Focusrite Scarlett 2i2 Audio Interface Gen 3 Review

[AnalogSteph]

Curious. I have a Gen 1. Quite happy with it for speaker testing, but I notice a big difference in your graphs than mine. My noise floor runs from about -120 dB @ 20 Hz, rising in a strait line to about -94dB @ 20K. You show a flatter noise floor. Is this a feature of the DAC technology, or maybe a feature of my spectrum analyzer? ( TrueRTA)

Pretty sure that's a software / processing artifact, I've seen similar when distortion over frequency was being measured. You can try recording some of your noise floor and using the spectrum analyzer function built into RMAA (N = 16384, Kaiser-Bessel window, beta = 20).

A/D dynamic range in the current generation is up to 111 dB(A) from 105 dB(A) in the first one. While a nice improvement, I suspect you'll generally be running into microphone and room noise first, and skilled use of the input gain dial would make a greater difference.

If you think it is not good enough for home recording, just think back a few years when the best we had was a Sure pre-amp into a Sony quarter track 7 inch reel to reel. This thing is better than the every best studios for when most of my CD collection was recorded.

If you are talking firmly 1980s and earlier, then probably yes. Studio-grade ADCs were reaching 18 bit performance by the early '90s. A few years later, the best ones reached 120 dB, and by the end of the decade SOTA had moved to 130 dB, and 96 kHz recording was basically standard.

 

[tvrgeek]

I will try Right Mark. The noise floor I am measuring is in loop-back, line out to line in. Or if I use an external DAC, and then line in, virtually no difference when using a good DAC. Totally irrelevant for my speaker testing, but does mask even mid-range electronics now.

Yes, I was thinking about 70's. I suspect CDs got a bad rap early as the first Sony mastering systems were 14 bit and the engineers needed to learn how to get the best of a new medium. They got better quickly.

 

[tvrgeek]

I get an error opening file for the download from both Rightmark and a mirror site. 6.4.5 Downloaded SoundCheck. It shows a flatter noise floor. Looking for other tools.

 

[Blumlein 88]

I get an error opening file for the download from both Rightmark and a mirror site. 6.4.5 Downloaded SoundCheck. It shows a flatter noise floor. Looking for other tools.

Did you try to open the file as an administrator? If not you'll see an error. They actually mention this on the download page.

In any case go get REW, it can do measurements too.

https://www.roomeqwizard.com/

 

[tvrgeek]

I have a hint it is how True RA works, but the other tools I downloaded only go to -100 dB.

 

[tvrgeek]

Ah, stupid windows, told it the first time but did not behave. Installed OK. I think I verified what I was seeing was an artifact of TrueRTA.
I do not see that with RightMark. New tool, lots to learn. Report says my version 1 meets spec. Imagine that.

Good news: verifiable results with RightMark. They correlate with TreuRTA. RM seems to have a finer resolution.
Good news, sort of: THE Scarlett outperforms the SMSL Scanscrit 10 and the Topping E-30 for distortion and noise. By a lot.
Good news sort of. My old MUSE does have a bit more distortion, higher above the floor so the reading is valid. But way better than the new ones.
Bad news: Right channel on my V1 is blown.
Good news for Focusrite, looks like I'll buy a V3 right away.
It seems odd the new super chip wonder DACS have shown up as terrible on my bench.

Loopback, RightMark. Left line out to line in for my old V1.

 

[AnalogSteph]

Good news, sort of: THE Scarlett outperforms the SMSL Scanscrit 10 and the Topping E-30 for distortion and noise. By a lot.

Both have unbalanced outputs, so I suspect your cabling may not be adapting that to the balanced input correctly and inviting a ground loop. You should have this for a pinout:
unbal signal --> hot
unbal gnd --> cold
(not connected) --> shield

Bad news: Right channel on my V1 is blown.

You have verified it's not the cable by swapping them over, right?

Good news for Focusrite, looks like I'll buy a V3 right away.

Gen 2 and Gen 3 also have a switchable 10 dB input pad which the Gen 1 does not seem to have had yet, which should make matching levels easier. I think the line-in on yours will tolerate >10 dB more than the max output level.

Do consider an upgrade to a Clarett 2Pre, budget permitting.

 

[tvrgeek]

8 inch cable, not a ground loop. Verified I blew it up. Probably as I was verifying power supply noise before and after re-capping some amps. Transient maybe. Of course, the box with Zeners in it from Digi-key is sitting on my workbench. My bad.

I made loopback for both balanced and unbalanced. Yes, slightly less low frequency noise with balanced, but with 8 inch cables not an issue. All my equipment is unbalanced, so to compare apples to oranges, using unbalanced. Doing a baseline with balanced would be comparing apples to cows.

I'll look at the big boy. Part of what I was doing was verifying config and performance of a couple new DACs. SMSL was clearly a defective sample. Topping I am measuring distortion way in excess of the Scarlet where the addition should be barely noticeable. ( spec is .0003%) Some are suggesting I am an idiot ( maybe true) but I have my suspicions it has more to do with how they designed the output filters to support the insane data rates and may meet spec at higher rate. I only run 44.1 and 96. The old Focusrite is far lower distortion than either of the two new ones. Noise floor in all is so low as to not measure with any confidence. -120 or so. Good enough.

So, if only for measuring a few more speakers, I would not buy new but limp by, but as an excuse for the main stereo upgrade, V3 or Clarett. I'll consider.

I was actually considering the 4 channel and do my crossovers on host, but the UIs for input switching and volume control are not convenient enough to pass WAF. Not yet.

 

[AnalogSteph]

I made loopback for both balanced and unbalanced. Yes, slightly less low frequency noise with balanced, but with 8 inch cables not an issue. All my equipment is unbalanced, so to compare apples to oranges, using unbalanced. Doing a baseline with balanced would be comparing apples to cows.

This line of thinking is flawed. It is fundamentally impossible to do accurate loopback testing of unbalanced outputs using an unbalanced connection without providing galvanic isolation between input and output. Which is why dedicated audio analyzers feature that. In an unbalanced playback setup, it is provided by the mains power supplies. The only way to mimic that would be splitting playback and recording between two computers, one of which must be battery-operated.

If galvanic isolation is not an option, even a poorly balanced connection (without taking the time to accurately balance impedances) as outlined above would generally still provide a workable amount of CMRR. Even 30 dB is one heck of a lot better than nil.

When you are measuring a USB DAC with a USB audio interface on the same computer using an unbalanced connection, you are inevitably creating a ground loop between them that is sharing paths with both devices' USB power supply return currents. An adapter cable as outlined above (not available off the shelf to my knowledge) is breaking the loop and letting input CMRR do its job to a certain degree; it could be optimized further by including a model for signal output impedance between "output gnd" and "cold", if known. (That's the way a lot of inexpensie studio gear implements balanced outputs, actually. It's called pseudo-balanced or impedance-balanced.)

 

[Music1969]

Anyone know if the volume of this Gen3 Focusrite is controlled in digital or analogue domain?

 

[tomeh]

Lot's of good comments. Thanks.

Question foir the "CFO", how can you compare devices containing DAC's when the actual front end is totally different such as a headphone amp containing a DAC versus a two channel, microphone preamp containing a DAC?

Isn't it hard to null the active, front end of each? Or should the general title for noise and distortion be SINAD of devices containing DACs?

I'm just thinking the casual reader may say this or that device is terrible because the SINAD measurement is worse than XYZ unit when in fact it may be not comparing apple to apples??

Cheers,

Stay well
Tom eh

 

[maxm272]

I'm about to order the 2i2 to digitize vinyl.
Because my phono preamp has RCA outputs, I don't have XLR cables. So, am I right to assume that I'll have to buy two mono TS adapters, one for each channel? The manual says that the 2i2 does accept TS inputs.
Sorry if my question seems ignorant, I just want to make sure I don't buy the wrong adapter and possibly increase distortion or damage something.

 

[Blumlein 88]

Yes two TS or TRS to RCA adapters. Goes from phono pre output to inputs on face of the 2i2.

 

[maxm272]

Great, thank you! Alternatively, connecting XLR adapters to my RCA cables wouldn't make much sense, would it?

 

[Xyrium]

Great, thank you! Alternatively, connecting XLR adapters to my RCA cables wouldn't make much sense, would it?

It would be the same as connecting trs to real RCA assemblies. I think you'd be fine either way, but ts to rca is the way to go if it accepts TS connections.

Edit: Corrected the "corrections" made by my phone....sigh...

 

[Blumlein 88]

Great, thank you! Alternatively, connecting XLR adapters to my RCA cables wouldn't make much sense, would it?

The 2i2 XLR inputs are for microphones only. So they'd provide gain you don't need. It was already provided by your phono preamp. So TSR or TS to RCA are fine while XLR to RCA is going to cause problems. The XLRs in the 2i2 also would have a rather low input impedance for your phono pre.

In some interfaces the XLR can either be microphone or line level via a switch. In the 2i2 they are microphone only.

 

[Helicopter]

Mic input impedance is 3kOhm, Line input is 60kOhm. Phono outputs normally are designed for 47kOhm.

I thought about making RCA-F to XLR-M with a 44kOhm series resistor, skipping the phono pre, going from TT to mic in and using software for RIAA de-emphasis.

 

[Helicopter]

The 2i2 XLR inputs are for microphones only. So they'd provide gain you don't need. It was already provided by your phono preamp. So TSR or TS to RCA are fine while XLR to RCA is going to cause problems. The XLRs in the 2i2 also would have a rather low input impedance for your phono pre.

In some interfaces the XLR can either be microphone or line level via a switch. In the 2i2 they are microphone only.

Agree. Phono preamp to and RCA-F to TS adapter is the easy way to do it properly.

 

[maxm272]

The 2i2 XLR inputs are for microphones only. So they'd provide gain you don't need. It was already provided by your phono preamp. So TSR or TS to RCA are fine while XLR to RCA is going to cause problems. The XLRs in the 2i2 also would have a rather low input impedance for your phono pre.

In some interfaces the XLR can either be microphone or line level via a switch. In the 2i2 they are microphone only.

I bought a Cambridge Audio Duo based on the review posted here on ASR. However, it has a nominal output of only 300mV, which is why I wondered whether the extra gain with XLR in could be useful. But I guess I'll use TS adapters and adjust the gain manually, using the knobs on the front then.

 

[Helicopter]

I bought a Cambridge Audio Duo based on the review posted here on ASR. However, it has a nominal output of only 300mV, which is why I wondered whether the extra gain with XLR in could be useful. But I guess I'll use TS adapters and adjust the gain manually, using the knobs on the front then.

If you are going to use the Duo for de-emphasis and if you don't want to solder a custom thingy together, then that (RCA-F to TS) is the way to do it.