Microphones with OpAmp and de-emphasis feedback network

[micolas]

@Itsme
Opamps can be very linear thus creating very little distortion (harmonics).
With the addition of a feedback network it is fairly easy to add arbitrary emphasis or de-emphasis on both ends of the frequency range.
I could post some simulations to demonstrate some variations of the circuit and explain how that works.

 

[micolas]

The topic is broad, but in particular I was thinking about an NFB adapted to OPIC / OPIC TX circuits of @rogs , that would do deemphasis for the bright k87 capsules.
Any idea is welcome.

 

[Itsme]

Before we start:
Be aware that simulations have some shortcomings but they allow to examine the behavior of simplified models of a real circuit.
What it doesn't show without explicitly modeling it, is the influence of parasitics i.e. elements that are properties of the PCB, the parts themselves aging, dirt and more...
To get deeper insights into these (mostly undesired) properties, they must be modeled in addition to the main circuit.

 

[micolas]

In addition to attenuating high frequencies, the NFB network would reduce distortions. The perfectly justified question is: will it improve SNR like in the case of the 87 topology?

 

[Itsme]

The perfectly justified question is: will it improve SNR like in the case of the 87 topology?

Yes, from a total signal chain perspective. The initial SNR is defined by the properties of the capsule, it's biasing and the input properties of the opamp attached to the capsule.
We can have a closer look at this topic once I have tidied up my simulation files.
In the meantime I suggest that people who have interest in simulating the circuit(s), download and install the latest version of LTSpice from Analog Devices' website if not already done so...
I will share the simulation files for everybody to play with, if interested.

 

[Itsme]

Here is the basic circuit with added feedback network. I changed the Opamp to a very similar one out of the LTSpice library.
I also added the phantom power network to keep the supply part realistic. If you spot any mistakes, please let me know so I can correct them.

 

[Itsme]

Here a variant with de-emphasis. Note: I picked random values for the filter characteristics.
I used the .step directive to demonstrate three different 'strengths' of the de-emphasis (R13).

 

[rogs]

In addition to attenuating high frequencies, the NFB network would reduce distortions......

A non-inverting buffer has a direct connection between output and inverting input. ... how does introducing extra passives into that path improve the distortion figures?

 

[Khron]

Here a variant with de-emphasis. Note: I picked random values for the filter characteristics.
I used the .step directive to demonstrate three different 'strengths' of the de-emphasis (R13).

You might wanna consider using "AC 1" for AC analyses - that way, the graph will be referenced to 1V = 0dB.

(Clipping or other such distortion is not an issue in .AC simulations)

 

[micolas]

A non-inverting buffer has a direct connection between output and inverting input. ... how does introducing extra passives into that path improve the distortion figures?

It was intended to be a question. The simulations were supposed to answer all of that.
The main goal is not to reduce distortion.
Opic Tx seems very attractive to me precisely because it introduces saturation

 

[micolas]

...
The main goal is not to reduce distortion.
Opic Tx seems very attractive to me precisely because it introduces saturation

It would be great if Opic/OpicTx were a modern, elegant alternative to the U87 topology.

 

[Itsme]

A non-inverting buffer has a direct connection between output and inverting input. ... how does introducing extra passives into that path improve the distortion figures?

Depends on which distortion is meant...
If the capsule itself is distorted a bit of de-emphasis would affect harmonics in the attenuated range.
If someone is not happy with the distortion of the opamp more open-loop gain helps. This would mean to introduce a second opamp which would be quite a rabbit hole...

 

[Itsme]

Due to their high open loop gain opamps are significantly more linear than JFETs or tubes resulting in much lower distortion. However fans of the 'tube sound' would certainly be disappointed .
Edit: I like tubes btw.

 

[micolas]

Due to their high open loop gain opamps are significantly more linear than JFETs or tubes resulting in much lower distortion. However fans of the 'tube sound' would certainly be disappointed .

The OpAmp sounds very clean, which is why I also named the @Rogg OPIC.TX circuit as an excellent candidate for these experiments with de-emphasis options.

I like tubes btw.

Who doesn't like the sound of tubes?

 

[micolas]

@Itsme
Can you do some simulations for the feedback network in the case of the @rogs OPIC.TX circuit as well?

 

[Itsme]

I tried to find a model or at lest a data sheet that contains sufficient parameters to derive a half-decent model - so far: no cigar
Does anyone have detailed data or a spice model for a similar transformer? The Neutrik NTE 1 data sheet is a joke btw...

 

[micolas]

I tried to find a model or at lest a data sheet that contains sufficient parameters to derive a half-decent model - so far: no cigar
Does anyone have detailed data or a spice model for a similar transformer? The Neutrik NTE 1 data sheet is a joke btw...

I was thinking of a good transformer, with clearly stated specifications, and just putting the secondary on it twice (secondary:secondary) just for simulations.

 

[micolas]

I also thought of something else.
For LDC I usually use a polarization voltage of the capsule of 70...80v, it depends on the capsule, circuit, source, etc.
If we increase this voltage, and we reduce the gain of the OpAmp, the input sensitivity, could we improve the SNR?
The feedback network would be adjusted later.

 

[Itsme]

If we increase this voltage, and we reduce the gain of the OpAmp, the input sensitivity, could we improve the SNR?

Yes, given the capsule is specified for that voltage.
That way the signal gets larger while the input (voltage-, current-) noise of the opamp remains the same ==> higher SNR

 

[Екатерина]

Good day，I measured an EE19 transformer, which is used for isolated output with a transformation ratio of 1:1, a single-sided DC resistor of 160, and an inductance of 32H 100HZ. Of course, its frequency band is 20Hz~20kHz ± 0.5db, 10Hz~40kHz ± 1.5db, which is better than NTE 1. These parameters can be used