-
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
- Thread starter amirm
- Start date
somebodyelse
Major Contributor
- Joined
- Dec 5, 2018
- Messages
- 4,495
- Likes
- 3,886
https://linearaudio.net/record-replay-riaa-correction-digital-domain covers the DSP in some detail, plus a section on coupling cartridge and interface.
As rja4000 suggested different gain range can lead to different real world results.
My suggestion:
1 First calibrate the input to relate V/dBu to the dBFS.
2 Set gain to 50dB (or pick one magic number from 45-55.
3 Measure the noise with resistor termination (150ohm is good) and refer back to the EIN.
4 Measure again at max gain to verify against the claim from manufacturers.
The EIN should still be A weighted because that's what matters. Larger low frequency noise or noise over 15khz are not going affect sound as much as noise around 3khz.
Preamp noise only matters at higher gain levels. For condenser mics, the sensitivity is high and noise is usually much higher than the preamp itself. The noise is only a big problem when SM7B is so insensitive and so popular. SM58 is also slightly better but also quite insensitive. Low sensitivity and high noise goes hand in hand. So measuring a curve is unnecessary.
Using a signal source and cable in between will inevitably introduce noise and interference or grounding issues into the mix which is not really dependant to the preamp but the setup and the environment.
My suggestion:
1 First calibrate the input to relate V/dBu to the dBFS.
2 Set gain to 50dB (or pick one magic number from 45-55.
3 Measure the noise with resistor termination (150ohm is good) and refer back to the EIN.
4 Measure again at max gain to verify against the claim from manufacturers.
The EIN should still be A weighted because that's what matters. Larger low frequency noise or noise over 15khz are not going affect sound as much as noise around 3khz.
Preamp noise only matters at higher gain levels. For condenser mics, the sensitivity is high and noise is usually much higher than the preamp itself. The noise is only a big problem when SM7B is so insensitive and so popular. SM58 is also slightly better but also quite insensitive. Low sensitivity and high noise goes hand in hand. So measuring a curve is unnecessary.
Using a signal source and cable in between will inevitably introduce noise and interference or grounding issues into the mix which is not really dependant to the preamp but the setup and the environment.
Audio Precision has too high of noise to test EIN for mic preamps. If it's for testing distortion and such, then it's fine. IMHO people I know don't cares about distortion with mic input even though E2x2 excels at this. Maybe just knowing when the input is clipping is good enough. But I'm ok with testing distortion. To test EIN, the kind of resistor termination I made is kind of a must.
I'm not sure if max gain for Hi-z input is really meaningful. Max gain + Hi-z = Crazy noise.
Input impedance for High-z is crucial.
And I think the max voltage input before clipping is needed because in the earlier days some pickups will drive the hi-z input to red and causing audible unwanted distortion.
Sorry guys Couldn't contribute this before
Here is what I think (first thoughts)
1. Focus on inputs
If you purchase an interface (or a mic preamp), your main interest is the Preamp/ADC part
DAC is also important for sure, but your main focus the input signals
2. Focus on Mic inputs
Most interfaces have different inputs: Mic input, line input and, for some, instrument input
My experience is that Mic input is usually performing better (line input is often using the same circuit with a pad attenuator)
So I suggest we focus on the Mic input, without any pad
3. "Gain" range and max input level for various gains
Getting the actual gain range is a key information, and it's usually not clear -or not always accurate- in the specs.
For an analog mic preamp (or any analog amp), gain is straightforward and easy to measure.
For an interface with a preamp followed by an ADC, there is no way for us to separate the Preamp gain from the ADC sensitivity.
So I suggest we standardize on an "Input sensitivity" measurement
For min and max gain settings,
measure the maximum input level (Vrms or dBV or dBu) that gives 0dBFS (when there is an ADC) or 0.1%THD (for Analog preamps, but also if the ADC is saturating before 0dBFS)
List 3 values: the Gain markings for min and max, and for which the maximum input level is 100mV rms
Notes: I propose 100mV since it's somewhere in the middle of typical gain, so it's likely all interface will allow it.
That could be any value, as long as it's standardized.
It could probably be deducted from the Min and Max gain sensitivity.... IF the actual gain is actually following linearly the gain markings. Which we don't know.
From the above, we may deduct a standardized "Input sensitivity range" for the interface.
A very important spec, since a user will want to compare it to his use case and the voltage he's expecting to get from his microphones.
4. Low level input signal
For low level input, what matters is the noise.
The standard measurement for noise is EIN
So let's measure EIN noise at high gain (for low level signal we need high gain).
Questions we'll have to answer:
- 150 Ohm load or Short circuit input ? The former is more realistic but may be less discriminant (and you may have to compensate for the temperature to take into account your load's own thermal noise), while the latter will show only the noise.
- What bandwidth ? I suggest 20Hz-20kHz, but that's arbitrary
- Weighted or not ? If the noise is perfectly pink, the relation between both is around 2dB. Weighted will give less importance to mains noise typically.
- For what gain ? That's the tricky part. Could be max gain (and then different for each interface) or the gain for X mV...
5. High level signal
Here, distortion matters.
So the typical ADC dashboard you're using makes full sense.
In my opinion, dynamic range is a very important measurement at low gain:
If one want to record classical music in a room with highly sensitive microphones, the dynamic range may be a key success criteria.
This is where it's important to align the levels between interfaces, IMO.
Again: for what level do we want to measure that ?
Note that the high sensitivity microphones are condenser microphones, which require phantom supply.
And I think input impedance may be very different when phantom power is switched on, so behavior may be affected.
But it's hard to measure with the Phantom power on.
6. Frequency response and THD vs frequency
Frequency response is, for sure, an important topic.
As quite a few interfaces or mic preamps are using input (and/or output, for the analog ones) transformers, THD vs frequency is important.
A faster way to measure both is to use Multitone at max level.
I'll come back later, but now I need to go back to work...
Here is what I think (first thoughts)
1. Focus on inputs
If you purchase an interface (or a mic preamp), your main interest is the Preamp/ADC part
DAC is also important for sure, but your main focus the input signals
2. Focus on Mic inputs
Most interfaces have different inputs: Mic input, line input and, for some, instrument input
My experience is that Mic input is usually performing better (line input is often using the same circuit with a pad attenuator)
So I suggest we focus on the Mic input, without any pad
3. "Gain" range and max input level for various gains
Getting the actual gain range is a key information, and it's usually not clear -or not always accurate- in the specs.
For an analog mic preamp (or any analog amp), gain is straightforward and easy to measure.
For an interface with a preamp followed by an ADC, there is no way for us to separate the Preamp gain from the ADC sensitivity.
So I suggest we standardize on an "Input sensitivity" measurement
For min and max gain settings,
measure the maximum input level (Vrms or dBV or dBu) that gives 0dBFS (when there is an ADC) or 0.1%THD (for Analog preamps, but also if the ADC is saturating before 0dBFS)
List 3 values: the Gain markings for min and max, and for which the maximum input level is 100mV rms
Notes: I propose 100mV since it's somewhere in the middle of typical gain, so it's likely all interface will allow it.
That could be any value, as long as it's standardized.
It could probably be deducted from the Min and Max gain sensitivity.... IF the actual gain is actually following linearly the gain markings. Which we don't know.
From the above, we may deduct a standardized "Input sensitivity range" for the interface.
A very important spec, since a user will want to compare it to his use case and the voltage he's expecting to get from his microphones.
4. Low level input signal
For low level input, what matters is the noise.
The standard measurement for noise is EIN
So let's measure EIN noise at high gain (for low level signal we need high gain).
Questions we'll have to answer:
- 150 Ohm load or Short circuit input ? The former is more realistic but may be less discriminant (and you may have to compensate for the temperature to take into account your load's own thermal noise), while the latter will show only the noise.
- What bandwidth ? I suggest 20Hz-20kHz, but that's arbitrary
- Weighted or not ? If the noise is perfectly pink, the relation between both is around 2dB. Weighted will give less importance to mains noise typically.
- For what gain ? That's the tricky part. Could be max gain (and then different for each interface) or the gain for X mV...
5. High level signal
Here, distortion matters.
So the typical ADC dashboard you're using makes full sense.
In my opinion, dynamic range is a very important measurement at low gain:
If one want to record classical music in a room with highly sensitive microphones, the dynamic range may be a key success criteria.
This is where it's important to align the levels between interfaces, IMO.
Again: for what level do we want to measure that ?
Note that the high sensitivity microphones are condenser microphones, which require phantom supply.
And I think input impedance may be very different when phantom power is switched on, so behavior may be affected.
But it's hard to measure with the Phantom power on.
6. Frequency response and THD vs frequency
Frequency response is, for sure, an important topic.
As quite a few interfaces or mic preamps are using input (and/or output, for the analog ones) transformers, THD vs frequency is important.
A faster way to measure both is to use Multitone at max level.
I'll come back later, but now I need to go back to work...
- Joined
- Feb 23, 2016
- Messages
- 22,492
- Likes
- 41,304
It certainly is the case that the great majority of interfaces use the same circuit for line and mic inputs with a pad being the only difference in line and mic.
OP
- Thread Starter
- #28
Of course we are not going to use the AP for noise measurement if the standard is 150 ohm. I thought he was referring to something else such as miscalculating the gain with high impedance (AP) source.
Can't you just use an appropriate voltage divider between the AP output and the MIC input, or would Johnson noise of those resistors be too high?
thecheapseats
Addicted to Fun and Learning
really? - music professionals - imagine that!... : )
I came here to shop for home audio stuff for my son and brother - and stuck around for the free food...
OP
- Thread Starter
- #32
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?
thecheapseats
Addicted to Fun and Learning
and aes wasn't supposed to cut you short and 'own' pub-rights on your paper... life's tough... : )
now I have a good excuse as to why I never joined aes...
- Joined
- Feb 23, 2016
- Messages
- 22,492
- Likes
- 41,304
60 db is pretty common, but if you want all of them, less than 50 db is very uncommon. Most are between those 50 and 60 db values.
The preamp gain doesn't mean much.
Let's say you should be able to saturate the input of most interfaces with 10mV at high gain.
That's 20dB below 100mV.
If your interface's 0dBFS for 0dB gain marking is 18dBu (EBU std), you'll reach 0dBFS with 10mV with a gain marking of approx 57dB.
If the 0dBFS for 0dB gain marking is 24dBu (AES std), you'll need to push it to 63dB gain to get 0dBFS from the same 10mV.
For some analog preamps, if you want to push them to the maximum output before clipping, this level may be too low: the Millennia in my plot saturates at 28dBu, in my measurements, and the maximum (measured) gain is around 60dB (61.5dB marking). So you'd need a -32dBu signal to use its full potential.
That's 20mV.
So if we want a general example, 20mV is probably a good target level.
It makes sense to standardize the test signal level, since it's similar to standardizing a given SPL recorded by a given microphone @ a given distance.
Last edited:
OP
- Thread Starter
- #36
Well it does if you want to measure EIN equally across all interfaces. I tested the Topping E2x2 and it gets its best number at max gain but then drifts lower as you reduce gain. I thought it was you that brought up this point.
Maybe we should try to pull the manufacturers in this discussion ?
@MC_RME
@signalpath
What do you think about this ?
@MC_RME
@signalpath
What do you think about this ?
It does, for sure, vary with gain.Well it does if you want to measure EIN equally across all interfaces. I tested the Topping E2x2 and it gets its best number at max gain but then drifts lower as you reduce gain. I thought it was you that brought up this point.
But "gains" between interfaces don't have the same meaning.
As @JohnYang1997 wrote, you don't know how much of this "gain" is actually analog and how much is compensated for in the digital domain.
Even for Pro interfaces, 0dB gain don't mean the same between brands.
RME uses EBU standard, I think (0dBFS at 0dB gain = 18dBu), Yamaha uses 20dBu, Focusrite uses 22dBu, most american builders use 24dBu (AES) for the same...
If I take the analog mic preamp Millennia HV-3C in my plot, you may vary the level by varying the sensitivity of the ADC. If you connect it to an ADC where 0dBFS is achieved for 4dBu, you'll get artificially 20dB more "gain" than if you connect it to an ADC accepting 24dBu.
But you'll also loose 20dB of dynamic range at low gain/high level doing so.
In the plot above, the ADC senditivity is changed. (NB: This is simulated, not measured)
The same gain marking of 61.5dB (max gain) on the HV-3C will give you a very different dynamic range for the different settings.
If you want to compare things, you'll need to set a reference which is the same for all, something meaningful in real life.
Like the input level.
And use the gain that gives you 0dBFS for that level, whatever it is.
Well, that's what I think.
That's how I aligned the plots for the different interfaces.
You see that at low level/high gain, on the right of the plot, all interfaces have more or less similar dynamic range. That's limited by the thermal noise, and that's normal.
If I didn't align the gains for the same sensitivity, that would have been all over the place.
Last edited:
In my plot above, 20mV 0dBFS corresponds to the gain at +14dB above referrence and your 4V corresponds to -32dB.
It seems to me those values are meaningfully.
At 20mV, you're well into the thermal noise limited range, and at 4V, the dynamic range differences show seriously.
+4dBu (AES nominal level) or 0dBu (EBU nominal level) are a bit low, IMO.
At low level, though, I don't think it's worth measuring anything else than the noise, so you probably don't even need a signal.
But aligning gains for 20mV=0dBFS makes sense.
It seems to me those values are meaningfully.
At 20mV, you're well into the thermal noise limited range, and at 4V, the dynamic range differences show seriously.
+4dBu (AES nominal level) or 0dBu (EBU nominal level) are a bit low, IMO.
At low level, though, I don't think it's worth measuring anything else than the noise, so you probably don't even need a signal.
But aligning gains for 20mV=0dBFS makes sense.
Last edited:
OP
- Thread Starter
- #40
Well, this is why I was trying to pick a lower value "max" gain as to avoid that.
Similar threads
