So, I have been using a Focusrite Scarlett 2i4 for quite some time. I have now changed to another cheaper interface, and I though it was a good occasion to draw my conclusions on what I think about the Scarlett.
Overall, my opinions are mixed. There are good pros about this device, but also few significant cons.
Although the product I am reviewing is not current anymore (there is a new generation out there now) I think it might still be useful to have a look into this, as it can give some insight about what kind of products and functionality we can expect from Focusrite.
My History with the Product
I got my first Scarlett on 17 June 2014 for 178 Euros. It lasted less than 10 minutes. the interface analog electronics stopped working as soon as I connected a condenser microphone and pressed the +48V button. I could see the interface registering under ALSA, and I could start JACK. However, no input, no output.
I concluded that some component was probably fried in some way by the +48V. I though I just got a bad unit. It was replaced with a new one for free, as it was under warranty, and I have been using that until last Christmas.
The Mechanical Aspects
The interface might look rugged at first sight, but that impression did not last long after I started using it. The switches are not really sturdy: it feels like a bad hit could break them easily. The potentiometers caps are easily removed, but they were not too loose, so I would be surprised if they could came off on their own. The pots themselves were well retained on the front panel by their mounting nuts. What I felt like it was the absolute worst were the audio connectors. The XLR/JACK connectors worn out very soon, by not even heavy usage (I used this interface very sporadically). During the last month of usage they would make awful clicking noises and often need some tinkering to ensure a reliable connection, like some gentle shake of the Jack plug of my guitar.
For my master dissertation (See here) I had a chance to measure this interface frequency response. More exactly, I used a technique, called swept sine technique, which allows to measure frequency responses of audio systems. To measure an interface we can loop its output back to its input, and stream a swept sine. The response of the output and input stage of the interface will distort the swept sine, and by comparing it with the pristine one we can find the interface response. The result is shown below.
Just focus on the orange curve (the blue just has a phase shift). We can see that, in the audible range, the interface response is -dead- flat. This is good, as we want neutrality of what we record. The drop at high frequency is actually were the bandwidth of my measuring swept sine ended. The phase doesn’t look bad either. Phase that rolls in that way, in fact, will introduce just an overall delay to the signal.
So, in this regards, the interface is very good.
However, there is a nitpicking: the potentiometer used to balance input and playback to the headphone would make some signal from playback leak into the headphone when it was set to completely to input.
I did not use the MIDI connectors a whole lot, but they worked just fine.
latency was acceptable. You can see many latency tests being reported in this thread on Linux Musicians.
The tests were performed with jack_iodelay (see this HOW-TO), which is a nice utility able to measure round trip latency. jack_iodelay outputs a multi-tone signal that can can be steamed to an output of the interface. This output of the interface can then be connected to one of its inputs through a cable. The signal at this input can then be patched into jack_iodelay input. By comparing the phase difference between its output and input, jack_iodelay can calculate the total loop-back latency. Hence, jack_iodelay will report the sum of the latency from input electronics and analog to digital conversion (ADC), the latency from hardware controllers, the buffering latency from the operating system audio stack, the latency of output electronics and digital to analog conversion (DAC). Given that the buffering latency can be calculated a priori, we can subtract it to find the latency that is hardware related only. This is called the extra loop-back latency, which for the Scarlett was around 8 ms, Given that the total loop-back latency, which includes all delays, was around 13 ms, we can see that the Scarlett was responsible for almost 62% of the total latency. Still, we can expect ADC and DAC to take the same time. This means that the Scarlett provided 4 ms DAC and DAC latencies. Which is a very good value.
Finally, by looking at the table in this post, we can see that 13 ms is just below the threshold at which latency starts to be audible as a delay for guitar players. So, the Scarlett was able to give a good latency performance, ensuring a stable JACK setup for which delays were inaudible. Other than comparing well with published experimental results, I can confirm I never heard delays when using the Scarlett, although playing on tight grooves felt weird sometimes.
Then the Interface Died
Yes, it died the day before Christmas 2017. How?
Well, I connected a couple of condenser mics, pressed the +48V, and it was gone just as the first one I had. I did not see this problem being reported a whole lot, so maybe I was really unlucky, but I would consider strange that 2 random devices would die in the same exact way. Probably there is something wrong in how the +48V power supply was designed or with how the PCB was manufactured.
Since the interface was not under warranty anymore, I decided to open it up to see whether I could spot something obvious. I couldn’t. but I found this:
I am not sure what part is represented in the figure, as I couldn’t quite locate the data sheet. Nevertheless, It has few pins shorted by a blob of solder (see the red squares). Also, many capacitors and inductors were bent. Although this did not seem to alter functionality, it made me think that Focusrite PCB manufacturing quality was probably not that superior to that of cheaper products back in the day. Perhaps, that’s why my 2 units committed suicide by frying themselves with +48V…
Nothing to say about it a part that it was great. Just plug and play. Awesome!
The Scarlett was overall a very good interface, with very flat frequency response and low latency. Not properly rugged, but not fragile either. It has been a great companion, and I even used it for my master dissertation. Linux compatibility was top notch.
However, the fact that two interfaces died in the same way, the soldering errors on the PCB and the low quality of the connectors made me believe that, at least for the series from which my interface is from, Focusrite did not offer higher quality with respect cheaper interfaces. For this reason, I decided to go for a Behringer UMC202HD as a replacement instead of a new gen. 2i4 (or 2i2).
I will let you know what I think about the Behringer too, after I get to use it for a while.