The most visited page on this site has always been the guide to distortion measurements, written when I had just discovered the thread on DIYAudio. That post, like many of my other posts here, was written more as a reference for me and this one fulfils much the same function. If it’s of use to other people, that’s good, but please remember, I’m a complete novice to this stuff and I’m just playing around! In any case, I would advise people to read through that original thread on DIYA as there is lots of useful information there and people with actual knowledge to communicate with.
After I surprisingly sold one of my amps, I had some spare cash. This has allowed me to buy the remaining components for another build (of which I really, really don’t need) and a nice new toy – the subject of this post. I think most knowledgeable people probably saw my original post and despaired, whilst novices likely read it and then asked the grown ups for advice. I never really went anywhere with measuring devices after my initial foray, but it’s often been in my mind and I had the urge to look into it again. It was prompted when a Focusrite Scarlett 2i2 3rd Gen came up for a very good price.
Overview of the Scarlett 2i2
This 24-bit, 192 kHz resolution interface (and its variants with different numbers of inputs) appears to be a very well regarded device with respect to its price. It has been reviewed and tested at various sites over the years and you can see examples over at ASR with a follow up by one of the members as well as numerous measurements of both this and the 2nd Gen version at the DIYAudio thread. For this first post, I’ll just be looking at the interface itself and my setup before following up with measurements of some of my equipment in later posts.
Of particular interest with this device is the versatility of the two balanced inputs which can accept both TRS 6.35mm (1/4″) Jacks or male XLR connectors (although care should be exercised here – read the manual!). The 2i2 also has a monitor volume control (lacking from my previous device). Initial setup is straightforward with ASIO drivers included.
Setting up the Scarlett 2i2 with Room EQ Wizard (REW)
As last time, I used Room EQ Wizard (REW) software which is provided free, but if you do use it, please consider donating to the author. As I hadn’t looked at this program for a while, I had to spend some time refamiliarizing myself with it. Below, I will provide the steps I took to get it set up.
Assuming you have the correct and most recent drivers for your interface, download the latest version of REW for your OS (V5.20.3 was used here). On the opening screen, click on the preferences tab on the right and you will find the screen below which I have already set for my device. Input 1 and output 1 correspond to the input 1 on the front panel and the left output on the rear panel. Note the ‘Help’ panel at the bottom of the screen which dynamically updates to provide relevant information. Much more detailed help is also available at the official website.

The next stage is to calibrate the audio interface by clicking on the ‘Calibrate soundcard’ button to perform a loopback test. It is worth fully reading the Help section of the tool to understand what is happening and for instructions. At this point an appropriate cable needs to be connected to the output and input previously selected in the preferences panel. What constitutes an appropriate cable? I provide some details below on different cables I have tried, you may get different results.
The next section asks you to set the input volume to match the output volume. If you are using JAVA drivers you will see options to control volume in the panel, but if using ASIO drivers this has to be done on the 2i2 (and/or device under test). You may need to alter monitor volume and gain. The image below shows my levels closely matched.

Clicking ‘Next’ advances the Help providing further information on the process. Once the process has finished you will end up at the screen below. You will land on the SPL & Phase screen but you can see other information, including distortion, by clicking on the small row of buttons adjacent to the SPL & Phase button which is pre-selected.

At the top of this screen are a number of icons – choose RTA (Real Time Analysis). Once that screen opens up, click on the gear icon on the right which opens the graph controls. Here you can control various display parameters, useful when doing like by like comparisons with other graphs. Hovering the mouse cursor over the display window will show a small drop down box in the left of the main window – here you can choose what the graph displays (dBFS seems to be the popular choice). Click on the red ‘record’ (circle) icon on the right to control playback and pause. The spectrum displayed shows the self noise of the 2i2 at the parameters chosen (e.g. 48kHz sample rate). To capture the image, click on the camera icon at the upper left.

If you go back to the original screen you should see an icon labelled generator – clicking this will bring up the control panel for a frequency generator. Check the boxes are appropriate and click on the green arrow to send a signal from the 2i2’s output to its input.

Switch back to the previous window to see your tone displayed. Click on the large ‘Show Distortion’ button at the top of the window to see the distortion results displayed in a box to the top left. Moving the mouse cursor to the bottom right shows a span button which alters the visible display to your required measurements (set on mine to 20-20,000). Here you can see the 1kHz tone and make out some of the harmonics, particularly the third. The distortion measurements in the top right show THD at 0.00051% and THD+N at 0.0074%. Audio Precision aren’t going to be troubled by these devices, but they are a fun, cheap way to start playing with your gear.

At this point, you can make a calibration file for the 2i2 by returning to the preferences window, clicking on the Make Cal file button and typing in any notes and a descriptive name for the file to be saved. This allows REW to be calibrated for the 2i2 and the ‘signature’ of the interface, the self noise etc., will be accounted for in future device measurements.
Cable choice for initial loopback test
From what I’ve read, you may have more luck (better results) with one type of connection or other. There are so many variables that this will always only be a rough guide, some people seem to be getting good results using interconnects that others deem sub-optimal, so it’s worth experimenting to see what works on your particular setup. Remember at this point I am only concerned with the loopback test, not with measuring external devices. Additionally, it is worth acknowledging that your computer and environment will probably impact on the results and that what I get on this PC will likely differ to what you get on yours. I have tried the test on my laptop, sitting in my kitchen, with and without AC mains connected, but I found my desktop PC was quieter.
Whilst there a plenty of self-test measurements to be found for this device, I don’t think it is overkill to include another set here as this provides a baseline for subsequent audio equipment tests and also shows a larger sample size of measurements of the same unit. I’ve tried the loopback test with three different cables as shown in the image below. The 2i2 was calibrated for each cable swap.

Cable 1 – loopback TRS to TRS with homemade 3m lead – Monitor volume up to full and gain at about 9 o’clock to get input level to match output level
Cable 2 – loopback TRS to TRS 1m Stagg SAC1PS DL lead – Monitor volume up to full and gain at about 10 o’clock to get input level to match output level
Cable 3 – loopback XLR to TRS 1m HosaTech STX-103m lead- Monitor volume set to 2 o’clock and gain set to zero to get input level to match output level



As you can see from the images, there are very slight differences between the cables summarised below:
Connections | Cable length | THD % | THD+N % |
Cable 1 TRS to TRS | 3m | 0.00044 | 0.0074 |
Cable 2 TRS to TRS | 1m | 0.00054 | 0.0077 |
Cable 3 XLR to TRS | 1m | 0.00034 | 0.0074 |
When I tried this on my laptop the 1m lead was noticeably noisier, but on the PC not so much. The most obvious difference is the suppression of the 2nd harmonic with the XLR connection.
Using an external oscillator to provide the signal source has been recommended by some. I have the Akitika 1kHz 2ppm and connecting this to the 2i2 with standard RCA cables gives the following results for the loopback test shown in the image below. Note the wider skirting around the fundamental frequency and more pronounced harmonics. There is also a slight bump at mains frequency (50Hz). Overall distortion figures are better with THD at 0.00015% and THD+N at 0.002%. The cables I used are shielded single core with the cable shield connected across pins 1&3. There has been a suggestion that rewiring the cables may reap some benefits. I have ordered some balanced cable to make a TRS -RCA cable with signal wires connected to pins 2&3, leaving pin1 unconnected and hopefully benefitting from the shielding. I’ll hopefully include the results from this in the next post.

Troubleshooting
Looking around various sites you will likely see widely varying results, even for loopback tests. It is important to compare like with like, so parameters for setup should be made explicit. I noticed that setting the sampling rate in the preferences screen made a large difference to THD and THD+N readings – the lowest distortion figures come from measuring with the lowest sample rates (e.g. 44-48kHz). To illustrate this, the image below shows the loopback using the XLR-TRS cable with the sample rate set to 192kHz. Note the increase in THD and THD+N.

If you are experiencing problems with this initial loopback test, particularly with regard to obtaining a stable reading on a Windows PC, it may be worth following the advice from DIYAudio user Kozard who found the following helped:
1. Run with realtime priority (see post #734). [Most important improvement.]
2. Turn off WiFi on the laptop during measurements.
3. Close every other other program.
4. Run 44.1k or 48k on a slower/older laptop.
I would also add to this that I remembered my mouse was wireless on my laptop, so removed the USB wireless dongle.
I think I’ll call it a halt there and continue in the next post once I have tried another cable and started some measurements.