• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Measuring Microphone Preamp Perf in Audio Interfaces

I'd also like to see the metering checked for accuracy across the frequency range. Is the clip light pessimistic like most clip indicators? Are the input/output indicators on the money?

The Topping interface has such coarse indicators (6dB) as to be pretty much useless for precision setting especially around 0dB.
 
I'd also like to see the metering checked for accuracy across the frequency range. Is the clip light pessimistic like most clip indicators? Are the input/output indicators on the money?

The Topping interface has such coarse indicators (6dB) as to be pretty much useless for precision setting especially around 0dB.
I agree, but you usually end up looking at the software meters on a laptop. I forget who, but they made some interfaces with no meters at all for a time. All in software. It ends up being a better way.
 
I agree, but you usually end up looking at the software meters on a laptop. I forget who, but they made some interfaces with no meters at all for a time. All in software. It ends up being a better way.

Good point.
 
So what is a pre-amp gain that is on the high side that is supported by all products? Is it 50 dB? Or lower?
@Julian Krause to the rescue once again:
index.php

(Strictly speaking, this graph should be labeled input sensitivity, but what's a factor of -1 among friends?)

Looks like -40 dBV @ 0 dBFS ought to be fine. It would, however, require that ADC dynamic range be safely above 90 dB, so let's say at least 100 dB. There's a handful of interfaces that would not make the cut:
jk-micin-dr-0.jpg


I don't think there is a single interface that couldn't get its preamp noise floor at least 10 dB above ADC noise one way or another when juggling both of the above, but there will be some variation in how it is achieved. Hence why I suggested <max gain>, -40 dBV, -20 dBV.
 
Looks like -40 dBV @ 0 dBFS ought to be fine.
That's 10mV for 0dBFS, which corresponds to +20dB above the reference gain in my plot above.

As I said, that may be a bit short for some ambitious pro preamps.
I'd use a slightly higher figure, like 20mV for 0dBFS (-34dBV)
 
Last edited:
@Julian Krause to the rescue once again:
index.php

(Strictly speaking, this graph should be labeled input sensitivity, but what's a factor of -1 among friends?)

Looks like -40 dBV @ 0 dBFS ought to be fine. It would, however, require that ADC dynamic range be safely above 90 dB, so let's say at least 100 dB. There's a handful of interfaces that would not make the cut:
View attachment 323403

I don't think there is a single interface that couldn't get its preamp noise floor at least 10 dB above ADC noise one way or another when juggling both of the above, but there will be some variation in how it is achieved. Hence why I suggested <max gain>, -40 dBV, -20 dBV.


Hmm, the Zoom dynamic range number seems high. Is this just the micamp or the micamp + ADC? If ADC is included, it must be a multi-path architecture. Zoom, Rode, Sound Devices, Salzbrenner (True Path) and others are shipping multi-path ADCs now, with a significant dynamic range improvement. The first multi-path DAC will ship next year, with around 160dB of dynamic range (think 50nV of quiescent noise). Multi-path microphones and power amplifiers to follow. At some point, we will realize a 160dB signal path, from mic to power amp.
 
(Strictly speaking, this graph should be labeled input sensitivity, but what's a factor of -1 among friends?)
Indeed. Strange that he states gain as a negative number.

Anyway, is there really goodness to having highest level of gain??? I am talking about ranking them. Wouldn't the goodness come from some kind of gain*(inverse of)noise level?
 
-40dBV could work. But I prefer keeping the convention of using dBu. So -40dBu. But not too big deal. That's directly convertible by adding 2.2dB.
-34dBV maybe slightly too high in level. At high gain is where the difference shows and when the noise becomes audible.

MAX, -40dBV, -20dBV works for me.
 
Last edited:
-40dBV could work. But I prefer keeping the convention of using dBu.
Users will want to compare this figures to microphone sensitivity, which is usually given in mV/Pa (94 dB SPL).
Or to dBu indeed, when using with a line level signal.
So dBu and mV both make sense.
-34dBV maybe slightly too high in level. At high gain is where the difference shows and when the noise becomes audible.
-35dBu then ?
That's -37dBV, as you said.
Or max gain if the interface/preamp can't reach that level.

Remember this is the max level. Average recording level should be -18 to -20 dBFS, with peak signal around -6 dBFS, if you follow the usual rules.
MAX, -40dBV, -20dBV works for me.
For EIN, representing low level performance, one figure should be enough.
Of course, more values are welcome -and a full plot vs all gain settings is very interesting, especially since sudden jumps are quite common- but that's time consuming.

On top of that, we need the full analysis at low gain, high input level.
For which gain ?
That's a good question.

If you look at microphones, their sensitivity range between around 0.8mV/Pa (like the very good Audix OM7) and 40mV/Pa (for the legendary DPA 4006). That's a 34dB range.
And, of course, source's SPL and distance to source can vary hugely.

So, at least 40dB less gain than low noise.
If we do it for 0dBFS = 4V (14dBu), that may be a bit high for some lower cost, 5V powered interfaces.
And that's low for a pro interface.

So, maybe, the best would be to measure at min gain (without PAD), with a signal at -6dBFS (which is supposed to be the peak value you'll using).
This will favor the high end gear, but they'll actually perform better in reality, given their higher input level tolerance.
This is beneficial for the user, so it makes sense to show it, IMO.

A similar approach could be used with ADCs:
Measure noise for a standardized input level setting, if you have a choice, but do the rest of the measurements near max level.
 
Indeed. Strange that he states gain as a negative number.
His X axis is the level to reach 0dBFS.
Yamaha, on the AD8HR, also uses negative numbers for "gain".
Anyway, is there really goodness to having highest level of gain??? I am talking about ranking them. Wouldn't the goodness come from some kind of gain*(inverse of)noise level?
1698859222547.png

EIN figures will flat out below a certain sensitivity:
Below around -35dBV for 0dBFS, they don't vary much.
(15dB more gain than my "reference gain", which is -20dBV for 0dBFS)

That's why I proposed the level above.
If you go below that (higher gain), some interfaces may not have enough "gain".
And if you keep it higher, you'll have other factors than noise at play.
(Look at the steep changes for the Yamaha)
 
Last edited:
My comment above was NOT in the context of EIN. He seems to be saying that the higher the gain of a pre-amp the better it is, regardless of its performance. That makes no sense to me:

index.php
 
My comment above was NOT in the context of EIN. He seems to be saying that the higher the gain of a pre-amp the better it is, regardless of its performance. That makes no sense to me:

index.php
What matters for the user is the range of input sensitivities.

You're right on the fact that, past a given gain, the noise will just be amplified with the signal, as my EIN plot shows, and you'll bring no benefit.
SNR ratio will remain the same.

Lesser preamps or interfaces may just reach this platter sooner, given they add more noise by themselves.

Getting more gain was important in analog, since if the preamplified signal remains too low, you'd have to amplify it again, and add another source of noise.
But not in digital.
 
Last edited:
My comment above was NOT in the context of EIN. He seems to be saying that the higher the gain of a pre-amp the better it is, regardless of its performance. That makes no sense to me:

index.php
At least in terms of gain, if you use ribbons or the infamous SM7b it does help. The sensitivity of the Sm7b is 1.12 mv/Pa. Some old ribbons can be half that though most are between 1 and 2 mv/Pa. Also with ribbons you don't want those as close to what you are recording because they could be damaged.
 
7
At least in terms of gain, if you use ribbons or the infamous SM7b it does help. The sensitivity of the Sm7b is 1.12 mv/Pa. Some old ribbons can be half that though most are between 1 and 2 mv/Pa. Also with ribbons you don't want those as close to what you are recording because they could be damaged.
But, as I said, and as experience teaches, for each preamp, there is a gain above which increasing it brings no benefit.
For good preamps, this threshold is the same for all.
 
7

But, as I said, and as experience teaches, for each preamp, there is a gain above which increasing it brings no benefit.
For good preamps, this threshold is the same for all.
Agreed, but part of the reason for testing is to ferret out units that don't work properly or well. So a preamp with not much gain, and a poor EIN at lower gain, that is what you want to find in testing. Some designs may not have enough gain to get you to the point of no benefit for something higher.

I had an interface with 75 db, and I cannot say I ran across any situation where 65 or 60 db is not enough. I know some number of stand alone pro units have 80 or in some cases 90 db. I don't know when those would be useful. Or if they are a side effect of optimizing some other aspect of the microphone preamp. Metric Halo preamps have a claimed 91.5 db gain.
 
Indeed. Strange that he states gain as a negative number.
It probably should be named "input sensitivity" or "line-up level" or something like that. But non-technical people might not understand this term, that's why I have settled for "maximum gain of mic input". The whole graph is a tradeoff between understandability for casual consumers and technical people. The negative number gives you the input level in dBV that is needed to reach 0 dBFS with the gain set to the maximum. The more negative (smaller the signal) the more "gain" the device has. Technical people can directly see the line-up level of the maximum gain setting in the graph and consumers tend to completely ignore the minus sign and just see something like "65 dB gain". Not ideal but the best I could come up with for now and I'm happy to hear suggestions for improvement. :)
 
Last edited:
My comment above was NOT in the context of EIN. He seems to be saying that the higher the gain of a pre-amp the better it is, regardless of its performance. That makes no sense to me:

index.php
Correct, more "gain" does not necessarily mean better. In most cases gain isn't even an issue as you can simply amplify the recording in post. But there are scenarios where more gain is benefitial. E.g. if you use a low output dynamic mic like the Shure SM7B and you are live streaming you want to be able to get the mic to a decent level otherwise your stream would be too quiet. Yes, you could also boost the signal in programs like OBS but in reality, not that many people know that this is possible and they want to be able to get their signal as close to their desired output with the gain on their mic input. Also, if the gain is too low, you might have troubles monitoring the audio in because you would need to crank up your headphone volume but might still not be able to hear yourself properly. That's why a certain amount of gain is needed but I also agree, that anything above -50 dBV line-up at maximum gain setting, is rarely beneficial. That's why most interfaces on the list already have a "good" or "excellent" rating as there are only a few where you might have issues due to their low amount of gain.
 
So most of the discussion has been about gain and EIN. Seems pretty clear that is what is worth adding to the ADC review along with what is already done.

Are we any closer to a consensus on that?

Sensitivity for 0 dbFS sounds worthwhile. Just need to pick a given signal level and gain. 40 db gain would probably work for this as everything has at least that much gain.

EIN at max gain(usually the only spec quoted by companies) plus one or two other levels. The other one or two levels is what needs to be decided upon. Maybe EIN at 20 db, 40 db and max gain? Most big changes in EIN are below 40 db gain levels. Since you'll set up for the sensitivity measure at 40 db gain might as well use that for one of the EIN levels. Also I find with most condensers 40 db gain is about the most I use. I suppose even 40 db gain and max would work. As John Yang mentioned, noise only becomes an issue with higher gain levels.

Or maybe more consideration about this, but it does appear input sensitivity, and EIN are what everyone finds useful.
 
Back
Top Bottom