Digital converter technology has advanced immensely over the last 20 years. With a wide range of excellent products to choose from, it can be a daunting prospect deciding on the best box for your particular needs. In this article, our technical manager Tom Shorter gives a summary of the key points to consider when choosing your weapon…
The clock source is arguably the most important component of a digital converter, and for most people this means minimising jitter.
So what is jitter? In essence, this is simple to understand... if you are recording @96kHz, your "real world" analogue audio signal is being sampled 96000 times every second. In an ideal world, perfect reproduction assumes that the period between each sample is exactly the same. However, in reality there is a small amount of variation, so your 96000 samples are being taken at very slightly different times - this is jitter. Samples taken at slightly the wrong time will be reproduced slightly incorrectly and won't truly represent the original source material. Apart from Nyquist sampling frequency considerations, this is one reason why higher sample rates produce better conversion - the same degree of jitter becomes less significant as the sample rate increases. The best converters use sophisticated signal conditioning (both software and hardware) to ensure the lowest possible jitter - the better this is, the more expensive it is likely to be.
Manufacturers often claim jitter specs in the region of a nano or pico (0.000000000001) second - sometimes even less! Although these headline figures are a useful indicator of the quality of the technology in use, they aren't the definitive variable. As with all audio gear, circuit implementation is the make-or-break of any converter. You might be reading a tiny and impressive jitter figure at the output of the PLL (Phase Lock Loop) module, but whether this number remains the same once it's travelled the entire length of the circuit board is another matter.
Once you have a converter with a world class clock, resultantly low phase noise and spectacular transient response, you don't want to ruin it by adding another AD/DA in the chain that is not synced properly. Most devices will allow you to pass a clock through the digital audio stream - but we usually recommend against it. Apart from it generally being more jitter-prone, it means you're limited in what can be the master clock and what's being fed where. Also, if you regularly change your setup, keeping track of what's feeding what can become tiresome.
Where possible, proper BNC Word Clock over dedicated cables is the better way to go. It’s good practice to keep the clocking separate to the audio - make sure it's done right the first time then you can simply forget about it. Star clocking is preferred but a chain is still a lot better than letting your converters lock to a digital stream – and the more complex your digital setup, the more this matters.
With the main focus on A/D and D/A quality, the amount and type of digital connectivity on your converter is often overlooked. Having enough of the right type of digital I/O future-proofs your system and allows a myriad of expansion options. These may include a 500 series chassis with ADAT I/O built in (eg. the new Cranborne 500ADAT), separate converters for monitoring (eg. Crane Song Solaris) or additional inputs from preamps with digital option cards in them (eg. Neve 4081). There's a world of accessories that can be simply and stably integrated into your system without needing to go through the inconvenience of freeing up or re-patching analogue I/O, or (in the worst case) having to consider a different interface.
AES/EBU, S/PDIF, ADAT, MADI, Dante, Ravenna - while there are deviations between the formats in terms of jitter and data transmission, it's best to consider your application first and consider these after. ADAT is better for multiple channels. AES/EBU is no-fuss and uses simple, robust XLR connectors - plus you can combine 8 channels into a DB25 connector (don't forget to check the pinout – Tascam is by far the most popular convention, but there are alternatives out there). S/PDIF and AES/EBU can be fed into each other assuming you ensure the impedance of the cable doing the hop has the correct termination impedance – ie. use 110 Ohm cable when running S/PDIF to AES, and use use 75 Ohm when running AES into S/PDIF. MADI is a hugely efficient protocol and can bus hundreds of channels up and down without breaking a sweat. Although still fairly niche in the world of audio interfaces, it's worth exploring if you need to connect many channels over long distances – which is why it is so well established in live sound applications. Dante and Ravenna add the obvious benefit of being able to network multiple booths, rooms and control rooms together with high channel counts.
Current devices will generally be Thunderbolt, PCIe, USB, Dante or Ravenna. Once you get into the world of high channel counts at high sample rates, choosing the host you want to use becomes particularly important. PCIe is pretty much always the best way to get audio in and out of a DAW. There are no shortcuts to the stability and low latency you can expect from a high-bandwidth data bus bolted directly to your motherboard. At this point it's worth mentioning that Thunderbolt is also a PCI-based format and you can expect similar performance - assuming that the Thunderbolt adaptor in the computer is properly implemented.
Dante can handle the extra load caused by a high channel count with impressively low latency, but you still need to think about how to get those streams in and out of your DAW. USB is generally not the place you should be looking when you're trying to close-mic an orchestra. However, it's the most ubiquitous connector on the planet and that brings its own advantages. A well-designed USB driver can handle a lot more audio than you would expect. Especially if it's receiving data from something like MADI. Plus, if you're on the road a lot and want a reliable box that can come with you, you know you'll be able to plug USB in to basically every machine on the planet and use your converters, pres and headphone amps. Not something to be overlooked for today's busy engineers.
When it comes to getting an analogue signal to and from your converter, you’ll see TRS, XLR, Phono (RCA) and DB25. If you need a high channel count or run everything in to patchbays, DB25 makes for the cheapest and simplest path. And you can always use breakout cables if you want to mix and match things between the banks of 8 audio paths contained in each DB25 connector. TRS is convenient and flexible for line inputs and, naturally, keep it XLR where possible for preamp inputs to avoid possible damage caused by mis-directed 48V phantom power.
The idea of networked audio seems to make a lot of people nervous. We were initially told that you can plug a red box into a wall and send an infinite number of channels at 1000khz with 0ms latency and everything will work perfectly. Of course, this is simply not the case!
However, with a properly set up network and the right selection of interfaces, you can have a very solid system with the added flexibility (and significantly reduced costs) of not needing any analogue patching or tie lines between rooms. As well as reducing signal noise, this means that if you're renting a retrofit space, you don't need to start pulling up the floor or drilling holes through walls to run chunky multicore looms. There's even a good chance there will be some CAT cable already installed which you can take advantage of.
If you're a converter snob like me, you shouldn't be afraid of networked audio - just go for boxes that have the right digital connectivity and use whichever standalone converters you prefer.
Ultimately, this is what it’s all about. The pace of change in the world of interfaces and converters is staggering. The major players fight it out constantly, the result being that consumers are able to enjoy better quality audio at increasingly affordable prices. Some of you may remember the days when improving on Digidesign’s “premium” (at the time) 888 Pro Tools interface meant having to buy a painfully expensive Apogee AD8000.
Even so, there remain vast differences in how these devices sound – just listen to the sound files in our A/D converter comparison. As the technology moves forward, it's likely that these deviations will only increase – reintroducing an element of subjective preference back into an area which has been largely considered “objective” until quite recently. Beyond a certain point, you simply can't say that one converter sounds “better” than another – just different.
Therefore our final bit of advice is this... treat interfaces in the same way you would when buying any analogue gear - work out if it is in budget and has the functions you need - then LISTEN and decide if this is how you want your audio to sound!
And if you’re still not sure, we’re always happy to discuss these things with our customers – just get in touch!