Microphone preamp measurements: How to get the signal level right

[Pluto]

When I used to **** around with mic amp builds, the use of those awful carbon jobbies as an input load was 2 or 3dB noisier than decent MO. Perhaps there was a tiny current flowing. Never checked for that. That's probably why my mic amp designs were never any good.

LATER -- thinking about it, there probably was a small current flow because I usually tested the electronics without the front end transformer (which was too expensive for students).

 

[Rja4000]

Anyway, I'm progressing.
I have now a clear view of "gain" for each of them, and I'm done with my measurement template.
Still need to finish the automation part, and I'll do the measurements...

 

[dfuller]

Old man Rupert claims that transformers are the key to the sound of Neve consoles of that era, and that he would not have used a solid-state approach to a mic. input stage even had a suitably low noise one been available at that time.

Yes, Rupert does love his transformers. He thinks they have certain sonic advantages and who am I to argue with him on that one? Part of his continued use is the other inherent benefits of coupling transformers - perfect (or near enough to not matter) DC isolation and passive (un)balancing of a signal - part of that is the particular sound signature they introduce.

 

[Rja4000]

particular sound signature they introduce.

Which is what Rupert assumes.
Not our current definition of "clean".

 

[Rja4000]

Rupert does love his transformers

The same transformers I introduced in my test, with the JDI

 

[Rja4000]

First measurements (just to give a taste of it)

A set of Mic preamps tested for SINAD @various gains (20mV 1kHz input)

(OK, that's a little crowdy.
Consider it as just a preview at that stage.)

For preamps with ADC, we align them with 0dBFS
The Motu actually saturates before 0dBFS

For the 2 full analog preamps (Mackie and Millennia), I've consider 0 dBFS to be +24dBu (max range on the RME ADI-2 Pro fs R)
Note that the Millennia goes beyond that (the Autoranger II is used to level it down for recording), while the Mackie can't reach it (on Insert out, which is the shortest and cleanest path).


Now we start to understand what makes the cost difference.

The Millennia is, by far, the preamp that gives you the widest Gain range with >90dB SINAD.
The cheaper designs are much narrower: Motu, Mackie, DM1000.

Don't worry too much for a 1dB difference in SINAD: The limit at that level is most likely set by the source signal.
I guess a change in input impedance may explain 1-2dB difference.
What I find more interesting is the trend profile.

The Yamaha AD-8HR is meant to be close to stage and remote controlled, so it presents a very wide range. Obviously, they use some range change to align with ADC, which gives this strange look.
The Focusrite also proposes a wide gain range, but you're far from max gain, at 20mV level (54 gain mark, while it goes up to 80 on 4Pre - the AD8HR is 51 out of 62)


A zoom on the usable section :

Comments welcome.
And more to come...

 

[Rja4000]

I let you imagine the other 33 gain settings...

 

[Blumlein 88]

First measurements (just to give a taste of it)

A set of Mic preamps tested for SINAD @various gains (20mV 1kHz input)

View attachment 77406

(OK, that's a little crowdy.
Consider it as just a preview at that stage.)

For preamps with ADC, we align them with 0dBFS
The Motu actually saturates before 0dBFS

For the 2 full analog preamps (Mackie and Millennia), I've consider 0 dBFS to be +24dBu (max range on the RME ADI-2 Pro fs R)
Note that the Millennia goes beyond that (the Autoranger II is used to level it down for recording), while the Mackie can't reach it (on Insert out, which is the shortest and cleanest path).


Now we start to understand what makes the cost difference.

The Millennia is, by far, the preamp that gives you the widest Gain range with >90dB SINAD.
The cheaper designs are much narrower: Motu, Mackie, DM1000.

Don't worry too much for a 1dB difference in SINAD: The limit at that level is most likely set by the source signal.
I guess a change in input impedance may explain 1-2dB difference.
What I find more interesting is the trend profile.

The Yamaha AD-8HR is meant to be close to stage and remote controlled, so it presents a very wide range. Obviously, they use some range change to align with ADC, which gives this strange look.
The Focusrite also proposes a wide gain range, but you're far from max gain, at 20mV level (54 gain mark, while it goes up to 80 on 4Pre - the AD8HR is 51 out of 62)


A zoom on the usable section :

View attachment 77408

Comments welcome.
And more to come...

Let me see if I understand correctly. You are feeding an input of 20 mv to the preamp, and if an ADC, arbitrarily raise gain until you get 0 dbFS with that signal? So your graph is that gain and lower. I wish it showed what the gain setting is to obtain 0 dbFS. Maybe in a separate chart.

What you are seeing looks similar to what I've seen testing a handful of interface mic pres. You have a 90 to 100 db wide range that follows the gain setting. Some manage that over a wider range of gain settings than others. Eventually you run into the upper max input limitation with enough gain, and as you lower the gain noise levels intrude at some point on the low end of the signal level.

For instance if you were to look just at noise levels, a pre might have 130 dbv EIN at 60 db gain, but often that will begin to deteriorate at and below 40 db gain levels. A preamp that maintains EIN over the wider range will give a little better result in practice.

Also I'd note it isn't uncommon for condensor and electret microphones to have outputs of 35 mv or more at 94 db SPL. So you might want to run the same test at 40 mv as well as 20 mv.

Also what output impedance was feeding the 20 mv signal? 50 ohms or 150 ohms or what?

 

[Blumlein 88]

View attachment 77466

I let you imagine the other 33 gain settings...

Is this result with the same fixed 20 mv input signal? You really wouldn't use a preamp that way normally.

 

[Pluto]

A preamp that maintains EIN over the wider range will give a little better result in practice

This is very true. In the real world, I am only interested in the front end noise performance when the gain knob is set at 50dB or more. Below that, the noise performance would have to be heroically poor to be of consequence. That said, some mic. amps. do themselves no favours by “enticing” the user to press a 20dB pad button rather than genuinely reducing the front end gain by 20dB!

 

[Blumlein 88]

This is very true. In the real world, I am only interested in the front end noise performance when the gain knob is set at 50dB or more. Below that, the noise performance would have to be heroically poor to be of consequence. That said, some mic. amps. do themselves no favours by “enticing” the user to press a 20dB pad button rather than genuinely reducing the front end gain by 20dB!

You are so right. I've assumed the proper use of a pad is when your mike can exceed the input max of your preamp. Like if you were miking drums with a mike that can take loud inputs and output a correspondingly high output signal. You pretty much have to pad it, but it keeps the mic pre from distorting as much.

 

[Rja4000]

Thanks for your feedback.
As I said, I'm exploring.

The 2 graphs are 2 different things: first is with a fixed input level of 20mV with x varying with the gain.

As comparing actual gains for preamps with no analog output is impossible, I thought I'd scale them with the digital level after ADC.
Of course, it raises with the gain setting, until overload (which happens before 0dBfs for the Motu).

To have the analog preamps on the same graph, I arbitrarily decided that 0dBFS for them would be +24dBu, the max range on the RME, which is also very common.

So, in my opinion, that tells me how flexible is my choice for a given input level.
In real life, we'll set the gain for highest level, to have enough margin below 0dBFS to avoid accidental overloads.
The margin will be 20 to 10 dB typically, for low to high confidence on high level accuracy.
With the Mackie or the Motu, I may want to increase that margin though or select tge +13dBu range for the ADC with the Mackie).

I planned to do the same at different levels (2mV, 20mV, 200mV, 2V)

As said, in real life, we set the gain and leave it. The input level varies with the music.
The second graph corresponds to that, and is very similar to any SINAD graph for any amp.

 

[Rja4000]

I understand EIN at 150 ohms and in short vs gain are most interesting.
I'm working on that.

The difficult part was first to identify the gains.

The Yamaha AD8HR, The Focusrite, the Motu and the Millennia all have fixed steps with marking, so it should be possible to reproduce a previously measured "gain".
The DM1000 has detends, but gain varies a bit every time. The Mackie gain pot is tiny and impredictable. So I'll have to remeasure it for each setting for noise measurement.


Finally, a full set of measurements at levels around -10dBFS (or lower for the Motu) may still be worth it.
SINAD, THD, IMD SMPTE and CCIF, 31 tones, Frequency response, crosstalk, gain matching (for the stereo HV-3C).

At minimum gain, we may try to vary level down to -50dBFS too. At higher gain, I suppose we would mainly see the source signal noise.

 

[Rja4000]

what output impedance was feeding the 20 mv signal? 50 ohms or 150 ohms or what?

The Radial JDI outout feeds the preamp for 20mV.
That's basically a transformer with a resistive Pad.
It's given for a realistic "typical 150 ohm", which is similar to a microphone.

I guess the only way to make sure is to create your
own pads.

I've build one H pad with some (cheap) 1/4W 1% metal film resistors in an XLR connector.
I've build it for 6k input Side and 150 ohm output side resistance and 66dB attenuation
I have results identical to the measures with the Radial SAT-2 attenuator for the same attenuation., except that the distortion raises suddenly at high level (not the case with the Radial).

I guess I should use higher power resistors and choose values for lowest power dissipation.

I plan to build a box with selector for some values (and optionnal ground lift).
Any recommandation is welcome.

 

[Rja4000]

some mic. amps. do themselves no favours by “enticing” the user to press a 20dB pad button rather than genuinely reducing the front end gain by 20dB!

And, indeed, that gives poorer results than real gain change.

But, as we can see on my first graphs, the Yamaga AD8HR and the Focusrite also have some major steps at some gains. Do you think that's because they insert a pad ? Given their gain ranges (14 to 80 for Focusrite and -10 to 62 for the Yamaha), that's probably the case.

 

[MC_RME]

If you are interested in measuring noise levels, why worry about a signal source. Simply place a low noise* 200Ω resistor from each of pins 2 & 3 of the XLR input down to earth.

That would mean 400 Ohm input impedance or termination. These are all balanced inputs, so it would be correct to use one 200 Ohm resistor between pins 2 and 3.

 

[Pluto]

What microphone source impedance do you optimize for? A powered mic. that will tend to have a higher output but lower output impedance or a dynamic mic. with an inherently higher output impedance but somewhat lower output and, therefore, desperately in need of the quietest possible gain?

 

[Blumlein 88]

That would mean 400 Ohm input impedance or termination. These are all balanced inputs, so it would be correct to use one 200 Ohm resistor between pins 2 and 3.

That's what I've done in the past. Except I use 150 ohm metal film. Seems to me 150 ohms is the most common condenser impedance.

 

[Pluto]

the Yamaha AD8HR and the Focusrite also have some major steps at some gains. Do you think that's because they insert a pad ?

It's obviously possible but without access to circuit diagrams…

Don't get too paranoid about the gain steps however as long as the results appear sensible. Even dear Uncle Rupert with his wonderful switched 5dB gain steps all the way from -20dB to +80dB used a combination of gain adjustment and source attenuation to keep things optimized; the trick lies in getting that optimization right

 

[Pluto]

Seems to me 150 ohms is the most common condenser impedance

Agreed, but should we not be optimizing our mic. amps. for the lower output (and higher impedance) of dynamic mics. that need all the ‘help’ they can get.

The dominant noise source in a condenser mic. chain is probably the mic. itself (i.e. its internal amplifier); for dynamic mics., the dominant noise is far more likely to be the mic. amp.