Hi,
I've been using the Focusrite Fcarlett 2i2 2gen interface for measurements for some time now.
The card seems to measure well. All the results from RMAA are at a similar level to the results I found online. I am posting a THD loop-measurement made with the RMAA.
I recently connected the output of the card to an oscilloscope (one of the differential outputs or the output from the differential to single ended shorting adapter - no difference).
To my surprise I saw a very noisy waveform. I'm posting screenshots from the oscilloscope for a 1kHZ -0.5dBFs waveform and for the output with no signal. As you can see, the noise reaches 200 mVp-p. I also include FFT spectra of these signals, where you can see that the noise spectrum is shifted to 30-40kHz which explains the good measurements in RMAA.
Is this normal for this interface?
I took measurements with the card connected to a laptop powered from the mains and battery - no differences
I will add that the output signal from other sources I own is clean in the same measurement setup .
Could someone confirm these results on their unit?
Marek
I've been using the Focusrite Fcarlett 2i2 2gen interface for measurements for some time now.
The card seems to measure well. All the results from RMAA are at a similar level to the results I found online. I am posting a THD loop-measurement made with the RMAA.
I recently connected the output of the card to an oscilloscope (one of the differential outputs or the output from the differential to single ended shorting adapter - no difference).
To my surprise I saw a very noisy waveform. I'm posting screenshots from the oscilloscope for a 1kHZ -0.5dBFs waveform and for the output with no signal. As you can see, the noise reaches 200 mVp-p. I also include FFT spectra of these signals, where you can see that the noise spectrum is shifted to 30-40kHz which explains the good measurements in RMAA.
Is this normal for this interface?
I took measurements with the card connected to a laptop powered from the mains and battery - no differences
I will add that the output signal from other sources I own is clean in the same measurement setup .
Could someone confirm these results on their unit?
Marek
Attachments
Isn't what you are seeing the actual self noise of the scope? They are not very clean and only 8 bit resolution systems.
//
//
Its the ultrasonic noise of the audio interface bunching up and showing because the scope sees up to several MHz.
I thought so too, but then how do you explain the fact that the waveforms from the function generator and other audio DACs are clean? I can post them here later when I get home.
None of the above, this is how noise-shaping DACs work, they push the quantizatiobn noise up into the ultrasonic region where it can be filtered out or simply ignored. In fact they generate more noise that other types, but less noise in the audio band. The noise spectrum is "shaped" by clever high-order delta-sigma modulation schemes.
That makes sense. This can also be seen on fft spectra. So the type of converter plays a role and this behavior is normal....
However, just to be sure, could someone take a screenshot of the waveforms from their scarlet 2i2 to compare?
I don't have a picture but I have seen a friend's Scarlet doing like that on a scope. That's what I meant with ultrasonic noise of the interface, the quantization noise. I have an older other brand interface which shows up even worse on a scope. If there's option to set a low pass digital filter in the tens of kHz region in the math menu of your scope you will see the noisy line getting thinner.
Thank you all for your input.
By the way, Has anyone tried 4th gen focusrite? is it better than 2nd gen for mesurement?
Or is there any other cheap measurement interface currently recommended?
By the way, Has anyone tried 4th gen focusrite? is it better than 2nd gen for mesurement?
Or is there any other cheap measurement interface currently recommended?
I recommend the Motu M4. Can reach 106dB Sinad from its back panel line inputs. Not cheap but not too expensive either.
Thank you Mark.
It looks similar although in your case the noise is about 2 to 3 times less amplitude
It looks similar although in your case the noise is about 2 to 3 times less amplitude
Thank you. btw. Nice filtering feature on your Rigol. Unfortunately my my Siglend doesn't offer this.
The new 12 bit DHO 800 economic series has such a math menu feature too among many. Here's the EMU's output signal again.
