Interfaces: USB 3 vs Thunderbolt

by Pablo Bellinghausen

There’s never been a better time to buy an audio interface, with more accurate converter chips and circuitry than has ever been possible in portable, inexpensive devices. The new generation of computer connections promises better speed than the industry-standard USB 2 and Firewire. However, many people are surprised at the almost complete lack of USB 3 interfaces, compared to the steadily growing number of Thunderbolt models. There are good reasons for this, so if you’re one of the many contacting us lately enquiring about this, look no further!


USB 3.0

USB 3.0 has been in the market for a few years, but to date there are only a couple of interfaces using the protocol natively, the most successful ones being the RME MADIface XT, which uses the extra bandwidth to achieve amazing track counts (almost 400 channels of in/out), and the Presonus Studio 192, which actually runs mostly at USB 2 specs.

The main reason for this is that even though USB 3.0 is a lot “faster” than USB 2.0, meaning it can carry more information at the same time, that information doesn’t reach its destination much quicker – the latency (delay from input to monitoring output) that is inherent to USB 3.0 is almost the same as what can be achieved practically through USB 2.
It’s like adding extra lanes to a motorway; one can fit more cars at the same time, but the driving time for each car will remain roughly the same.

USB 2.0 is already very good for audio; as of 2011, Core Audio (Apple Mac’s implementation of all computer sound) can achieve a latency as low as that of FireWire, although in many cases it is still noticeable.

PrintReal latency calculation for a DAW-reported latency of “2.9 ms” under USB

A lot of people confuse the audio driver’s buffer size with latency; although a bigger buffer size will definitely add audio latency, even the smallest practical input buffer (usually 32 or 64 samples) has to be added to the output buffer, converter latency, and USB bus clock, all of which add to a bare minimum of at least 5 ms “mic through software to headphones” latency on any USB device; and in most real world scenarios, the lowest usable latency will easily go over 20 ms. Many people think they can easily detect very low latencies, but this is often caused by them trusting the reported latency of their software, which often omits the extra delays caused by the bus clock and drivers in general and is therefore often a lot lower than the real figure.

The only reason why USB 3.0 would be better for 4-channel interfaces or smaller would be in order to properly bus-power the product, since USB 3.0 can provide 4.5W of powered as opposed to the 2.5W of USB 2. It is worth noting that in most cases, current bus-powered stereo USB 2 interfaces will either have underpowered headphone amps or not provide full 48V of phantom power (a surprisingly common problem in many budget models). High-quality ones will all provide options to deal with this limitation (the Apogee Duet lowers the headphone volume automatically when microphone current is too high, and products like the Audient iD22 and the Focusrite Forte don’t supply phantom power at all without a power supply).

For most other applications, high-quality USB 2.0 is for all intents and purposes as good as USB 3.0, and in most situations the added power and track count are not enough for manufacturers to switch to the USB 3 standard, which is growing in popularity but isn’t as commonplace yet.



Thunderbolt is the fastest external connection currently available, being basically able to carry the internal PCI Express connection in a cable. This is almost like plugging straight into the motherboard – if USB 3.0 track counts are already in the hundreds, theoretically Thunderbolt’s bandwidth allows over a thousand channels of high-quality audio, although that isn’t really its selling point. It is more expensive to implement and has licencing costs (which is the real reason why there are still relatively few computer models sporting the socket) but there really are several real-world benefits that make it the ideal choice for high-end audio interfaces:

Lowest latency


Since Thunderbolt is allowed almost straight access to the CPU, it is able to lower the “round trip” audio latency from about an absolute minimum of about 4.5 ms through USB, to under 1 ms. This less time than it takes for sound to travel from a snare drum to the drummer’s head! Even at larger buffer sizes common during multitrack mixing, tests show that on a Thunderbolt-enabled computer the latency remains stable under the 6 ms mark (some even staying under 3 ms at 128 sample buffers), which would be completely impossible to get with any USB interface under a similar workflow. This makes it easy for product manufacturers to add many DSP effects into the interface to be used as plug-ins without any noticeable delays or problems; the Apollo series by Universal Audio is a great example of this, and the sound of these effects is fantastic to boot.

Better reliability

Since the protocol has been guarded by Intel so fiercely throughout its development, there have been no implementation struggles between different companies, which are usually where compatibility problems start; a Thunderbolt-ready interface can be guaranteed to connect perfectly to a Thunderbolt-ready computer – whether the software drivers are fully compatible with the operating system is another matter though!

More power for portable devices

The very first iteration of Thunderbolt gives 10W of power, more than double of what USB 3.0 can do. This allows for ample and reliable juice to power high-quality stereo microphone preamps and headphone amplifiers.

Incentive for high quality

At this point in time there are a few interface ranges sporting true Thunderbolt implementation, but they all have something in common – they are all the flagship models of their respective brands (Avid, Apogee, MOTU, Universal Audio and Zoom come to mind). The analogue-to-digital converters of all of these models are all top-of-the-range, the preamplifier circuitry a cut above all of their own USB of Firewire interfaces, and the build and driver quality in general is as high as each of these companies are able to provide to date. This is in order to justify the higher price associated with Thunderbolt; a budget interface sporting this connection wouldn’t make much sense!

A special mention needs to be made of the newest flagship models by Focusrite, called the Clarett range; the new 8Pre and 8PreX offerings are a feat of technological achievement, with a transparency and musicality far beyond any of their previous interfaces, and up there with the very best in the market. And with the 2 and 4-input versions coming in the next few months (2Pre and 4Pre, respectively), these should be at the top of the list for anyone looking for a top-notch interface today.

Clarett-Group-FlatThe new Focusrite Clarett 8-way interfaces are some of the best models available right now


USB 3.1, Thunderbolt 3, and USB-C: The future is bright

For anyone reading this who remains unconvinced about the future of computer audio and the role that different connection types will play in it, there are some great news for you: USB and Thunderbolt are not exactly merging, but they are becoming partly compatible, and better than ever. This is through the release of their latest incarnations (USB 3.1 and Thunderbolt 3, respectively), which both use the same connector, called USB-C. This is a very new development, with only a couple of computer models sporting USB-C ports, none of which are Thunderbolt-enabled at the moment, but they will start becoming available to the public as early as the beginning of next year.


The USB-C connector is actually not tied down to the USB interface at all, but is the culmination of years of development, and can be used with many protocols, including Thunderbolt and DisplayPort. It is cheap to make, completely reversible (no more scrambling at night or at the back of the computer trying to plug it the wrong way around), and a far more secure fit than any other USB or Firewire plug. It can also carry up to 40 Gbps (8 times more than USB 3.0) and can carry up to 100W of power – enough for several products in a completely mobile, battery-powered system.

USB-C ChromebookThe tiny new USB-C port featured on the latest Macs – here displayed on the Chromebook Pixel

The future seems like it will look like this: budget PCs will have USB-C ports carrying USB 3.1, to which any USB 1, 2 or 3 interface can be plugged in, and high-end PCs and Macs will have only Thunderbolt USB-C ports, to which any interface will be able to be plugged in with the best possible performance, all of this with just the right lead. Thunderbolt’s advantages in latency and power over USB will still remain, but any current sound card of either type is now guaranteed to be future-proof, with backwards compatibility provided throughout the range.

At this point Thunderbolt interfaces cannot work with USB 3.1 ports even if the connector is the same, but it’s not inconceivable that newer Thunderbolt interfaces could be created that could revert to USB 3.1 mode should they detect that Thunderbolt is not available.

In any case, the sound quality and performance of all the latest models is so high that it’s hard to go wrong nowadays, so go get the best model for your needs and start recording some music!

Leave a Reply

4 Comments on "Interfaces: USB 3 vs Thunderbolt"

Notify of
Sort by:   newest | oldest | most voted

USB 3 vs Thunderbolt interfaces –

Mark Sto

Well, what about USB 3.0 SuperSpeed protocol’s ability to work with data flows which e.g. allows the USB Attached SCSI (UAS) interface for modern mass storage devices? Aren’t such abilities also suitable for audio interfaces from their developers point of view? It is worth saying that exactly this data flows vs. packages approach in data transfer was an essential historical difference between USB and FireWire (IEEE 1394) standards and thus the latter one was a de facto standard in portable audio recording for a long time.



This article is woefully inaccurate. A good USB interface can achieve close to the round trip latency of a Thunderbolt interface. For example 1.2ms measured round trip latency with a Zoom UAC-2 unit running on USB3 with a sample rate of 192kHz with a 32 sample buffer size. Click and pop free here on a laptop. It’s 2.2ms at 96kHz with the same buffer size. 3.3ms at 48kHz with a 24 sample buffer. Again, this is the measured round trip latency, not the numbers the drivers give (which are accurate in the case of this device, but this is far… Read more »
Matt Parsons

Two points to note:

USB3.0 allows up to 100Watts of power (10 times that of Thunderbolt 2)

I have been using a Zoom UAC-2 for one year on a USB3.0 bus and found it to have a lower CPU overhead than on a USB2.0 bus (on Mac, not tried on Windows) when operating at an input latency of ~2ms.