elektret mike amplification

Or a reasonable explanation of why it would be better. If no better counter-arguments come up, then I don't need to show anything. Like in a thought-experiment.

Don't hold your breath. :)

joe

Any idea how much equivalent sound level the Brownian noise contributes? Some of the MEMS miks are really tiny.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Yes, thanks.

joe

What is the advantage of multiple mics in each array?

Mikek

The coherent wanted sound signal increases by 6dB each time you double the number of capsules, but the random Brownian and head-amplifier noise only increases by 3dB. This gives 3dB improvement in S/N ratio for each doubling of the number of capsules.

My design used 16 capsules in each block, giving 12dB improvement over the basic capsule and getting the diaphragm area and the Brownian noise into the same category as the 4038, which I was trying to emulate.

--
~ Adrian Tuddenham ~ 
(Remove the ".invalid"s and add ".co.uk" to reply) 
www.poppyrecords.co.uk

** Sound waves impinging on the diaphragm of a condenser mic cause capacitance changes - while the stored charge remains fixed.

Capacitances changes cause a signal voltage to be generated that mimics the sound wave.

... Phil

Yes, that's the electrical engineering model. I think the physical explanation is what I wrote in another reply:

That, due to the fixed charge getting closer to the plate opposing the diaphragm, the electrical field changes which makes that opposite plate more (or less) polarised.

I propose that it is this charge that generates the voltage over C_el+Cgs (the electret capacity resp. + Cgs), but maybe I'm a lousy physicist and got this all wrong (that's why this is a 'proposal', to be shot down if you find a hole).

If I'm right however then it would give a better SNR if the gate were kept at zero by an nfb loop with a value of Cfb as transcapacitance as this would generate a voltage gain with respect to the old situation of (Cgs+C_el)/Cfb and with the same noise. Hence the improved sensitivity and SNR.

joe

The field between the plates is fixed, because the geometry is parallel plates and the charges on the electret can't move. It's the voltage that changes. From a physicsy perspective, the moving charge creates a displacement current equal to q*velocity, which is what's sensed by the other plate.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

...but if the voltage were kept fixed and charge was allowed to run in and out of the low impedance terminal of a suitable amplifier/feedback arrangement, the system would work equally well (provided that the voltage output of the amplifier corresponded to its current input).

Because there would be no change in voltage on the parasitic stray capacitances, they would not need to be charged and discharged by the 'working' (variable) capacitance and would effectively be non-existant. The parasitic stray capacitances are considerably larger than working capacitance of most types of capacitor mic, so that effectively removing them from the circuit should give a significant increase in sensitivity.

The question is whether this would give any improvement in S/N ratio compared to the high-input-impedance source follower, bearing in mind that an entirely different low-noise amplifier configuration would be needed. There is also the question of stability when the 'virtual earth' node of the system is effectively connected to earth by the stray capacitance.

--
~ Adrian Tuddenham ~ 
(Remove the ".invalid"s and add ".co.uk" to reply) 
www.poppyrecords.co.uk

That's the TIA vs bootstrap distinction. It turns out that (1) the noise is the same, because the strays only contribute by the e_N*C_in mechanism, an d that operates the same way in both cases; and (2) you can combine the two approaches and get about a 3dB improvement.

Bootstrapping is great in DC-coupled situations because you can use a quiet device such as a BF862 single-ended without screwing up the offset.

Cheers

Phil Hobbs

Ok, so you agree it's a charge displacement.

joe

No. If you've forgotten your fields classes, GIYF: 'displacement current'.

Cheers

Phil Hobbs

Exactly, that's the goal: keep the voltage on the electret fixed. The charge amplifier has a very low impedance if the loop gain is sufficiently high.

If one would follow Phil Hobbs' reasoning, the movement of the diaphragm causes a displacement current. Then, if one would insist on working according to E. Nordholt's design philosophy, a transimpedance amplifier (with impedance = resistance) to convert and amplify this current into a voltage.

Right on, that was my whole point.

Yes, that would be an interesting comparison. Mind you however that my proposal only involved an improvement with respect to the current, cheap, electrets with the common source JFET and the two-wire drain+source connection to the amplifier.

An elementary set-up would be to just put a small nfb capacitor between drain and gate in the current electret, together with a 'suitable' (mind the noise) biasing circuit.

Not sure what you mean? Do you insist on using a single stage source follower?

thanks,

joe

I seem to recall that loading a virtual-earth input with a capacitor to real earth can give some unpredictable effects in practice (because the system cannot achieve theoretical perfection). Phase-shift outside the wanted frequency band could lead to a peak in the response or complete instability.

A purely capacitive feedback circuit into an input parasitic capacitance to earth will impose heavy loading on the amplifier's output stage at high frequency and may lead to slew-rate limiting and all sorts of other evils. This could cause bursts of oscillation on steeply-rising wavefronts.

If you are trying to work out a practical design, this is something to beware of.

--
~ Adrian Tuddenham ~ 
(Remove the ".invalid"s and add ".co.uk" to reply) 
www.poppyrecords.co.uk

Ok, got it.

There are also parasitic resistances and the capacitances are quite small, and the frequencies are audio only.

Thanks, I'm not. I was just proposing and frankly am still trying to work out how and electret exactly does work, as in visualizing, not just by writing rot x H = J + dD/dt or so. :) I want to 'see' where the charges are coming from to find out whether it is a charge 'pump' charging a capacitor, or whether it's a (magical) voltage source.

joe

When you move the plates of a charged capacitor, the voltage across it changes. AIUI that's all a condenser mic is. When you load that with external R or C, charge naturally flows in & out, reducing voltage amplitude.

NT

My problem was that the capacity does not change in the elektret microphone where the polarised membrane moves between 2 capacitor plates. Or maybe I got the construction details wrong?

joe

Normally in a capacitor microphone there is one fixed plate and one moving plate, so the capacitance does vary.

(Most electrostatic loudspeakers use two fixed plates and one moving one; this may be where the confusion has come from.)

--
~ Adrian Tuddenham ~ 
(Remove the ".invalid"s and add ".co.uk" to reply) 
www.poppyrecords.co.uk

In such a construction C from one fixed plate to the moving one does change. Having 2 fixed plates making 2 series caps that change in opposition makes no sense that I can see. But I'm no expert on them by any means.

NT