Focusrite Scarlett 2i2
Following my review of the Calibrated Dayton EMM-6 Microphone in the last issue of HifiZine, I needed to come up with a recommendation for a suitable USB audio interface (aka “sound card”), for people interested in doing acoustic measurements, whether to optimise their system’s bass response or as an aid to building loudspeakers. The main criteria were ease of use, good performance, and a reasonable price. After reading as much as I could on anything in the under $200 range, I decided on the Focusrite Scarlett 2i2.
There are more complex and versatile units out there. In the next issue, I will be publishing a review of the MOTU Microbook II, which has a lot of interesting functionality for a modest increase in cost. However, that functionality also brings increased complexity, which works against the ease-of-use goal. There are also cheaper units around; but low cost was not the only goal. Reliable operation on both Windows and Mac, 24/96 operation, and good measured performance, were also key attributes.
The first thing you notice about the Focusrite Scarlett 2i2, on taking it out of the box, is that it is not scarlet. I’d say it’s more like fuchsia… at any rate, it looks nothing at all like the pictures on the Focusrite webpage! The case is a “brushed” aluminium and looks quite nice, I suppose, with “Focusrite” painted on the top. The knobs feel a little plasticky but I’d be surprised if anyone had issues with appearance or “build quality” of the Scarlett at the price.
Other than the unit itself, the box includes a USB cable, a “Getting Started” guide, and a CD/DVD with drivers and some other software. My first thought was to download the latest drivers from the Focusrite website. But heck, this is a USB Class 2 device, and I’m writing this on a fairly recent Mac laptop, so what happens if I just plug it in? Voila! The Focusrite showed up in the System Preferences as “Scarlett 2i2 USB” and in Audio Midi as a 24-bit device capable of 44.1, 48, 88.2, and 96 kHz operation, for both input and output.
Plugging a pair of headphones into the socket on the front and pushing out some audio through Audirvana Plus confirmed that it was “working fine.” However, I did later run into trouble getting the 2i2 working with Room Eq Wizard, and installing the driver seemed to fix the problem. So, I’m not entirely sure what the driver does over and above what Mac OS X is already doing, but you do need to install it.
Moving onto the front panel, there are two “combo” jacks that will accept either a male XLR for microphones, or a balanced 6.3 mm TRS jack for high-level sources. Next to each is a gain control and a switch to select between “line” or “instrument” input levels, which according to the Focusrite specifications have +20 dBu (7.75 V RMS) and -3 dBu (0.55 V RMS) maximum input levels. This switch matters only if you plug in a 6.3mm jack – if you plug in a microphone with an XLR connector, it makes no difference.
Around each combo jack is a circular LED indicator. Relative to full-scale input (that is, digital clipping), it turns green at a signal level above -24 dB, and red at exactly 0 dB i.e. at digital clipping. (It flashes orange briefly between green and red.) My older Edirol unit doesn’t have any meters at all, and this is now one of my most favorite features of the Focusrite, as it’s sooo easy to get your signal level “good enough” for quick frequency response measurements – just run a sweep and move the gain knob to get green. And since it turns red at 0 dB, keep an eye on the meter while making a measurement – if it turns red, you know you need to adjust your levels and try again.
Moving over to the right, there is a switch to select 48 V phantom power. You will want this on when performing measurements with a microphone, and off when performing loopback calibration. The two inputs don’t have independently selectable phantom power. Below that is a switch labeled “Direct Monitor” – you will want to set this to Off, as if you set it to On while a microphone is connected, the mic signal will route straight to the outputs and you may get feedback (“howl”) through the speakers. At best, your measurements will be suspect. Then we have the big volume control for the main outputs and the small one for the headphone output.
Since it’s got a USB input and a headphone output, the 2i2 can be used as a USB DAC/headphone amplifier. With Audirvana Plus and Sennheiser HD-650s, the 2i2 has a mellow, slightly “romantic” sound, with individual instruments and vocals “larger” and less detailed than with more expensive equipment in the same role. But it’s a lot better than being harsh-sounding. The small size of the headphone volume control knob is somewhat inconvenient, but you get used to it quickly enough.
The back panel couldn’t be simpler. There is a type B USB connector, and two 6.3mm output jacks. These are balanced outputs, but most “pro” gear doesn’t care if you short one half of the balanced output to ground. In fact, the 2i2 manual says that unbalanced 6.3mm jacks here are fine.
The simplicity here is appealing. My earlier interface, a Roland UA-25EX, has lots of buttons and switches and knobs, on both the front and back panels, few of which add much value to me. And a block diagram on the top, so you can figure it all out! Most of this is just an inconvenience, with things that I have to remember to set properly every time I use the unit (the direct-monitor switch being the main one, as it resets itself every time you plug the unit into USB). The 2i2, however, is very much plug and play.
A loopback test measures performance of the device with the output connected back to the input. While it doesn’t separate measurement of the output section from the input section, useful as a basic characterization of the device and is arguably representative of its use in a measurement application. It also provides a simple way of comparing the basic technical performance of different audio interfaces.
Figure 1 shows the frequency response at 44.1, 48, and 96 kHz sampling rates, as reported by FuzzMeasure Pro, with the output connected back to one of the line inputs with a 6.3mm balanced jumper cable. There is nothing to complain about here – running at 96 kHz, the 2i2 is essentially flat from 10 Hz to 40 kHz. (Note: each small division on the scale is 0.1 dB.)
While it’s often recommended that loopback tests used to “calibrate” an interface such as this are performed with the microphone input, I’m not sure that is always correct. If the circuit has an output coupling capacitor, then the lower input impedance of the microphone input may affect the low-frequency rolloff (in a way that it will not be affected if a microphone is used as the source). At any rate, I also ran the above measurement into the mic input as well, and the 2i2 is a hair less than 1 dB down at 10 Hz.
While not shown on the graph, the impulse response graphs indicated a round-trip latency of around 12 ms when running at 96 kHz.
To perform some basic distortion measurements, I used the Electroacoustics Toolbox from Faber Acoustical. I’m not trying to verify or challenge the manufacturer’s specifications here, but to provide measurements that I can use to compare “like for like” across different audio interfaces. I use the line input for these measurements, and set the output and input gains so that 1.0 VRMS is equal to a full-scale digital signal. Figure 2 shows harmonic distortion and noise with a 1 kHz sine wave, with the signal level set at -1 dB. The Toolbox calculates a THD of -102 dB.
In addition to the distortion components, there are some additional spikes further up in the spectrum, at multiples of roughly 8 kHz. These are not related to the signal and are present at all sample rates. I suspect that they are bleed-through into the output stage from an internal switching power supply (as all such bus-powered devices must necessarily have). Regardless, they are very low in amplitude, all being under -100 dBV.
The maximum signal output level of the 2i2 is specified at 10 dBu. I measured 9.4 dBu with my cheap multimeter. Distortion levels do rise a little at full-scale output, with the Electroacoustics Toolbox calculating a THD of -95.8 dB in that case, mostly due to a rise in the third-order component.
To test intermodulation distortion, I used the CCIF method with 19 and 20 kHz sinusoids. To avoid clipping in the signal generator, each sinusoid was scaled down to 50% and the peak signal level set to 1 dB below full-scale. Figure 3 shows the IMD spectrum of the 2i2 under these conditions. Again, that’s exceptionally good, with all distortion components 110 dB down. The power supply (presumed) noise is present again, which makes the plot look a little odd but still, it’s very low.
To check the mic input, I connected an external DI box with a total of 46 dB of attenuation between the line output and the mic input, so that the signal level was about 0.5 mV. The mic gain was adjusted for -1 dB FS input. The result is very clean, with only the 4th harmonic just slightly visible above the noise floor (of the DI box, I believe) at -100 dB FS.
The digital noise artifacts, while extremely low in level, do make the 2i2 unsuitable for applications where you might want to use it to perform electronics measurements. For acoustic measurements, you are never going to see them.
The photograph below shows the complete set of cables that I recommend for performing acoustic measurements with the Focusrite. (Most audio interfaces will also work with these.) They are:
- A standard microphone cable. This should be good quality but it doesn’t need to be super high quality, as you won’t be running extreme lengths or moving the cable around while measuring. A length of 6 meters should be enough to enable you to position the microphone pretty much anywhere in your listening room.
- A pair of RCA cables. These need to be long enough to reach from your audio interface to your preamp inputs. 5 meters should be enough.
- A pair of TRS-RCA adapters. These adapt from the balanced 6.3mm (aka 1/4″) jack outputs to a regular RCA cable as above. Shown at lower left of centre in the photograph below is a pair of “unbalanced” adapters with a single RCA socket on the back. Most audio interfaces have balanced outputs but are designed to work fine with unbalanced jacks (but check your manual before connecting). Alternately, a pair of stereo adapters, as shown lower right of centre, can be used. In this case, just connect your RCA cable to the red output.
- A TRS-XLR loopback cable. This is a short (50 cm long) cable with a TRS (balanced 6.3mm) jack on one end, and an XLR (balanced) connector on the other. You connect this from one of the outputs of your interface to one of the inputs when you do what a loopback test to the mic input.
- A TRS-TRS loopback cable. This is a short (50 cm long) cable with a TRS (balanced 6.3mm) jack on each end. This is useful for calibrating the frequency response, and might be more accurate in some cases than using the mic input. It can also be used as a loopback on the second channel of the audio interface for making timing measurements (depending on the software that you have).
You will of course also need a USB cable. One is provided with the Focusrite. I used to use a long (4 m) USB cable, but found it unreliable when running at higher sample rates. So now I use a short USB cable and long RCA cables.
Connecting the Focusrite to perform acoustic measurements is simple. I usually connect both outputs to the left and right channels of an amp, and select which channel gets played in the measurement program. In some cases (such as with REW), you will need to disconnect one channel or the other when you need to measure just the left or right speaker.
Performing acoustic measurements (Mac)
Performing acoustic measurements with the 2i2 on the Mac was a snap. The 2i2 is capable of running at 96 kHz, so this needs to be configured in Audio Midi Setup:
In FuzzMeasure Pro, it is then simply a matter of selecting the output channel that you want to drive (either 1 or 2, to measure from the left or right speaker), and the input channel that the microphone is connected to.
Here’s a measurement grab of a cheap “supertweeter” purchased for doing experiments like this. I used a calibration file with the Earthworks M30 that (in theory) makes the measurement system’s response flat to 36 kHz.
Do we really need to be able to measure to 36 kHz and beyond? Not when doing in-room measurements, but there are some reasons why we might when measuring speakers. One is that it’s a bit easier to see the high frequency behavior of a loudspeaker without the abrupt cutoff at 22 kHz. Another is that, when applying equalization with a unit that operates at 96 kHz (like the miniDSP 2×8, for example), it’s helpful to be able to measure what the EQ is doing. Finally, a while back I measured a supertweeter (the Townshend supertweeter) to see whether it did what it claimed to do. At the time, I was able to jerry-rig up a 96 kHz measurement system, but it would have been much easier with the 2i2.
In FuzzMeasure Pro, you can have the delay through the measurement system automatically removed by connecting a loopback cable. With the 2i2, this should be a 6.3mm TRS-TRS loopback cable, as you cannot turn off phantom power individually for the two inputs. The photograph below shows the cabling.
Sometimes, though, I revert to Room Eq Wizard (REW), as it has some analysis capabilities that FuzzMeasure Pro doesn’t have. On the Mac, REW supports only 44.1 and 48 kHz (and I wasn’t able to get it to run at 48 kHz). To get running with REW on the Mac, open Audio MIDI Setup first and configure the 2i2 for 44.1 khz. In the System Preferences, set the default input and output to “Scarlett 2i2 USB.” Then start REW and choose the Default Device for input and output. Running the REW calibration procedure produced the same curve as shown in Figure 1 above. With that done, it was easy enough to take a couple of sample measurements.
Performing acoustic measurements (Windows)
I’m not habitually a Windows user, but for the sake of reviewing a component like this I felt that it was important to confirm proper operation with Windows. I was also curious to try a couple of Windows-only measurement programs that I had read about, and this seemed like a good opportunity.
For testing under Windows, I used Windows 7 x64 running as a virtual machine under VMWare Fusion 5 on an iMac with 12 GB of RAM and a 2.8 GHz i7 processor. While this setup is not what most people are likely to use, I figured that it’s unlikely that anything that works on the VM will not work on a regular Windows 7 machine.
Installation of the driver was straightforward, and seems to be necessary for Windows to even recognize the 2i2. Once done (and rebooted), a small program shows up in the Start Menu that allows you to set the buffer size. I left it at the default 10 ms.
On Windows, the 5.01 Beta version of REW supports ASIO drivers, and thus runs at 44.1, 48, 88.2, and 96 Khz. It’s a simple enough matter to select the driver and the desired sample rate:
Running with ARTA took a little more experimentation. Again, once the Focusrite USB 2.0 Audio Driver is selected, running at 96 kHz was no problem at all. HolmImpulse was a little trickier again but once set up correctly, I was able to run loopback sweeps at 96 kHz.
I didn’t spend a lot of time doing measurements on Windows, just enough to confirm that it seemed to be working. As a second data point, Paul Spencer was able to install the drivers on his Win7 laptop and run a sweep in REW within a few minutes.
I also tried Windows XP SP3, running on an older Asus laptop with 1 GB RAM and 1.7 GHz Intel Pentium processor. The Asus laptop has USB 2.0 ports – if you are thinking of using an older machine, I’d recommend that you verify that it has USB 2.0 ports on it. This worked fine with REW and ARTA. I did, however, run into problems with a second Win XP laptop (not mine), which we were unable to get running with the Focusrite. I suspect it was something to do with the drivers for that laptop’s audio-related hardware, but we never figured it out.
It is worth mentioning the additional software included with the 2i2, although it’s not relevant to acoustic measurements. Ableton Live is a DAW – “Digital Audio Workstation” – that is used for recording and multitracking. The “Lite” version comes with the 2i2. How it compares to other products of this type, I have no idea. For those reading HifiZine, it might be a way to experiment with stereo recordings (you will of course need a second microphone) or for digitizing your analog material.
Also included with the 2i2 is a set of Focusrite VST plugins. If you have read Oliver Masciarotte’s book and are interested in trying out plugins in either Pure Music or Fidelia, here’s an interesting way to get started.
In a way, the Focusrite Scarlett 2i2 is pretty boring. It doesn’t have any fancy gizmos or tricks, it just does what it’s supposed to do, without much fuss or excitement. But I like the Scarlett 2i2 for this reason, and it makes it easy to recommend for those getting started as well. While the level meter around the input gain knobs could be distracting in some situations, for this purpose it’s a huge advantage, and having (and paying attention to) these in the past would have saved me some embarrassment. The look of it doesn’t appeal to me much, I have to admit, but it’s a functional unit, not a decorative one.
I haven’t spent any time evaluating its quality as a recording device – it’s designed usage – but I’m sure that there are plenty of other people who will. For acoustic measurement purposes, the only “want” feature that the 2i2 doesn’t have is a precise, repeatable gain setting on the inputs. This would be helpful when moving the mic between different measurement positions – such as nearfield and in-room – where different gain settings are required. However, no other “budget” unit that I’ve found has this feature either, the closest being the Microbook II. Since the 2i2 does have two inputs, a workaround is set the two inputs to have different gain settings and swap the mic cable between them as needed.
There is a minor concern with the low-level digital noise on the outputs, but this is not uncommon for inexpensive bus-powered devices and in any practical application that doesn’t involve actually looking for a problem, it’s still within the noise specification. To me, the 96 kHz option, the technical results, the simple and straightforward interface, the ease of getting up and running on both Windows and Mac, the future-proofing you get with a USB class 2 device, and the very reasonable price, make the 2i2 well worthy of consideration for those looking for an easy-to-use interface for their measurement setup.
HifiZine gratefully acknowledges the support of:
- Faber Acoustical, for the use of the Electroacoustics Toolbox
- Super-Mega Ultra-Groovy, for the use of Fuzz-Measure Pro
- Earthworks Audio, for the M30 measurement microphone calibration file