empirical audio or lynx aes 16 sound card

You can run a long AES/EBU cable from either of these. Long cables are not exactly a great thing but the higher voltage of AES/EBU is the better way to go - even better than a long toslink.

I found the Empirical Offramp 3 with Superclock 4 to sound a touch better than the Lynx aes16. The Lynx sounding a little flatter, the Offramp a little more vivid.

That must be a very good DAC. But a claim to be completely immune to jitter is a very big one, and one that has many times before been made, only to prove to be untrue.

I would be very interested to hear more about the so-called 2-D jitter reduction. I read what is said on their site and it is pretty obtuse, but you get that with web sites - it doesn't mean they are not onto something.

Reading between the lines, they seem to be using the same jitter reduction process as used in a number of DACs, but maybe with a twist. I am guessing they are doing the over-sampling trick used by Bel Canto and Benchmark, where you upsample to a very high rate and then downsample just before the DAC chip. While this more or less eliminates jitter it simply maps jitter to broadband noise - ie jitter turns up as bit errors. The sound is at first impressive but ultimately sounds unnatural, particularly timbres of real instruments.

The process can also add errors due to any imprecision at the very high clock rates used. What they appear to be hinting at is that they do the same thing as say the Benchmark but then fix the bit errors. This is conceptually possible since the original jittery signal can be used as a reference, since it is likely to be bit perfect. But there is a "so what" here. Eliminating jitter is theoretically possible without using this particular strategy. There are many such strategies, and using ethernet for bit transport is probably the most promising of them. The tough thing is implementing it in real time, and the ethernet products we know like the Logitech products are not at all perfect. So far, I don't think anyone has done it, including Wavelength with their asynchronous USB, but it would be a massive breakthrough if someone did.

I wouldn't throw around the "must be a 2-D DAC" mantra too much. All they are saying is the DAC should attempt to eliminate jitter - that is all they mean by 2-D. Therefore there are a lot of 2-D DACs out there. Their implication that no other DAC attempts to deal with timing issues is not a fair comment. The only interesting thing, is their implementation - if it does indeed work as completely as they claim.

From your comments Big-amp, they have had some success. Experience to date makes me remain a skeptic about the claims of complete elimination of jitter, for the moment at least - but I now want to listen to one of these DACs, if I can get my hands on one at this end of the world.

I think everyone is in agreement that x-axis, or timing, plagues digital audio, and that it ought to be able to be dealt with. Getting bit perfect transport of bits is very much achievable and so regenerating the clock data ought to solve the problem better than expecting perfect cabling and interfaces.

The key issues in digital audio to resolve are:
1. How do you eliminate jitter just before, or in the DAC chip itself, to avoid jitter creeping back in subsequent transmission steps. Ethernet makes the most sense to me, because of how it works. Its just that not many audiophile firms have the knowledge to exploit ethernet so they faff about with USB, SPDIF, AES/EBU etc.
2. How high does the sampling rate need to be to make quantisation error immaterial?
3. How do you deal with digital filtering at half the sampling rate. Use DSD? Increase the original sampling rate? Upsample before conversion? Use a filter, not use a filter? If you filter, then what sort of filter?

It is pretty clear to me that the Playback guys mean dealing with these three issues as being 2-D, with the first issue being x-axis and the second and third being y-axis. Where they are claiming to be different is in the x-axis, ie in eliminating jitter. A claim many have made before them.

The only thing they are being clear about is that you should do it in the DAC, not separate from it. I think they are saying they are mapping jitter to change the bits, which isn't new, except that it is usually done in a separate stage to the DA conversion. It is not clear to me why doing it in the DA conversion step is such a good thing as it is generally better to keep the DA conversion step as simple as possible.

But until the Playback guys tell us a bit more about what they are actually doing this is just guesswork.

I agree Blindjim. None of these interfaces eliminate all of the jitter created upstream from them, in my experience.

Airport express does sound good Lenny_pt and especially for the price. What can make it sound better is something like the Pace Car in mode 3 with the Airport Express modded and slaved to it. But the Pace Car and the modding process is expensive unfortunately.

Blindjim, it would be nice if you were right, that a DAC can eliminate jitter. I haven't found one that is not affected by changes in jitter performance. I also don't see improving jitter performance as just changing flavour. I have generally found that improved jitter performance improves everything - smoother, tighter, more resolution, less glare, better dynamics.