Focusrite Scarlett 2i2 (4th Gen) Interface Review

[SMJH]

@bennybbbx Response from Focusrite support confirms what you said:
"Thanks for getting in touch about the 4th Generation of Scarlett interfaces.
The 18i20 features a digitally controlled monitor output, 2 digitally controlled headphone outputs and digitally controlled analogue inputs.
The 18i16 and 16i16 features a digitally controlled monitor output and digitally controlled analogue inputs.
The 4i4 and 2i2 only feature digitally controlled analogue inputs.
The Solo uses all analogue-controlled inputs and outputs."

By the way, Apollo has the ESS IMD hump. Don't know if it's audible, but either way I cannot justify the purchase of hardware I will never use, such as extra audio inputs, or internal UA VST processors.
And as much as I would love importing SMSL or Topping DACs, I cannot risk it again due to poor QC.

Focusrite support adds:
"In our experience, this left-right imbalance is generally found only at the minimum of the pot travel as soon as the volume is increased to a useable level, this imbalance is generally not perceivable."

I will invest in the cheapest possible 4th Gen Solo. It will last me a couple of years at least until I find something better or go for a surround setup...

 

[bennybbbx]

Focusrite support adds:
"In our experience, this left-right imbalance is generally found only at the minimum of the pot travel as soon as the volume is increased to a useable level, this imbalance is generally not perceivable."

I will invest in the cheapest possible 4th Gen Solo. It will last me a couple of years at least until I find something better or go for a surround setup...

when you need no inputs you can use hifi DAC. they have a position indicator in that way that they show db values in display. it depend mostly how often change volume and how loud hear and which pot position use . how often do you think you change volume per day or week and how loud you hear ? because the focusrite have a input you can measure channel unbalance. plug the output of left channel to left input channel measure and then plug the right output channel to leftr input and measure too. then can do this 2 months later, or 6 months later. maybe should a own thread with audio examples how it sound when left channel is 0,5db less than right channel. only when diffrence is 2 db or more it can clear hear that it is out of mid. but also on small diffrence there is strange stereo image which can force to make louder that it sound better. make louder to have good sound is no good idea, because you can not buy new ears.

from presonus there is now a digital control 2 in 2 out interface under 200$. it have also a position LED. really sad that focusrite do this not in the 2 in 2 out.

 

[bennybbbx]

here i have done a small testpiece to test if can hear 0,7 db louder right channel. In studio one set pan R8 mean 0,7 db right is louder. I verify this with the moscilator which output a sine tone and look at the level meter how much is output in left and right channel. in the screeshot see left channel output -14 db. right channel -13.3 db. this mean right channel is 0,7 db louder. for me it is very much hearable that guitar sound strange and no room feeling as with the center version. if possible try the audio examples on a headphone which is digital control. the lower the hear volume is when you play the more is the diffrence hear also with digital control. when you have no digital control headphoen output, then you can hear diffrence between the 2 versions. So in fact hear loud make a better sound on worse pots, but is not good for ears. on digital control sound good when loud or not loud. balance compensation help not, because worse pots have depend on position and range use mostly diffrent left right balance and this change after 1000 pot moves 0,7 db easy. MOscilator can free download and you can test in a DAW yourself how much db diffrence Pan R8 have. https://www.meldaproduction.com/MOs...iBfxIzMky20BIwD4bcEwqQ9BB8WZ58wuoPVS57oYuAi1f

EDIT: testfile i remove because vst instruments use round robin and timing randomisation for humanize. i upload a new one later

 

[SMJH]

@bennybbbx Currently I have digital volume. (straight from PC motherboard line out) So I am using Jriver internal 64bit volume -> Jriver DSP -> motherboard line out -> Adam T8V with their gain knobs set to maximum.
I can control the volume of the L and R channels in Jriver DSP.

For my experiment, I set the audio to dual mono (so both channels get the same signal), then applied negative volume change on one channel. I could not notice -0.7 dB, but at -1 dB I can hear it no matter the channel and no matter the output volume.

Let's hope the Focusrite pot is not so bad. I'll update the thread when I get the interface. But I am going on summer vacation away from home anyway, so I think this will have to wait at least 2 months. Maybe I'll find a better interface during that time too...

 

[bennybbbx]

For my experiment, I set the audio to dual mono (so both channels get the same signal), then applied negative volume change on one channel. I could not notice -0.7 dB, but at -1 dB I can hear it no matter the channel and no matter the output volume.

in stereo left to right diffrences are more hearable as strange sounding reverb and clarity. even more on headphones when use a speaker simulate for headphones as realphones

 

[SMJH]

in stereo left to right diffrences are more hearable as strange sounding reverb and clarity. even more on headphones when use a speaker simulate for headphones as realphones

Interesting, can you recommend songs you hear this reverb and bad clarity in? I did dual mono because the center "ghost" image shifting is all I hear when I change the left-right balance.

What you are describing sounds like phase shift, possibly caused by realphones algorithm when you feed it imbalanced audio.

 

[bennybbbx]

Interesting, can you recommend songs you hear this reverb and bad clarity in? I did dual mono because the center "ghost" image shifting is all I hear when I change the left-right balance.

What you are describing sounds like phase shift, possibly caused by realphones algorithm when you feed it imbalanced audio.

have you test the centertest.zip audio wav test i upload in stereo ? . there is original and the 0,7 db wav. such problems can hear on most songs that are not very dry mix and have reverb in the record as classical music, rock, pop or metal. the centertest.zip contain wav that use a double tracked guitar. it is as 2 guitar players play 1 guitar player play on left 1 guitar player play on right. this is used on many songs. the diffrences in the play make the guitar wide it is called wall of sound. when left and right channel play exact same then it is small and in middle. when do mono and left and right channel are not same level, there can hear before it go out of middle as if there come reverb from left or ride side. so if you want test mono then listen to reverb more left that you hear not in the center version

it also it is good when use headphone to test so position of left, right to ears is same. try also lower levels 60 db or so. the 0.7 db i hear without realphone algorithm. wqith headphones you also can hear better when the channel inbalance create a fake reverb tail. the fake reverb tail cause loss on clarity. you can also use sine tone mono with no reverb to hear this.

 

[SMJH]

have you test the centertest.zip audio wav test i upload in stereo ? . there is original and the 0,7 db wav. such problems can hear on most songs that are not very dry mix and have reverb in the record as classical music, rock, pop or metal. the centertest.zip contain wav that use a double tracked guitar. it is as 2 guitar players play 1 guitar player play on left 1 guitar player play on right. this is used on many songs. the diffrences in the play make the guitar wide it is called wall of sound. when left and right channel play exact same then it is small and in middle. when do mono and left and right channel are not same level, there can hear before it go out of middle as if there come reverb from left or ride side. so if you want test mono then listen to reverb more left that you hear not in the center version


it also it is good when use headphone to test so position of left, right to ears is same. try also lower levels 60 db or so. the 0.7 db i hear without realphone algorithm.

I tried your center test files, and could hear a difference in the guitar reverb and snare attack, but it is so small I do not know if I am imagining it.
This is why I tried doing a null test next, to check for phase issues.
For some reason your files are different enough to not output an empty left channel, which should be the case if only a right +0.7 dB change was made. I also tried volume-matching the left channel then doing null test but it also failed, so in the end I simply made my own +0.7 dB Right file, and that passed the null test. So, what's up with that?

I am sharing a new zip that contains 3 additional files:
- My new +0.7 dB Right file.
- Original null test: output of the original center file with phase inverted original +0.7 dB Right file.
- New null test: output of the original center file with phase inverted new +0.7 dB Right file.

Let me know if you hear that issue with the new file, and if you have an explanation for the failed null test. The samples look aligned to me, and the volume has been matched so that the left channel has equal peak value between all files.

 

[bennybbbx]

Let me know if you hear that issue with the new file, and if you have an explanation for the failed null test. The samples look aligned to me, and the volume has been matched so that the left channel has equal peak value between all files.

In your test are volume and bass much diffrent to original. I not understand how you change the files. when play this on player in mono the left right diffrence is lost from the file. so you need always play in stereo to do blind test there is foobar 2000 player the abx comparator

foobar2000: Components Repository - ABX Comparator

foobar2000

www.foobar2000.org

after install can mark 2 files and right click for popup menu over the selected files then choose utility->ABX tracks

I have do new testfiles with sine tone in middle and +0.7 db right. 1 mistake i make at 5.th test but it is clear hear. this output abx . you see also in levelmeter that can show 0.1 db thatb file is correct

foo_abx 2.1 report
foobar2000 v2.0
2025-06-06 19:38:57

File A: sine tone center.wav
SHA1: 18af69c586ff0d72871fd32a9901f1791cecb336
File B: sine tone right +0.7.wav
SHA1: 88b1b10cc6893fbc35e5f15533289e1758eca357

Output:
Default : Primary Sound Driver
Crossfading: YES

19:38:57 : Test started.
19:39:21 : 01/01
19:39:35 : 02/02
19:39:55 : 03/03
19:40:13 : 04/04
19:40:32 : 04/05
19:40:54 : 05/06
19:41:09 : 06/07
19:41:26 : 07/08
19:41:26 : Test finished.

----------
Total: 7/8
p-value: 0.0352 (3.52%)

-- signature --
7a07ca16c63cbb1b1de4dcce67a7bc4aeafd2b6b

also need not underestimate ear tire after longer hear or test with bad balance

 

[SMJH]

In your test are volume and bass much diffrent to original. I not understand how you change the files. when play this on player in mono the left right diffrence is lost from the file. so you need always play in stereo to do blind test there is foobar 2000 player the abx comparator

Did you try to ABX your original centertest file (with no +0.7 dB) with my new +0.7 dB file?

The files marked with 'null' will of course have missing frequencies. Are you familiar with null tests?
I shared with you the output of 2 null tests, marked with 'new' and 'original'. These null test outputs show you the difference between the input files:

- 'original' null test file = difference between your centertest file and your +0.7 dB file.
It is showing there is a difference in guitar and snare, this is not normal.

- 'new' null test file = difference between your centertest file and my new +0.7 dB file.
It is showing that the only difference is in the Right channel, this is normal.

How I created the files:

- Null test files: using Audacity "invert" phase tool.

- New 0.7 dB file: Simply modified the original centertest file by adding +0.7 dB to the Right channel in Jriver DSP. Nothing else.

 

[bennybbbx]

Did you try to ABX your original centertest file (with no +0.7 dB) with my new +0.7 dB file?

The files marked with 'null' will of course have missing frequencies. Are you familiar with null tests?
I shared with you the output of 2 null tests, marked with 'new' and 'original'. These null test outputs show you the difference between the input files:

- 'original' null test file = difference between your centertest file and your +0.7 dB file.
It is showing there is a difference in guitar and snare, this is not normal.

- 'new' null test file = difference between your centertest file and my new +0.7 dB file.
It is showing that the only difference is in the Right channel, this is normal.

How I created the files:

- Null test files: using Audacity "invert" phase tool.

- New 0.7 dB file: Simply modified the original centertest file by adding +0.7 dB to the Right channel in Jriver DSP. Nothing else.

ah now i understand. my mistake was, i export original in DAW and change the pan value and export in DAW . I forget that the VST instruments i use use round robin and humanisation, so it sound slightly diffrent in timing and pitch . but the result of worse pan is the same I check your file with ABX. 1 error i get . maybe it is more easy to detect when i remove crash on right side when play start, because this crash, sound wide because of reverb. I upload later a new testpiece. the correct way is export the original in DAW, then load it in change pan and export it

foo_abx 2.1 report
foobar2000 v2.0
2025-06-07 09:53:59

File A: testpiece center.wav
SHA1: 9b0a04ac7f538e0c629b959ba4d33ca17b91add8
File B: testpiece center - new right +0.7 dB.wav
SHA1: 87ef94556759e8098e1273b73fda89cb3a54f006

Output:
Default : Primary Sound Driver
Crossfading: NO

09:53:59 : Test started.
09:54:21 : 01/01
09:54:40 : 02/02
09:54:51 : 03/03
09:55:01 : 04/04
09:55:17 : 05/05
09:55:43 : 06/06
09:55:56 : 06/07
09:56:06 : 07/08
09:56:06 : Test finished.

----------
Total: 7/8
p-value: 0.0352 (3.52%)

-- signature --
173be6fa8cb59f04488480a6047eb291b5b88b66

 

[bennybbbx]

I can now not edit old posts so sorry for another post about this. I do new version which is now 0,9 db on left louder(left because all my audio devices get with wear and tear of analog pots left more loud. I measure the foobar 2000 stereo balance DSP effect how much db left to right change depend on number setting, so can easy test songs you like how disturb you are on left right diffrences

foobar2000: Components Repository - Utility DSP Array

foobar2000

www.foobar2000.org

foobar Stereo balance value to db diffrence left to right
Center = 50

45 = left + 1,5 db
46 = left + 1,3 db
47 = left + 1,0 db
48 = left + 0,6 db
49 = left + 0,3 db

EDIT: I measure the balance settings from foobar this way : I play a sine tone in foobar and use the sessionwire as output device in windows. in the DAW i use the VST3 plugin from sessionwire sessionwire recieve. then i read the output when the sine tone play on the DAW output meter and calc diffrence. so audio is digital transfer and no diffrences of left right due to AD converter can happen

 

[SMJH]

I got myself a 4th Gen 16i16. Sounds wonderful, and we can assume its headphone volume knobs are digital despite having a white indent line. I can obviously hear digital volume 'clicks' when adjusting them. What's bad with it is latency in non-ASIO apps, so things like web browsers and games. Setting its sampling rate to 192Khz does result in acceptable latency, but on-board realtek audio is still faster... Overall, not a big deal, I am very happy with the performance and versatility.

 

[bennybbbx]

I got myself a 4th Gen 16i16. Sounds wonderful, and we can assume its headphone volume knobs are digital despite having a white indent line. I can obviously hear digital volume 'clicks' when adjusting them. What's bad with it is latency in non-ASIO apps, so things like web browsers and games. Setting its sampling rate to 192Khz does result in acceptable latency, but on-board realtek audio is still faster... Overall, not a big deal, I am very happy with the performance and versatility.

the 16i16 headphone is analog because of the white ident line.

you can see it too in the answer from focusrite you post

The 18i16 and 16i16 features a digitally controlled monitor output and digitally controlled analogue inputs.

it have only digital control analog inputs and digital control monitor outputs. So the 2 headphones are analog control

the digital control have no ident line on pot use LED or on hifi digital DAC a - db value.
The latency with windows audio can get better. see on right side in windows the tray of the focusrite. press over it right menu button and you get menus. there can choose which outputs can use with windows and it is possible to enable disable a buffer addition. you can disable it if you get no crackle

 

[SMJH]

the 16i16 headphone is analog because of the white ident line.

you can see it too in the answer from focusrite you post

it have only digital control analog inputs and digital control monitor outputs. So the 2 headphones are analog control

the digital control have no ident line on pot use LED or on hifi digital DAC a - db value.
The latency with windows audio can get better. see on right side in windows the tray of the focusrite. press over it right menu button and you get menus. there can choose which outputs can use with windows and it is possible to enable disable a buffer addition. you can disable it if you get no crackle

The volume behavior is very clearly digital, I can hear the steps clear as day. I think Focusrite support is talking about the ability to control digital volume via software (Focusrite Control 2). I am contacting them again to confirm.

 

[MaxwellsEq]

The volume behavior is very clearly digital, I can hear the steps clear as day

Could it be a relay-switched analogue resistor network. There are a number of analogue attenuators based on this concept - the audio is not digitised in the process.

 

[SMJH]

Could it be a relay-switched analogue resistor network. There are a number of analogue attenuators based on this concept - the audio is not digitised in the process.

I'll bring that up to support, but I cannot hear any relays from the unit, apart from when switching sampling rates.

 

[MaxwellsEq]

I'll bring that up to support, but I cannot hear any relays from the unit, apart from when switching sampling rates.

It may also be just a stepped fixed resistor attenuator like the Dact types? Or a shaft encoder to something like a PGA231?

 

[SMJH]

It may also be just a stepped fixed resistor attenuator like the Dact types? Or a shaft encoder to something like a PGA231?

No clue, but a digital volume would be the most cost effective.
As for the indent, a lot of encoder knobs can have them too. The Tonex pedal comes to mind, as well as Focusrite's own Clarett+ range. See here:

While we wait for support, here is everything I noticed:
- the volume steps are more easily noticeable at low volumes.
- the volume clicks are noticeable throughout the entire volume range when playing a low sine wave test tone (the clicks are high in pitch, so a low Hz sine wave does make them easier to notice)
- the volume clicks coincide perfectly with the audible volume steps.
- the clicks and steps happen in both directions.

I still have to check whether or not these are audible without any signal playing, to be fair I was testing the CS43198 for class H quircks, the knob findings were unexpected.

 

[MaxwellsEq]

No clue, but a digital volume would be the most cost effective.
As for the indent, a lot of encoder knobs can have them too. The Tonex pedal comes to mind, as well as Focusrite's own Clarett+ range. See here:

While we wait for support, here is everything I noticed:
- the volume steps are more easily noticeable at low volumes.
- the volume clicks are noticeable throughout the entire volume range when playing a low sine wave test tone (the clicks are high in pitch, so a low Hz sine wave does make them easier to notice)
- the volume clicks coincide perfectly with the audible volume steps.
- the clicks and steps happen in both directions.

I still have to check whether or not these are audible without any signal playing, to be fair I was testing the CS43198 for class H quircks, the knob findings were unexpected.

All of the discrete resistor volume controls such as the Dact will create an audible click when switching. Unless you use some clever zero-crossing detection algorithm.