Focusrite Scarlett 2i2 Gen 3 Performance Measurements

[mediatechnology]

I've been using the Focusrite Scarlet 2i2 as my test and measurement front-end and found a quirk that users should be aware of.

The left channel of my 2i2 has very poor common mode rejection:



The left channel CMRR measures 13.5 dB; the right channel is almost 48 dB.

With the input cables shorted the difference in hum signature is pretty large:



The input controls are at the 10 o'clock position which is unity gain.

The poor common mode rejection of the 2i2's Left input makes a THAT1646 balanced output appear noisier than it is.
Both channels are connected to the same 1646 output:



The left channel measures about 13 dB higher because the common mode noise in the 1646 output is being converted to a differential signal by the 2i2's "bad" input. (FWIW a DRV134 or SSM2142 would show similar characteristics due to all three parts having high CM gain.)
The right channel is near the noise floor of the 2i2 - the 1646 noise is a few dB lower.

When the 1646 OutSmarts circuitry is defeated the common mode output noise drops and the noise measurements are close to one another:



I don't know if I have defective 2i2 or not.
It seems hard to believe that component tolerance could be bad enough to make CMRR that low.

I've got a second 2i2 on order and am anxious to test it.

I would recommend everyone check the CMRR of their 2i2 inputs because it introduces measurement errors with active balanced outputs and has greatly reduced hum rejection.

[mediatechnology]

I received a second Focusrite Scarlett 2i2 and it has good common mode rejection:



Left channel CMRR 68 dB; Right channel 58 dB.

I'll use the first unit in an unbalanced environment and use the "good" 2i2 on the work bench.

[mediatechnology]

I've been using the Focusrite 2i2 recently to make twin tone (19/20 kHz 1:1) IM distortion measurements.

My area of interest is the f2-f1 IMD component at 1 kHz.

This is the loopback IMD performance with dual tones generated by ARTA.
Both tones are "internally" summed rather than producing 19 kHz on the left and 20 kHz on the right with passive summation.
Internal summing provides an indication of the entire chain of D/A, output stage, input and A/D.

The output level was +10 dBu RMS.

The first measurement is with a 10 dB pad in the loopback path in order to operate the input stage at unity gain and near the A/D's 0 dBu, -9 dBFS, "sweet spot."



This is with no pad and the input level reduced to accept the +10 dBu output.



The overall IMD performance of the 2i2 is pretty good.

In another post I'll explain what I've been measuring.

[mediatechnology]

I've found different IMD results between channels for my test and measurement Focusrite 2i2.

With a 0 dBu output 19/20 kHz 1:1. IM figures are f2-f1 (1 kHz)

Left Channel Output feeding Left and Right Inputs.

L = -115 dBc
R = -124 dBc

Right Channel Output feeding Left and Right Inputs.

L = -114 dBc
R = -120 dBc

The left channel A/D has about 6-9 dB more IM than the right.

For the lowest IM measurements I'll use the Left Channel Output feeding the Right Channel Input for a residual level of -124 dBc.
In the right channel there could be some cancellation between Left D/A and Right A/D making it measure slightly better than it is.
There's some obvious uncertainty: -120 dBc is probably a safe limit.

I may flip the polarity of the inputs to see if the results change...

[mediatechnology]

I may flip the polarity of the inputs to see if the results change...

And they do change. Significantly.

The best combination is consistently about -120 dBc IM2.
If I reverse the polarity it rises to about -105 dBc showing that the D/A and A/D are cancelling/adding to one another.

Today I generated 19 kHz on the Left output and 20 kHz on the Right.
I resistively summed them using 1KΩ series resistors.

At 0 dBu I now read:

Left -119 dBc
Right -114 dBc

The results are now that of the Focusrite 2i2's input stage and A/D in isolation and are consistent regardless of polarity.

I think I now have a better idea of what the minimum measurement capability is.

[mediatechnology]

Focusrite 2i2 Gen 3 192 kHz Sample Rate Loopback Frequency Response

I typically use the Focusrite 2i2 with a 96 kHz sample rate and have never swept the analog hardware in loopback at 192 kHz to see what the analog filter limitations might be.
Today I measured the loopback high frequency -3 dB point to be about 58 kHz.
In another measurement, not shown, I found the 20 Hz to 20 kHz response to be within ±0.05 dB with a 96 kHz SR.
The 2i2's input is AC-coupled: The loopback low frequency -3 dB point appears to be below 1 Hz which is outside of AudioTester's range.
The -1 dB point is approximately 2 Hz.


Focusrite 2i2 Gen3 192 kHz Loopback Frequency Response

[mediatechnology]

There is now a Focusrite Scarlett 2i2 Gen 4: https://us.focusrite.com/products/scarlett-2i2#

A few improvements:

Line Input DR now spec'd at 115 dB A-weighted (2i2 G3 Measured 104 dB unweighted.)
Line Output impedance 100Ω. (Was 300Ω/leg.)
Has external USB-C power input in addition to bus power. 900 mA requirement.

Neither my dealer nor Amazon have Gen 4 in stock as of today. Focusrite's store is quoting 4 day shipping.

I think the extra 8-10 dB of DR may make getting one worth it for Test and Measurement.
That frees up a 2i2 Gen 3 for use on my Linux machine.