Scarlett 2i2 vs 18i8 gain range

[Wonks]

But your recording space will still be far noisier, so unless you are recording something quiet a long way away with a dynamic mic, then the space will be far noisier than any mic or pre-amp noise if you've used sensible input gain levels.

[Hugh Robjohns]

So all things considered, I should get better (less noisy) recordings with the 2i2 then (especially with gain hungry mics and quiet sources)?

I doubt you'd actually be able to tell any difference in 'the noise' between the two interfaces in normal use simply because, as Wonks points out, the ambient acoustic room noise is likely to far outweigh the electronic noise floor contorbuted by the preamps.

Okay, so it appears that the EIN noise floor specs are lower for the 2i2, and it would be logical therefore that this model should be notionally quieter (electronically) for a given digital output level.

However, the important fact to remember is that neither of these interfaces is designed or intended for use with gain-hungry mics or quiet sources; they both have too little gain for that, quite intentionally. They're actually intended for common close-miking techniques on vocals, guitars etc, in typical project studio applications with common dynamic and capacitor (condenser) mics -- and from the feedback I've received they both seem to meet those needs pretty well.

H

[fladd]

So all things considered, I should get better (less noisy) recordings with the 2i2 then (especially with gain hungry mics and quiet sources)?

You could have just looked at the input noise spec to reach the same conclusion. The 18i8 claims an input noise of -122dBu at maximum gain while the 2i2 claims >-127dBu at 60dB gain. While the gain figures may be pretty meaningless in this context as they could send any level they like to the A/D convertor, the noise figures are absolute values that are easily compared.

Mmh, not sure where you got the noise figure of 122 dB for the 18i8 from. The official specs at the Focusrite website quote an EIN of 127 for the 18i8 (https://us.focusrite.com/usb-audio-inte ... 8#techspec). Are these measurements you did yourself?

I also don't quite understand where you got the 60dB measurement for the 2i2 from (it does not appear in the official specs on the Focusrite website, nor does the preamp of the 2i2 deliver 60dB of gain anyway: https://us.focusrite.com/usb-audio-inte ... ifications).

But in any case, I would have certainly expected better specs on the upper range devices...
I also wrote Focusrite support to ask about this issue, and they seem to be confused, too, stating that the preamps should be pretty closely matched.

[fladd]

So all things considered, I should get better (less noisy) recordings with the 2i2 then (especially with gain hungry mics and quiet sources)?

I doubt you'd actually be able to tell any difference in 'the noise' between the two interfaces in normal use simply because, as Wonks points out, the ambient acoustic room noise is likely to far outweigh the electronic noise floor contorbuted by the preamps.

Okay, so it appears that the EIN noise floor specs are lower for the 2i2, and it would be logical therefore that this model should be notionally quieter (electronically) for a given digital output level.

However, the important fact to remember is that neither of these interfaces is designed or intended for use with gain-hungry mics or quiet sources; they both have too little gain for that, quite intentionally. They're actually intended for common close-miking techniques on vocals, guitars etc, in typical project studio applications with common dynamic and capacitor (condenser) mics -- and from the feedback I've received they both seem to meet those needs pretty well.

H

Yes, you are of course right. I also do intend to use the 18i8 in this way. My quietest mic is an SM58. I was just a bit surprised to see this huge difference in (digital) level, as I always assumed that, if anything, the higher range model should deliver a "hotter" signal.

Btw. if anyone is interested, I uploaded the test recordings (since Focusrite support also wanted them). I recorded the output of my mobile phone's built in speakers into an SM58 with the mic and mobile phone position (and volume) constant between the recordings. I merely unplugged the mic cable from one interface and put it into the other. Both interfaces where at max gain. I know this was not the best test scenario, but it was the most constant test I could do where I am right now (not at home), and it should be good enough for measuring the difference in level. Here are the recordings: http://stuff.fladd.de/scarlett/.

[James Perrett]

Are these measurements you did yourself?

I did a very quick Google search and those figures appeared in the text that Google displayed for

https://uk.focusrite.com/usb-audio-inte ... ifications

and

https://uk.focusrite.com/usb-audio-inte ... cification

I realise that I attributed the 60dB gain condition to the wrong device though.

[fladd]

Are these measurements you did yourself?

I did a very quick Google search and those figures appeared in the text that Google displayed for

https://uk.focusrite.com/usb-audio-inte ... ifications

and

https://uk.focusrite.com/usb-audio-inte ... cification

I realise that I attributed the 60dB gain condition to the wrong device though.

Oh, I see. The specs you looked at are from the 1st gen 18i8, though. Mine is a 2nd gen 18i8. Apparently the noise figure has changed.

The 60dB are still strange, since the old 18i8 also only reached 55dB

[fladd]

No response from Focusrite anymore. Guess they are also confused.

[fladd]

Mmh, I now did get an answer from the Focusrite support. They also measured it with the same two devices, sending a sine wave with a generator and got different results.

For the 18i8 (2nd gen), they found the level to be -9.2dBFS. For the 2i2 (1st gen), they found the level to be -5.0dBFS. They then state that this closely matches the spec (not sure which ones exactly).

That is quite some difference to what I measured (4.2dB vs. 10.5dB difference). I will repeat my test in a different way tomorrow. They suggested to measure a sine wave from a third interface connected to the XLR inputs. I don't have a third interface, but I have a Circuit and a JD-Xi, I only don't know if connecting them to the XLR inputs will be safe (since these inputs are meant for much lower level mic signals).

What do you think?

[Wonks]

Got a DI box?

[fladd]

Got a DI box?

No. Is that needed?

[fladd]

I now conducted a further test with all inputs on both interfaces: https://we.tl/Ok3LWDLn3Q.

A sine wave from Z3ta+ running on iPad was recorded through the iPad's speakers into a Shure SM58. In the first run, the sine wave hit the microphone at 76.5 dBA (measured with an external volume meter). In the first run, the sine wave hit the microphone at 88.5 dBA (measured with an external volume meter).

Notably, in the second run, I tried to match the dB reading of the 2i2 in Reaper to that of what Focusrite had in their test (i.e. 5.0dB).

The results are consistent. The 18i8 is around 10.5dB quieter. I am not sure how Focusrite got their results (I wrote them again, so maybe I will know soon).

Is there a way to calculate, based on the measured volume, the specs of the SM58 and the specs of the interfaces, what the expected level in the DAW should be?

[Hugh Robjohns]

Is there a way to calculate, based on the measured volume, the specs of the SM58 and the specs of the interfaces, what the expected level in the DAW should be?

The sensitivity of the SM58 is published by Shure... but it's not actually relevant here as it is common to all your tests, and you are complaining about the different sensitivities of the two interfaces, not the mic in comparison to other mics.

However, 76.5dB SPL at the microphone should provide an electrical output of about -70dBu, and 88.5dB SPL would produce about -58dBu. Those signal levels will be raised by the analogue gain in the interface, and then converted into a digital signal according to the converter alignment as previously discussed.

Unfortunately, there is no way to calculate the digital signal level recorded in the DAW from a given sound pressure level reaching the mic from first principles because, as I explained much earlier, Focusrite don't state the internal alignment of the analogue/digital signal levels through the converters in the two interfaces.

What you can do, though, is measure the interface's overall conversion level -- as you have done -- by noting what digital level (in dBFS) is achieved for a certain analogue input signal level with the user gain control at a certain setting.

I'm still at a loss as to what exactly you're trying to achieve through all this, though.

H

[Matt Houghton]

After writing that I went back and searched for the review on the web site. It was Matt's review of the Scarlett 8 Pre that I was thinking of. However, now that I was able to see all the web links in the review I could trace back through Matt's history of the various Focusrite preamps and I see he mentions the issue in a previous review. This doesn't help those of us who mainly read the print edition but at least I can now see that he has covered the problem previously.

Numbers would still be useful though - I seem to remember in the ancient past that some magazines would have 2 sections to a review with a technical evaluation done by another reviewer - although I could see that could get a bit onerous if Hugh got lumbered with that job for every review.

Point taken, James. As Hugh said above, striking the right balance between being diligent and informative, and being repetitive and long-winded can be tricky. But I do try at the very least to include references/links to relevant previous reviews when I have to keep things brief!

[fladd]

Is there a way to calculate, based on the measured volume, the specs of the SM58 and the specs of the interfaces, what the expected level in the DAW should be?

The sensitivity of the SM58 is published by Shure... but it's not actually relevant here as it is common to all your tests, and you are complaining about the different sensitivities of the two interfaces, not the mic in comparison to other mics.

However, 76.5dB SPL at the microphone should provide an electrical output of about -70dBu, and 88.5dB SPL would produce about -58dBu. Those signal levels will be raised by the analogue gain in the interface, and then converted into a digital signal according to the converter alignment as previously discussed.

Unfortunately, there is no way to calculate the digital signal level recorded in the DAW from a given sound pressure level reaching the mic from first principles because, as I explained much earlier, Focusrite don't state the internal alignment of the analogue/digital signal levels through the converters in the two interfaces.

What you can do, though, is measure the interface's overall conversion level -- as you have done -- by noting what digital level (in dBFS) is achieved for a certain analogue input signal level with the user gain control at a certain setting.

I'm still at a loss as to what exactly you're trying to achieve through all this, though.

H

Thanks for the explanation.

Given the measurements of the Focusrite support, together with the statement that their measurements closely match the specs (whatever these are), one has to come to the conclusion that one of my two Focusrite interfaces is not functioning correctly (since my measurements are off by ~ 6 dB). What I am trying to achieve is to figure out which one.

Of course one could argue that it does not matter, as I can always increase digital gain, but in that case I would not understand what the point of analogue gain is in the first place (or even more so, why most people want a lot of it), because you can always increase the signal in the digital domain after recording it. Why would one need a preamp at all then?

[James Perrett]

Why would one need a preamp at all then?

Most mics have a wider dynamic range than all but the very best A/D convertors can accommodate. In addition, A/D convertors often require a signal of a few volts to achieve full scale. The preamp therefore has to change the output level of the mic to match the input level of the A/D and ensure the optimum gain structure.

Mics with digital outputs exist but they are expensive compared to their analogue equivalents which reflects the need for a high grade A/D convertor.

[Hugh Robjohns]

..., one has to come to the conclusion that one of my two Focusrite interfaces is not functioning correctly (since my measurements are off by ~ 6 dB). What I am trying to achieve is to figure out which one.

Okay. I'm clearer now about your concerns. However, if one preamp was not functioning correctly I'd expect the fault to be obvious and significant, not just an apparent small difference in overall gain.

From what you've said, Focusrite acknowledges that the two interfaces are designed to have different overall max gains, but there seems to be some confusion over which is which.

I would not understand what the point of analogue gain is in the first place (or even more so, why most people want a lot of it), because you can always increase the signal in the digital domain after recording it. Why would one need a preamp at all then?

The analogue preamp raises the very small microphone signal up to a line level suitable for the A-D converter to handle. Without the analogue gain stage, the mic signal would lie too close to the nose floor of the converter, and would have a very restricted dynamic range. Boosting the level in the digital domain would also boost the converter noise floor, and you'd still have a poor dynamic range.

In other words, it's about gain staging to maximise the dynamic range capability and minimise unwanted noise.... And James has said it in fewer words while I was typing!

H

[fladd]

..., one has to come to the conclusion that one of my two Focusrite interfaces is not functioning correctly (since my measurements are off by ~ 6 dB). What I am trying to achieve is to figure out which one.

Okay. I'm clearer now about your concerns. However, if one preamp was not functioning correctly I'd expect the fault to be obvious and significant, not just an apparent small difference in overall gain.

From what you've said, Focusrite acknowledges that the two interfaces are designed to have different overall max gains, but there seems to be some confusion over which is which.

I would not understand what the point of analogue gain is in the first place (or even more so, why most people want a lot of it), because you can always increase the signal in the digital domain after recording it. Why would one need a preamp at all then?

The analogue preamp raises the very small microphone signal up to a line level suitable for the A-D converter to handle. Without the analogue gain stage, the mic signal would lie too close to the nose floor of the converter, and would have a very restricted dynamic range. Boosting the level in the digital domain would also boost the converter noise floor, and you'd still have a poor dynamic range.

In other words, it's about gain staging to maximise the dynamic range capability and minimise unwanted noise.... And James has said it in fewer words while I was typing!

H

Mmh, but doesn't this imply that high analogue gain and a high conversion point is always better as it brings you further away from the converter noise floor and increases dynamic range? I think I'm still struggling with this concept somehow. While I can somehow imagine that one might decide to decrease the conversion point to prevent possibly hitting the ceiling in the digital domain even with the gain knob all down (with very loud signals maybe...not sure how reslistic that is), I still don't seem to understand why one would ever want to do that without adding more analogue gain to compensate (as Focusrite apparently did on the 18i8). It seems that this just shifts the problem from possibly hitting ceiling to possibly hitting noise floor.

Thanks for being so patient with me btw. This forum is one of the most helpful ones I've seen!

[James Perrett]

Mmh, but doesn't this imply that high analogue gain and a high conversion point is always better as it brings you further away from the converter noise floor and increases dynamic range?

You are making the assumption that the convertor is noisier than the preamp - which isn't necessarily true. In budget interfaces like this it would be better to think of the whole system as a black box rather than a separate preamp and convertor. There could be all kinds of interactions between the convertor circuit and the preamp circuit that wouldn't happen with separate units.

[Hugh Robjohns]

... but doesn't this imply that high analogue gain and a high conversion point is always better as it brings you further away from the converter noise floor and increases dynamic range?

Nope. Different sources generate different sound pressure levels. Different microphones have different sensitivities. Different applications require different headroom margins. There is no one perfect set of parameters to suit all situations, so the gain needs to be user-variable and the digital alignment either adjustable or chosen carefully to suit the typical application.

It seems that this just shifts the problem from possibly hitting ceiling to possibly hitting noise floor.

Yep... the art and skill (of both the user and the equipment designer) is to optimise the signal level somewhere between the two without falling foul of either!

H

[ef37a]

Sorry to be so late back on parade! I no longer have the means to generate an accurate ~1mV balanced signal so had to build a box to do it by indirect means and I eff'ed up the first time and just recorded hum. Mkll is better.

I don't have 2i2 but thought a comparison between an 8i6 and a KA6 might be of interest?

This HP laptop played a 1kHz sine into 'The Box' at 51.2mV and that put, as best I can calculate, 1.1mV into the 8i6 set to max gain. That gave an indicated -14.5dBFS in Samplitude.

Same input for the KA6 gave -25.4dBFS a gain difference of almost 11dB which surprised me as 'operationally' there seems no great difference. However, running a RightMark plot for both exported.wavs showed the KA6 to be only 3dB noisier (ref the 1khZ signal, the KA having a ~9dB lower overall noise floor but lower gain). Both figures are more than good enough for most application?
Both plots show a lovely 50Hz hum spike some 60dB above the noise floor but no hum is present when my lashup is disconnected!

The 8i6 has, IMHO a bit to MUCH gain and the control is very hard to set at the last 10% of travel but I dare say the new ones are better? The 8i6 is after all no longer made.

If anyone wants details of The Box PM me. I can also donate a ( cheap and fairly nasty!) 1:1 traff to the cause.

Dave.

[Drew Stephenson]

Sorry to be so late back on parade! I no longer have the means to generate an accurate ~1mV balanced signal...

I can't help but feel that most people would have stopped there!

[Hugh Robjohns]

Sorry to be so late back on parade! I no longer have the means to generate an accurate ~1mV balanced signal...

I can't help but feel that most people would have stopped there!

Indeed! I have the means too... but not the inclination!

...and I have to be careful about that restraining order....

[ef37a]

Right!
**************I'll get me coat.

Dave.

[Hugh Robjohns]

[fladd]

I thought I would give some feedback on how this played out.

After Focusrite Support asked me to do several tests that included loopback recordings as well as an external device (Novation Circuit) and send them the results, they concluded that my new 18i8 is performing according to the specs, but that my older 2i2 is not. To be more precise, it appears that my 2i2 has much more gain than it should have, and this explains the large difference to the 18i8 that I noticed.