Microphone preamp measurements: How to get the signal level right

[Rja4000]

Hello

I'm still trying to measure microphone preamplifiers.
Will the difference between an one-chip-based 100$ mixer mic input and a renowned, clean, 1000$/channel preamp be measurable ?

What makes the difference ?


I own a few microphone preamplifier channels.
Actully, counting all, I have at least 5 models.

But first difficulty is to create a signal with the right level.

Looking at microphone specs, output level range may vary from 0.8mV/Pa (50 Ohm) for a good dynamic microphone to 28mV/Pa and more (200 Ohm) for some world-famous large diaphragm condenser microphone. (1 Pa = 94dB SPL)

That's over 30dB difference, for same SPL level.
Then comes the music.

So, if I want to measure it, I need to be able to measure values from a level around 1mV rms up to 1V rms,
or over a range of 60dB.


So I decided to have a look with what I have at hand.
The only pure analog Microphone preamp I own is a very old Mackie 1402 VLZ (non Pro) from 1999.
(Picture is not from me. Mine is much dirtier.)

Measuring this with an RME ADI-Pro fs BE,

I get best measurements by doing this :
RME ADI-2 Pro fs Main out 1+2 (XLR)
=> XLR Y cable (summing both outputs)
=> Mackie Mic input
=> Mackie Insert out
=> XLR Y cable (splitting outputs)
=> RME ADI-2 Pro fs Analog In

RME Analog input is set to +24dBu range, M/S ON, to average both inputs on channel L
RME Main output is set to +4dBu range

I always measure using Virtins Multi-Instrument 3.9.
I paid the Virtins license.
(REW gives similar results, but I'm able to automate Virtins, which is what I do.)

Best results may be very correct for this venerable mixer (1999)
Here at minimum gain.
SINAD for the loop is 101.3dB

But, of course, if I lower the level to get around 1mV rms, results are much lower.

Here is the SINAD for the RME output set to around -60dBFS (-56dBu)
Mackie Gain is around 45dB (around 2 O'clock)

But what is the limiting factor ?

Obviously, noise is the limiting factor here.
Isn't it the noise of the RME DAC itself ?

And 58dB SINAD + 56dB to Full scale would give a loopback SINAD of around 114dB (DAC + ADC), which is what I measure for a loopback.
(Which is pretty impressive, by the way)

Would a (passive ?) attenuation of some sort be more efficient ?
And, then, allow me to measure deeper ?

That's what I tried.
Results in next post...

 

[Rja4000]

So I've tried almost every device I had at hand to attenuate the signal between the RME outputs and the Mackie Mic Input.

Here is what I tried:


Apex-Audio dBDI-2
That's a stereo Active DI with transformers.
I own it for several years and it always performed well.
I don't know if they still sale it.
It may be powered by Phantom 48 or by 12V DC.

Radial Engineering J48 DI
That's a mono Active DI
It has to be powered by Phantom 48
It also exists as a stereo model.

Radial Engineering JDI Duplex
That's a stereo fully passive DI, using Jensen transformers (I opened mine to make sure)
As it's fully passive, it doesn't require power
Note that they also have a mono model.

Radial SAT-2
That's a small stereo fully passive Attenuator

SPL Volume 2
That's an active volume control.
As such it is connected to mains.
(That's quite a beast, compared to the other here.)

Linear Audio Autoranger II
That's an active mono measurement kit you may purchase from Linear Audio
(@jan.didden is the guy who invented that genious piece of hardware)
The goal is to automatically switch some attenuation (or gain), to measure amplifiers with a sound card.
It's limited to around 100Vrms input level.
What's of interest here is that it may be manually forced to any attenuation up to -42dB, by 6dB steps

So, here we have the actors...

 

[Rja4000]

And here are the results

First, with a level of around -36dB at the Mackie input

The direct SINAD, just by attenuating the RME output, is around 79dB

We see that the first 2 DIs (APEX and J48) don't provide any significative attenuation (-5dB and 0dB).
So that's not of any interest.

We also see that the best result is given by the 2 fully passive devices:
Radial JDI Duplex DI and SAT-2

The Autoranger II is far from ridiculous, and bests the RME - which is impressive.

The Volume 2, while not bad, is coming last.



How practical ?
Obviously, there are practical differences between the solutions.

The SAT-2 allows a continuously variable level.
More comfortable between around -2dB (minimum) and around -30dB attenuation.

The JDI is fixed attenuation: The transformer and a serie (fixed internal) 15dB pad leads to that 38dB of attenuation.


About impedance matching
On top of it, the JDI being a DI has an output impedance more in line with what a Mic input expects
(spec says 600 ohm).


Best of both worlds
So if we want to be flexible, while combining the benefits of both, the best option is probably to combine both.



I then lowered the level to around -56dB at the Mackie input

First line is without the Mackie, looping directly the RME. Ouch.
All other lines are with the Mackie gain set around 45dB.

The best result is also the one that makes the most sense:
Combining the SAT-2 first - to allow accurate level setting- and the JDI - for proper impedance.

We can prove that by looking at the exact reverse combination: JDI then SAT-2 gives much lower results.
Autoranger set @42dB attenuation is, again, not ridiculous, in this role, so far away from its main target.


I also tried inserting the JDI twice (as it's a stereo DI). The attenuation is then above what we want to achieve, so I had to change the RME output range.
Results are very similar to the SAT-2 + JDI.
So no benefit. And that's very impractical.


Checking JDI with other frequencies
But that's all at 1kHz.
What about lower frequencies ?

Easiest way to check was to push a 100 tones through the combo.

Given the level, that's still pretty clean.

If we check the same for the double JDI loop, it's not that sexy looking

On top of the lower TDN figure (Virtins is able to compute a "Total Distortion Noise" figure on a multitone measurement, which is pretty handy)
we see that the low frequencies are significantly "dirtier".


Mic Preamp @+60dB gain

Now we have it
The SINAD figure for the old Mackie with gain around maximum. That's around 60dB gain.
We see that noise is the limit, of course.
And we also see that the results are very similar to the results above, for same input level but at 45dB gain, by the way.

Analog oscillator ?
For cross checking, I also tried with Victor's 1kHz Low distortion oscillator, at minimum output level.
As you can see in the table above, there is no benefit.

 

[SIY]

Nice to see!

FWIW, for low level signals, I always use resistor attenuators with the source impedance set to match the likely source impedance of whatever is expected to be plugged into that input. If the source impedance is too low, you miss the current noise contribution; if it's too high, you can exaggerate the current noise contribution.

 

[Instrumental]

I have no experience in microphones but in small signal testing of vibration transducers
and charge amplifiers I used T-pads with an impedance matching the systems tested,
A chain of T-pads switched in/out with DPDT switches with 10,10,10, 6, 3, 2, 1 dB attenuation
gives all combinations between 1 and 40 dB. The balanced pads require 4PDT switches.
https://www.electronics-tutorials.ws/attenuators/t-pad-attenuator.html

 

[Rja4000]

Nice to see!

FWIW, for low level signals, I always use resistor attenuators with the source impedance set to match the likely source impedance of whatever is expected to be plugged into that input. If the source impedance is too low, you miss the current noise contribution; if it's too high, you can exaggerate the current noise contribution.

I adedd a few more details above

 

[Pluto]

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. This, in effect, is your reference source. A microphone amplifier should be designed to give the best possible noise figure (NF) when its gain is at the highest (say ≥ 60dB). NF is, broadly, defined as the amount by which the noise at the output of the amplifier is worse than the theoretical noise level of the input resistance alone plus the amount of gain (in this case, 60dB). A really good mic. amp. can achieve a NF in the region of 1dB.

When checking mic. amps. within a mixer, be aware that the virtual earth mixing buses can themselves be quite noisy, so if you are testing the mic. amps you really ought to be measuring the mic. amp output at the insert points, not the mixer outputs.

ISTR those Mackie VLZ amplifiers as being rather good at the time.

*NB this is truly important

 

[Rja4000]

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. This, in effect, is your reference source. A microphone amplifier should be designed to give the best possible noise figure (NF) when its gain is at the highest (say ≥ 60dB). NF is, broadly, defined as the amount by which the noise at the output of the amplifier is worse than the theoretical noise level of the input resistance alone plus the amount of gain (in this case, 60dB). A really good mic. amp. can achieve a NF in the region of 1dB.

When checking mic. amps. within a mixer, be aware that the virtual earth mixing buses can themselves be quite noisy, so if you are testing the mic. amps you really ought to be measuring the mic. amp output at the insert points, not the mixer outputs.

ISTR those Mackie VLZ amplifiers as being rather good at the time.

*NB this is truly important

Hi
I know. I've done that too.

Here is the noise level with 60dB gain, on 150 ohm.
-93.62 dBFS on a +24dBu range, with a 59.9dB gain, that gives us an EIN figure of -129.6 dBu

But what I want to measure is not just the noise.

If noise was the only difference between those preamps, than we just have one figure to look at.
We all know that's not true.

 

[pozz]

Nice work.

Do I understand you right that the measurements you posted so far are only for the Mackie, not for the Millennia?

 

[Rja4000]

Nice work.

Do I understand you right that the measurements you posted so far are only for the Mackie, not for the Millennia?

Yes, you're right. That's the old Mackie only for now.
Now that I have the test chain defined, I'll start testing more.

What I have at hand are
- The Mackie
- Yamaha DM1000 Mixer
- Yamaha AD8HR
- Focusrite Liquid 4 Pre
- Motu 828 MkIII (firewire only version)

Except the Mackie, all the others have no direct Analog out after the preamp.


Unfortunately, I don't own the Millennia (yet).
I've been looking at a second hand HV-3B last saturday, but unfortunately, one channel was out of specs by a comfortable margin.
So if I want one, I'm afraid I'd have to bite the big bullet.

But let's try the other ones first...

 

[Pluto]

an EIN figure of -129.6 dBu

Impressively good for a unit costing about €350. The ‘rule of thumb’ for mic. amps. was, ISTR, that -130 was “as good as it gets”.

 

[Rja4000]

Question for all: How to compare gain for preamps with no analog out ?

Gain for an analog amp is a straightforward notion. You measure signal before and after the amp, and you have it.
What about an amp when you may only look at the output through an ADC ?

I guess what you want to know is the minimum input level to get a 0dBFS output at max gain.

Maybe we can still guess approximatively the analog gain by looking at the EIN-like figure ?

 

[Rja4000]

Impressively good for a unit costing about €350. The ‘rule of thumb’ for mic. amps. was, ISTR, that -130 was “as good as it gets”.

I paid it 150€ second hand 10 years ago.

And one of my (then) "friends" borrowed it for a few months, "to use as a monitoring mixer in his studio"...
until I found it back in a Rock Cafe, where it was used for live (and much drunk) events for months.
I don't want to tell you in which state it is now.
With bent cursors, sugar (beer) inside, and missing buttons. Not to speak about dust.

That's Rock and Roll, man.

 

[Pluto]

How to compare gain for preamps with no analog out ?

A pro analog output stage should be able to deliver about +22dBU (perhaps a little more) so it's reasonable to align +20dBU with 0dBFS (digital). This means that -10dBFS aligns to +10dBU which feels about right. In a “mixed economy” (i.e. digital and analog co-existing), operationally, you would ensure that you are working with significant digital headroom as you really don't want to go anywhere near 0dBFS (not in a live situation, anyway) but for test purposes, simply decide on a reasonable alignment and stick to it.

NB there is no absolute standard to compare digital and analog levels. “Mixed economy” houses will operate with differing amounts of headroom depending on how close to the wind they like to sail!

 

[Pluto]

bent cursors, sugar (beer) inside, and missing buttons. Not to speak about dust.

But still working nonetheless.

Commendable.

 

[Rja4000]

By the way, one of the benefits of inserting the JDI in the loop is the "Ground lift" button

Before "Ground lift"

After "Ground lift"

RME ADI-2 Pro fs +4dBu range
=> Radial SAT-2
=> JDI Duplex (Grnd Lift)
=> Y cable
-> Focusrite Liquid 4 Pre, Mic input HiZ, "49" gain, FLAT preamp

Both the RME and the 4Pre get their clock from a Dante network interface (Focusrite Rednet 3) through AES/EBU

 

[Rja4000]

Unfortunately, I don't own the Millennia (yet).

I wanted to buy a "return" HV-3C yesterday.
I whish I did: It was discounted by over 500€. Today, it's back to normal price.
"Mistake on the web price" they say.
:-S

 

[Blumlein 88]

Question for all: How to compare gain for preamps with no analog out ?

Gain for an analog amp is a straightforward notion. You measure signal before and after the amp, and you have it.
What about an amp when you may only look at the output through an ADC ?

I guess what you want to know is the minimum input level to get a 0dBFS output at max gain.

Maybe we can still guess approximatively the analog gain by looking at the EIN-like figure ?

Those are the specs listed at least sometimes.
Minimum input to get max output. And max input at minimum gain.

 

[Pluto]

I guess what you want to know is the minimum input level to get a 0dBFS output at max gain

Nobody doing real mixing on a digital mixer will go anywhere near 0dBFS, at least not intentionally. The usual conventions dictate that there should be at least 10dB between the very highest levels likely to be encountered and 0dBFS. 20dB would be a better safety margin to aim for so I would be inclined to take whatever margin you decide on, into consideration when setting up your operating conditions for measurement.

 

[Rja4000]

Nobody doing real mixing on a digital mixer will go anywhere near 0dBFS, at least not intentionally. The usual conventions dictate that there should be at least 10dB between the very highest levels likely to be encountered and 0dBFS. 20dB would be a better safety margin to aim for so I would be inclined to take whatever margin you decide on, into consideration when setting up your operating conditions for measurement.

Sure.
With 24 bits digital and current ADC resolving 18 to 20 bits without issue, -20dBFS should be used as a target.
(Or anywhere between -10dBFS and -20dBFS, depending on the recorded signal expected level variations.)

But that doesn't change the question.

We need to compare the capability of the preamp+ADC as a whole.
For that, we need to know what's the "gain" they allow.
And the only way I could think about to achieve such a comparison is to check what's the dBFS level at maximum gain for, say, a -60dBV (1mV rms) signal.

Of course, that doesn't give us the analog stage gain, because we don't know the dBV/dBFS correspondance after the analog stage, at the input of the ADC.

But that's a figure we can compare.

And if we want to compare to an analog preamp, we have to take the analog preamp headroom into account (which is quite high for the Mackie, as an example).
Because if we add an ADC like the RME behind the Mackie, that's what we will do: align 0dBFS with the max level before unacceptable distortion of the preamp analog stage, as far as possible.

So I'm measuring that.
On top of noise measurement.
(Which may give us a clue of what the actual analog gain is)