ItemAudio

Top tip of the day: buy a 20cm USB injector cable (they're on eBay for around $/£5) and a USB A/B gender changer (a bit cheaper); plug them together and have it cryo'd: it will be by far the best cable a few dollars/quid can buy, and opens up the world of USB power supplies and batteries for whatever DAC or SPDIF converter lives on the end of your cable. We do cryo lots of digital stuff (including whole motherboards), and have made USB cables. The former works brilliantly (when it doesn't kill it); the latter, for us, not so well.

Sorry! The straight dope on that question turns out not to be very straight . . . it's really tough to generalise in a way that would be useful to you, or accurate. You won't go wrong by limiting your DAC search to those with top-notch USB implementations. Anything that uses an XMOS driver/chipset is a good start. Supporting Class 2.0 on Linux or Mac is also a good sign (this will usually require a driver for Windows). There's no reason not to be choosy in 2013, and select only candidates that offer async 24/192: it's not a guarantee of USB excellence, but it's not a bad indicator.

It's a bit of myth - based on a misunderstanding of Intel's published 'TDP' figures - that the Atom is meaningfully more energy-efficient than other processors. We spent quite a lot of time in the last year measuring real-world draw on a number of different boards, and found that N2600-, N2800-, and N2550-based boards draw almost identical current in-situ in typical audio usage. Emboldened by these findings to widen the search to more powerful processors, we discovered that the sweet spot for high-resource/low-power no longer lies with Atom at all in 2012, but with the latest 500-Series Celerons. The TDP figures of the Cedar Trail designs look appetising (3.5W, 6.5W, 10W) versus the forbidding 35W and 65W figures of the Celerons, but CMP were right all along - and here's the newsflash - with very little tweaking, the real-world draw of the more powerful - and much more efficient - 500-Series is practically identical. That means there's no need at all to put up with the OOE and bus limitations of Atom. TDP is simply a guide to the theoretical cooling power required to keep the CPU at nominal temperatures. But, again, when you look into it, that's also a crock: the reality is that you buy Atom boards with integrated heatsinks . . . the ones we used to favour did not have fans, for obvious reasons. Many therefore run at 50-65° - well within tolerances, but still pretty hot. Moving to a socketed board, you don't have to put up with a low-rent heatsink: you can run a much better processor at the same current much cooler: 35-45°. And it turns out that heat is as big an enemy of 'low-noise' operation as HF grunge. Cool is good. Core Audio have been doing great work in this area recently. As we've said before, cryo-treatment helps even further. The bottom line is that your 'CAPS V2+' doesn't have to be disabled with regard to digital EQ, software crossover, room correct, etc. You can do it all one machine. And it will draw 12-15W at the wall just like an Atom ITX board. Welcome to Q3 2012!

2A is probably right at the edge of the operational envelope there: I would definitely recommend something with more capacity, even for a short trial. You don't want one of those going rogue on your mobo.

That's bang-on advice: SPDIF can - potentially - be a DAC's 'best input', but only when paired with a top-flight transport. And that will be costly. The beauty of USB done really, really properly is that it gives you a similar level of performance with a cheaper, and more versatile, computer via a single USB cable. The horse has to come before the cart: first, choose the DAC that sounds right, then try SPDIF converters and USB cables and different transports with a view to making that converter sound most like itself: without the transport imposing a sonic signature on it. Going from USB to SPDIF (or, better, AES/EBU) is something you should never rule out prejudicially, though: the Wyred 4 Sound DACs present an interesting dilemma here: the £899 DAC1 'only' has 24/96 async USB, whereas the £1299 DAC2 has a better 24/192 USB and AES/EBU. Although we never expected it to pan out that way, after lengthy comparative auditions, many of our customers have chosen the cheaper DAC and spent the remaining £400 on an SPDIF converter rather than go for the better USB: purely because it sounds better that way to them. If that seems like a complete contradiction to the first paragraph, I can only apologise: the inconvenient truth is that it's almost impossible to generalise: the DAC writes the rule book, and every DAC is a law unto itself. It also raises a Catch-22: how can you audition for the right DAC without knowing whether the transport is holding it back? It's easy to write off a converter as sub-par without having heard it through its 'favourite' input, or with a lossy cable. The only way to be really sure is, if possible, try it paired with the best (most transparent) USB and SPDIF transports and cabling you can beg, borrow or steal: give it no excuse not to be great. Another inconvenient truth! But yeh: bottom line, USB with a good DAC is usually better value. In this month's HiFi Choice, the MD of Perreaux was asked whether all inputs on the DP32 performed identically, and he responded with a clearly preferential hierarchy: optical < RCA < BNC < AES/EBU . . . with USB potentially topping the tree, providing the computer is setup properly. However, given that the DAC is not 100% galvanically isolated on that input, rail noise from the motherboard will play a fairly significant role in how that DAC sounds. Again, each DAC writes its own little rule book.

Chris has made clear that the idea of the CAPS series is to leverage off-the-shelf components and encourage the widest possible readership to build a computer player that aims to mitigate some of the problems presented by stock computers. The genius of the CAPS recipes is that they have been so successful in popularising the subject, and creating such a strong community of creative developers. They have given people confidence in DIY building. And the internet is better for it. But to apply for CCL, you have to prove originality: or at least your own, identifiable 'special sauce'. In this regard, I think we're all indebted to CMP whose 2007 recipe blazed the trial - followed two years later by our first DAT1 recipe and, in 2010, the first generation CAPS. Although neither was novel enough to warrant copyright protection, they were at least easier to follow than the rather daunting CMP instructions! The June 2011 CAPS 2.0 followed on from our second generation DAT1 shortly after the launch of the SOtM card that, at the time, offered a breakthrough in USB performance. Again, we're all indebted to SOtM for that one. Similarly, our third generation T1 recipe contains little that is original and nothing that is proprietary: the breakthrough this time comes courtesy of another card manufacturer and the advent of Atom-bettering processors: thanks to Adnaco and Intel! We're all fascinated to see what you've been cooking up over there, Chris . . . but as he has said, the real beauty of these things is putting them out there to see how they evolve into something smarter. As TimBL tweeted at the Olympic opening ceremony a few months ago: 'this is for everyone'.

Chris has made clear that the idea of the CAPS series is to leverage off-the-shelf components and encourage the widest possible readership to build a computer player that aims to mitigate some of the problems presented by stock computers. The genius of the CAPS recipes is that they have been so successful in popularising the subject, and creating such a strong community of creative developers. They have given people confidence in DIY building. And the internet is better for it. But to apply for CCL, you have to prove originality: or at least your own, identifiable 'special sauce'. In this regard, I think we're all indebted to CMP whose 2007 recipe blazed the trial - followed two years later by our first DAT1 recipe and, in 2010, the first generation CAPS. Although they were not really original, they were at least easier to follow than the rather daunting CMP instructions! The June 2011 CAPS 2.0 followed on from our second generation DAT1 shortly after the launch of the SOtM card that, at the time, offered a breakthrough in USB performance. Again, we're all indebted to SOtM for that one. Similarly, our third generation T1 recipe contains little that is original and nothing that is proprietary: the breakthrough this time comes courtesy of another card manufacturer and advent of Atom-bettering processors: thanks to Adnaco and Intel! We're all fascinated to see what you've been cooking up over there, Chris . . . but as he has said, the real beauty of these things is putting them out there to see how they evolve into something smarter. As TimBL tweeted at the Olympic opening ceremony a few months ago: 'this is for everyone'.

An ATX plug has 20 or 24 cables. You'll see from the pinout here: ATX power supply connector pinout and wiring @ pinouts.ru . . . that it's quite feasible to tie together the 3.3V, 5V and 12V rails and attach them to separate supplies. We've been a voice in the wilderness advocating this approach for five or six years now, but there's absolutely no doubt that it is the Gold Standard for a primary computer PSU. It literally takes the benefit of a single linear PSU into a single-rail board and multiplies it by three. You're also then well setup to tap the 5V and 12V rails directly from source to power the SOtM or Adnaco - or a PCIe SPDIF card. Anything else is fundamentally broken. A three-rail linear supply need not be expensive, either: about the same price as a single Paul Hynes PSU.

In the 80s there was a famous bumper sticker: 'To err is human; to foul things up completely requires a computer.”

That's exactly where I think we are! Although over here in the UK, major chunks of the market have only just progressed past the Technology Trigger. In fact, a segment of it hasn't spotted there is a graph yet. The penny is only just beginning to drop seriously that a computer has to be designed like a good old-skool CD transport (without the disc spinner) for uncompromised performance.

In the recent review of Mac Mini-based audio 'servers', our own HiFi World magazine concluded that you can take a Mac a long way from stock with regard to its performance as a transport, but privately the reviewer told me than the best Mac-based system still doesn't compare with a properly designed Windows or Linux system. Rather than take offense at that, I agree completely: even if you use a top linear PSU, SSD, stripped OS, convert for passive cooling, lose the optical drive, add Stillpoints fabric and a SATA filter and run Audirvana Plus, it stubbornly persists in sounding 'Mac-like’. A big part of the problem is definitely the single-rail PSU that leaves primary DC-DC switching at the mercy of high-noise components. The Mac OS and hardware management is also a culprit, I think: it's never really quiescent. If you can bear to switch from Mac (I never have completely!), you will find that an Intel/ATX board is a more versatile basis for an audio computer going forward. I recommend the Intel DH61DL as an ideal starting point for a low-noise, no-compromise board: it's about £65 and you should be able to build a really effectual transport around it for less than £400, complete with linear PSU and operating system. I know a number of people using such a system with the Audiophilleo.

Yay! You finally joined the Power Revolution, Chris! Several of the CAPS3.0 variants now have optional PSU upgrades: ATX or single rail? Looking forward to seeing more specs on your new recipes . . .

It's standard ATX spec, but actually we can get by without it fine. The Intel DH61DL board has a separate 4-pin ATX for the processor that draws no more than a few hundred millamps. You could have a separate rail just for the CPU, but it clocks up the expense: 'star-earthing' a line back to the main 12V rail seems to work just fine. Before you know it, you'll have us doing a Five-Rail PSU . . . Hmmm.

I'm sure Paul gave you chapter and verse on this: he knows what he's talking about. But in principle, it's always best practice not to introduce the primary 'contaminant' in the first place, rather than seeking ways to mitigate it once it's become a problem. In the case of poor-quality AC-DC conversion, and also primary DC-DC conversion, that 'problem' propagates immediately and pervasively throughout the system. You have to tackle it first, before looking to clean up the board or card output. Anything less is a sticking plaster for a broken leg!

If you're feeling adventurous, you can remove it. Otherwise, there's no harm in 'damping' it with glue or - less irreversibly - a small piece of self-adhesive bitumen flashing: £5.99 from B&Q - also works well to control vibration of the case itself: especially across joints.