It's a tested conclusion, not a misconception.
Now here's an actual misconception. There's no real correlation between voltage in the analog domain and decibels from full scale in the digital domain. There are reference points accepted as common such as -12, -14 and -18, but they're not actually a true correlation since they can all be an option depending on the use case. You can calibrate your analog gear to whatever digital reference you want and it's entirely preferential to the end user and the particular gear in question.
All right, I should have stated that it *usually* works out at -18dBFS (and it's more a question of how the ADC/DAC is calibrated). The manufacturer will usually (and should) state where the sweet spot (0VU) is. I have an older interface too (and actually the one I use the most) and they state 0VU as being -12dBFS IIRC. This is mostly academic, hey? If you're recording rock or pop, it is all mostly immaterial, you're going to get much more noise from your guitar/bass (amp) and you're going to compress and saturate everything and anything.
Of course they have SNR, and measurably in many, if not most consumer-tier products such as the Scarletts being referenced here the noise is appreciably higher relative to the signal when operating at or near maximum available gain.
Which is telling us that whatever is before the gain pot has a certain noise, and if you have the choice to get a stronger signal in (use a more sensitive microphone, for instance, or get it up closer) as opposed to cranking the gain, by all means do so, unless you need that specific character or you have no real choice (I'm thinking about passive ribbon microphones or field recording), because that way you are maximizing the SNR before the gain pot. Also, according to measurements I've seen on the web (Teardown, few basic measurements and personal thoughts of the Focusrite Solo Gen3) cranking the gain pot does cause a 2dB rolloff at 20Hz (and a 1-ish dB rolloff at 40Hz - honestly, inconsequential), which means that they probably chose a coupling capacitor somewhere that is just below the ideal value for the circuit as used in the worst case scenario and that they do not expect the gain pot to be ever maxed. But, again, if you have no other choice, then you most probably are better off maxing the gain instead of keeping it lower and adding gain digitally, because that "free" added gain is free for a reason, and that reason is that there is no (zero, nil) added data, and if you have a signal that tops out at -30dBFS and you need to bring it up to -10dBFS, say, that 20dB boost is going to be, well, the converter's own noise 20dB higher and that's it.
Literally, all the gain pot does is lower the signal of whatever is coming into it. So you're not taking the noise out, you're just lowering its level together with the rest, *and* you're adding the gain pot's own noise because when it's maxed, the gain pot's noise is mostly shunted to ground, when it's not it seeps into the audio signal. In other words, the gain pot in 99.99% of the cases is just a volume pot. When isn't it a volume pot? In a Tubescreamer, for instance. Or possibly in a Rangemaster, or is it a Tonebender?, since the pot, IIRC, is wired as a variable resistor on the transistor's collector. (The collector resistor doesn't really set the gain, it's an interaction between Rc, Re and Rb, to an extent, but you get the idea).
Last edited:
