially to compare my Sowtered amp to a stock one but then things moved onto CD players and I demonstrated my CD6000OSE/Hawk DAC combination. It sounded fine. Then they connected up the Arcam and the difference was amazing. At that point I wanted NOS 1541A!

The Hawk DAC is a very good DAC too. It's at the other end of the spectrum though, CS4390 chip, 128 times oversampling. Completely different sound. Some might say "refined" I guess. In comparison the 1541 is more exciting and "in your face". I think the 4390 is actually "veiled" in something I can't quite put my finger on. It's measured performance is wonderful of course. I just don't like the sound it makes as much!

So then I built the Pedja Rogic dac based on the TDA1541A and cobbled it together with the class A discrete buffer board that was part of the Hawk DAC. I'm still not sure that was the correct thing to do but it sounds better (in my system) than the unbuffered or JFET buffered outputs of the Pedja DAC. Given time and effort I would probably want to improve the I/V and maybe add a valve buffer instead. I'm using what is labelled as an S2 Double Crown DAC but of course I have no way of knowing if it's genuine, it sounds great to me! I don't use any filtering on the output, the valve amp OPTs filter it out ok. Looking on the scope the output (with sine wave test signal) is a nice load of steps running up and down the sine wave.


I could go on and on but I've no doubt bored everyone solid by now...

[Nick]

Yes, that described what I heard exactly Tony. When I first tried the Cambridge DAC, my first thought was "Oh, thats a very early sort of CD sound, not at all refined, very in your face" .

After a few tens of minutes the "in your face" subsided, either because it warmed up, or I got used to it, urt whatever it was it started to sound rather good, more tonal colour and energy.

but the real shock was going back to the berisford, the sound was so polite flat and grey.

[iansr]

The DAC in HiFi world used 16 1543s. I suspect its very similar to the DDDAC

Pah! Only 16 DACs? Try 128 - see post 164:

http://www.diyaudio.com/forums/showthre ... genumber=4

John Westlake was the designer of the Pink Triangle Da Capo DAC that many people still seem to rate highly. I've got some time for him not least because he is very happy to support product that's more than 10 years old from a company that no longer exists, in several cases making repairs at no cost. Based on his posts he is not so full of s**t as many in the hi fi business. So when he says this forthcoming DAC is something special I can't help but be intrigued.

[Andrew]

Right, where did we get to?

[Nick]

Well, I have a DDDAC now, I need to build a better supply than the switcher its using at the moment. Then we need to have a listen to some I think. The DDDAC sound good, but I am not sure its as good as the Cambridge yet.

[Andrew]

You sly old devil you, how many stacked chips?

[Nick]

Just the 12 at the moment

[Andrew]

Be good to hear that, still hoping to get to Owston.