Got Focusrite Scarlett 2i2 (2nd gen) for measurements. What's next?

AX tech editor
Joined 2002
Paid Member
have you ever open the unit? kind of interesting inside. output connector is filtered by 3 pin filter to earth, even the signal gnd was filtered. when I measure op amp voltage relative to earth, +-5V, relative to signal gnd is +-2.5V. Don't know if I remember it right, but it puzzles me.

Very often the opamps run on a single supply of +5V. The output signal is therefor centered on half the supply, +2.5V. That is also the reason for the max output signal of somewhere between 1 and 1.5V RMS.

Jan
 
Just another Moderator
Joined 2003
Paid Member
Your circuit works super duper, it eliminates those weird spikes. can't thank you enough.

Glad it works well for you! :) Note that I'm pretty sure I did some experiments later increasing the 2K resistors to higher values of 10K or so to make the opamps life a bit easier. I was worried using higher values would make the noise floor higher but it wasn't an issue if I recall. I still haven't done the verro board version :rolleyes:

edit: here was where I discussed a bit more. https://www.diyaudio.com/forums/pass-labs/324902-p3-adjustment-average-hobbyist.html#post5488717 Note that something went wrong with my test setup, my multimeter went on the fritz and my scope went wacky and I never got back to doing the measurements I was going to. Something for the new year perhaps.

Tony.
 
Last edited:
Further to my post #76 on the previous page, I have now graphed the information I plotted.

It can clearly be seen how the input voltage on the 2i2 affects THD and noise performance.

This is with a single ended input only.
 

Attachments

  • distortion.png
    distortion.png
    18.3 KB · Views: 608
  • noise.png
    noise.png
    17.1 KB · Views: 600
Thanks Tony - I ended up doing the graphs in LibreOffice, as I just couldn't seem to get them looking how I wanted them to in MS Office (2007)

I've also been following your comments in the other FocusRite thread with interest - and plan to build your suggested single ended to balanced converter, to use on the input side.

I did buy this balanced to single ended converter to use on the output of the Focusrite, however I was plagued with IMD issues when using it.

Additionally, I have found that the output stage of the FocusRite is a tad noisy, and suffers from distortion as well.

I have been using a DAC as my audio source (instead of the FocusRite output) for my test signal(s) as I find both the noise and THD is lower than that of the FocusRite output stage - it's definitely optimised for recording.

Finally, I recently purchased an Asus Xonar D1 sound card, and it beats the FocusRite hands down for s/n ratio and THD - but the bandwidth above 48kHz sampling rates sucks.

I've added test results for the Asus sound card - easy to see how they beat the FocusRite....
 

Attachments

  • WaveSpectra.png
    WaveSpectra.png
    57.3 KB · Views: 582
  • 2018-12-19 15_03_06-Test results.png
    2018-12-19 15_03_06-Test results.png
    14.5 KB · Views: 560
Last edited:
Hi there,
I'm new in the measurement side of audio equipment but I have a scarlet 2i2 2nd gen and I want to measure my fully balanced vacuum tube OTL (transformer-less) amplifier that I designed and built (monoblock).
Can you guys tell me what software to use and maybe if possible a schematic of the attenuator you use for the input so you don't fry it?
Best regards,
Silviu
 
Last edited:
I don't know a great deal about valve stuff, however if it's fully balanced, then does that not mean that both speaker terminals are effectively 'hot' ?

Which in turn means you can't connect either side to ground?

The FocusRite does have a balanced input, but I'm not sure just how good the isolation is...

As for an attenuator - you could use a dummy load with a 0.5ohm (or so) resistor to tap off (making a potential divider part of your load), or a potential divider with suitable wattage / values.

See the link for a calculator.
 
Hi there Audio_Tony, the amplifier is fully balanced (circlotron topology) with no feedback. I have build the voltage divider but aded before it 1uF/630V decoupling caps to avoid any dc from the amp. I do my measurements with ARTA.

Everything seems to be ok.
What’s the best (distortion wise) input level in mV?

Do I need to have twisted and shielded cables for the xlr going in the amp and coming out to the sound card (now I have 2 single shielded cables one for hot and one for cold going to the amp input and an unshielded but twisted coming from the tube drivers (85Vrms) or output (40Vrms).

Regards,
Silviu
 
<snip>
What’s the best (distortion wise) input level in mV?

Do I need to have twisted and shielded cables for the xlr going in the amp and coming out to the sound card (now I have 2 single shielded cables one for hot and one for cold going to the amp input and an unshielded but twisted coming from the tube drivers (85Vrms) or output (40Vrms).

Regards,
Silviu

Assuming you are using the balanced input of the Focusrite then that sounds ok, and twisted pair cables should be fine in a balanced configuration.

See this post for best input voltage vs distortion - this post has additional graphs.

For the best distortion measurements, use ARTA STEPS (provides a swept distortion graph).

You could also use WaveGene / WaveSpectra, however you would need to perform 'spot checks' of distortion at different frequencies - but I always find it useful to use at least two measurement methods, as it provides a 'sanity check' to show you are getting the correct results.
 
Last edited:
Hello,
I am new in measurements and I have many questions.
A few days ago I bought Focusrite 2i2 2gen. card. I installed drivers and I installed RMAA software.

On the computer I set "playback" and "recording" devices to "Focusrite".

In the RMAA software, in the "test options / sound card" section for drivers I checked the "driver model - MME" and I checked "use WDM drivers (if posible). Are they good setups?

The problem is when I measure anything below the sempling of the 192Khz frequency, then I can not set the signal level and always writes "the input level is low." and a red square comes out. Only for the sempling frequency of 192Khz I can see the "record level", I can not see everything below that frequency. The card performs measurements but I can not raise the signal level either by "master" potentiometer or "gain" potentiometers. Does anyone know why this is happening?
You can see in pictures.

I hope someone knows what it's all about and you help me. Thank you!
 
I want to wake up this thread again, since I got a Scarlett Solo Gen3, and it seems to be similar to the 2i2. I'm planning to use it mostly to measure unbalanced amps, so I think I will mostly use one channel in and one out. However, balanced in/out is new to me, so that is where I'm looking for advice. I did some quick loopback tests yesterday, and it all seemed ok using a balanced loopback. Connecting one of the signals to ground however made distortion jump to unacceptable levels. I also tried using a resistor (abt 400ohm) to connect one (or the other) signal to ground, and that seemed to have little impact on the performance.

I see talk about balanced to unbalanced converters, but if possible, I would like to keep it as clean and simple as possible without influencing performance significantly. I would rather take 3dB more distortion or noise, than have extra op-amps, cables, power supplies or batteries that run empty when least expected. I think distortion components below -100dB are not so relevant/audible.


I have not yet tested measuring an amp, but hope to do something during the day.

I hope you can share your experiences in the matter, especially when it comes to simple workarounds for balanced/unbalanced.
 
Just looking at the Scarlett Solo and 2i2 3G specs and they show an instrument input with 1.5Meg impedance. No earlier generation specs on the Focusrite site identify the instrument input impedance.

A significant virtue of having a 1Meg input impedance is the ease of using a 10:1 and 100:1 generic scope probe for bench work without the need for any further hardware setup than two metal connectors (6.5mm to RCA, and RCA to BNC), and a metal grounding clip to whatever shielding arrangement is useful.

Has anyone seen a spec for earlier generations, or tried using a scope probe with the Gen3's ? Sometimes the scope probe compensation adjustment runs out, requiring a bit of smt surgery to either remove the installed shunt cap (not the variable cap), or increase its value, to get the best raw spectrum response (before using software calibration to flatten the response).

The earlier gens are staring to show up cheap, and seem reasonable for a benchtop test tool for amplifier work.