Why are almost all audio interfaces still USB 2 devices? And does it matter?

Audio interfaces mostly connect to computers over Universal Serial Bus, or USB. Back in the day, computers used the rectangular Type‑A USB socket, while peripherals used the house‑shaped Type‑B socket. Today, both computers and interfaces use the smaller Type‑C socket.

Depending on how recent your computer is, you might find your Type-C sockets labelled USB 3.1, USB 3.2, USB 4 or Thunderbolt. But delve into the specifications of your audio interface, and the chances are you’ll find it’s actually a USB 2.0 device. How can this work? And why are interface manufacturers apparently stuck in the past?

What’s The Difference Between Type‑C And USB 3?

The lettered connector type and the numbered USB standard refer to related but different things. The connector type determines how electrical links are made between the host and destination devices. The USB version determines how those paths can be used to transfer data, and hence how audio and MIDI flow from interface to computer and back again. It’s even possible to use Type‑A, B or C connectors purely for charging, without supporting any form of data transfer.

One of the central pillars of the USB standard has always been backwards compatibility. The idea is that older USB devices should never be left behind as computers move to newer versions. When you plug a USB 2 device into a computer’s USB 3 port, it’ll be recognised for what it is, and the computer won’t attempt to do anything that isn’t supported under USB 2. What type of physical connector it has makes no difference.

In this sense, then, the fact that most audio interfaces are USB 2.0 devices isn’t a problem. There is no risk of them becoming orphaned by newer connectors or computers. But even so, it’s reasonable to ask why that’s the case. Since the rest of the world has moved on to newer versions of the standard, why haven’t interface manufacturers followed suit? And wouldn’t Thunderbolt be even better?

Is USB 2 A Bottleneck?

Some interface manufacturers will tell you the reason they haven’t moved to USB 3 or later is because there’s simply no need. The USB 2 standard is more than adequate for audio interfaces, and thanks to USB’s backwards compatibility, a USB 2 interface can be connected to any modern computer. Because USB 2 ensures maximum compatibility without compromising performance, there is no reason to use anything else.

There’s a lot of truth in this answer, but it isn’t quite the full story.

It’s certainly true that USB 2 offers plenty of bandwidth, at least on paper. If that bandwidth could be fully exploited by an audio device, it would be able to handle more than 400 streams of mono 24‑bit audio at 48kHz. In practice, though, this isn’t possible, because the ability to control the data flow is exceeded long before the data flow itself becomes overloaded. In terms of channel count, one of the highest‑performing USB 2 interfaces on the market at the time of writing is RME’s MADIface USB, which provides 64 inputs and 64 outputs at 48kHz.

That’s just about at the limits of what is achievable with USB 2, and is more than most people will ever need.

So, in that sense, USB 2 is certainly ‘good enough’ — except that if you look around, you’ll find that very few other interfaces get anywhere near this. The large majority of USB 2 interfaces offer only 24 ins and outs, or less. Why?

What Other Limitations Are There?

One reason is that most audio interfaces are constrained by other factors not directly related to the USB 2 specification.

To explain why, let’s take a quick detour into the world of vehicles. Suppose you are in the market for a vintage sports car from the early 2000s. You might test drive a Lotus Elise and a Vauxhall VX220 and find them surprisingly similar... except that it wouldn’t really be a surprise, because they are built on the same chassis. Figuratively speaking, the same is true of most USB audio interfaces.

At the heart of an interface is a hardware processor that has a lot of complex jobs to do. It handles communication with the computer over USB. It takes audio from all the various inputs, manipulates it in various ways and routes it to different outputs, whilst keeping everything in perfect sync.

This hardware processor is a key part of the interface, but few interface companies have the resources to manufacture custom hardware. Instead, they have two choices. One is to take a general‑purpose processor such as an FPGA (field‑programmable gate array) and implement their own mixing, routing and USB controller code from scratch. The other is to use an OEM product from a third‑party manufacturer that specialises in this field. This makes development much easier, because the difficult low‑level coding has already been done by the third‑party manufacturer.

Nearly all manufacturers take the second route, and as a consequence, interfaces from many different companies are built around the same chassis — because, behind the scenes, the market is dominated by a single provider. That provider is a British company called XMOS (www.soundonsound.com/people/mark-lippett-xmos).

However, this widely used platform is designed to be affordable and reliable, not to exploit the USB 2 standard to its very limits. So, if you’re wondering why there are hardly any 32‑channel USB interfaces on the market, or why the internal mixer on your interface loses some features at high sample rates, that’s not because USB 2 can’t support it in principle. It’s because the hardware platform that is used in the majority of interfaces has finite processing resources.

What About Latency?

Latency is a word you’ll hear a lot in discussions of audio interfaces. In this context, it refers to a delay between the signal entering the interface and being heard again at its output. This can make it hard for singers and musicians to properly hear themselves whilst performing, so one of the most important features of any professional audio interface is the ability to keep the delay as short as possible.

Most of the delay arises because the audio signal has to follow a convoluted path before it reaches our ears. From the microphone or instrument, it goes first to an analogue‑to‑digital converter. The processor described above then pipes it down the USB cable to the computer, where our DAW software takes care of writing it to a drive, before spitting it back down the USB cable. Finally, the hardware processor in the interface routes it to our headphones or monitors.

In terms of latency, the most problematic points in this chain are where the signal enters and leaves the computer. To ensure smooth signal flow with no dropouts, the signal needs to be ‘buffered’, which means that samples are pooled into batches before being released. Making the buffer smaller can reduce the latency, but makes the computer work harder. Make it too small and the buffer may not be able to supply data on demand, and you’ll hear a glitch.

Does the choice of USB 2 make it hard to achieve low latency? Not necessarily. How low a buffer size you can effectively work with depends on several factors, but one of the most important is the audio interface driver. This is a piece of low‑level code that runs on your computer and handles the transfer of data over USB.

It’s not only the connection protocol that determines how good or bad the low‑latency performance is, it also depends on how efficient the driver code is. And interfaces that are built around a generic processing platform often rely on generic driver code. On Apple machines, most manufacturers opt to run the interface in Class Compliant mode, which means they can use Apple’s own Core Audio drivers. On Windows computers, the generic ASIO driver is supplied by a third‑party provider to work with the processing platform in the interface. Both work reasonably well, but manufacturers who go to the trouble of writing their own custom drivers usually achieve better low‑latency performance.

Does Type‑C Make Any Difference?

As we explained earlier, the choice of connector doesn’t determine what USB standard is used. Thus it’s perfectly possible for an audio interface to be a USB 2.0 device, but have a Type‑C port for connecting to the host computer. And that’s exactly the way it works with most current interfaces.

This switch to Type‑C ports brings with it some concrete advantages. One is that the connector and cable are both reversible, so you need never again fumble around trying to get it the right way round. A more substantial plus is that the Type‑C standard can deliver much more power from the computer to an attached device. Twenty years ago, bus‑powered USB interfaces were very limited as to channel count, processing power and headphone volume, simply because they couldn’t draw enough current from the computer. Today’s bus‑powered interfaces are a lot more capable and better‑sounding, although larger rackmount models still usually need a dedicated mains supply.

Is Thunderbolt Better?

You’ll sometimes hear that Thunderbolt is superior to USB for connecting audio interfaces. Both Thunderbolt and later variants of USB offer far, far higher data bandwidth than could ever be required for audio, and eliminate the USB 2 control restrictions that might prevent this bandwidth from being exploited fully by an audio interface. When it was first launched, Thunderbolt also delivered more electrical power than USB, but since both now use the Type‑C port, this is no longer an issue. The remaining important difference is that Thunderbolt integrates into the computer architecture at a lower level than USB. Technically, in fact, Thunderbolt is an ‘externalised’ version of the PCI Express protocol.

Put very simply, what this means is that there are fewer hoops to jump through to transfer data over Thunderbolt. Unlike USB 2, there hasn’t been a generic platform and driver code available off the shelf, but transferring data without delay or interruption is intrinsically more straightforward. Hence Thunderbolt interfaces usually perform better than USB 2 interfaces based around generic processors and drivers, though they can be matched by a well‑written custom USB driver.

However, there are also down sides. First of all, although all modern Apple computers support Thunderbolt, the same has not always been true of Windows computers (though this looks likely to change with the latest generation of USB 4 chips, which support Thunderbolt 3). And second, developing and manufacturing a Thunderbolt interface is more expensive than developing a USB 2 interface, in part because the hardware components are more costly and in part because there are no generic platforms available.

It should also be pointed out that crucial aspects of audio interface performance such as dynamic range and frequency response are not dictated by the choice of platform. It is up to the manufacturer to choose converters and design the best possible circuitry around them. Two USB interfaces based on the same processing platform won’t necessarily have identical audio specifications, and nor will a Thunderbolt interface necessarily outperform a USB interface in this respect.

Many of the limitations of current interfaces are actually down to the implementation rather than to the USB 2 standard itself.

Is ‘Good Enough’ Good Enough?

Ultimately, then, USB 2 does have intrinsic limitations, and these can make it unsuitable for professional applications where large numbers of channels need to be handled. But many of the limitations of current interfaces are actually down to the implementation rather than to the USB 2 standard itself. Manufacturers who choose not to use an off‑the‑shelf solution, and instead design the entire system from end to end, can achieve higher channel counts and lower latency. They are also able to realise more diverse or unorthodox designs.

However, the existence of a widely used generic platform has advantages too. Most of all, it has brought down the cost of multi‑channel audio interfaces. Interfaces based on this platform more than meet the needs of most users, and they do so at a highly competitive price for the features on offer. For most of us, what this platform offers under USB 2 is indeed good enough. If you’re one of the minority who needs something more or different, you have the option of buying an interface based on a different platform, whether that be Thunderbolt or a custom USB controller — but be prepared to pay significantly more.