Question about condenser mic capsules & non-linearities

Hello,

Please allow me to start with:

C = ( eps_r * eps_0 * Area ) / ( platedistance - membranedisplacement )

So as known the capsule is a 'non-linear capacitor' w.r.t. membranedisplacement.
('inversely proportional and a shift')

Potentially dumb question ahead: is this a indeed a significant/noticable
contributor to microphone performance flaws ?

Any tricks existing to tackle this (eventual) problem ?

And where could I perhaps read more ?
I saw various AES-articles about models etc, might be time for a subscr.


Thanks !

  Peter

clintrubber said:

Hello,

Please allow me to start with:

C = ( eps_r * eps_0 * Area ) / ( platedistance - membranedisplacement )

So as known the capsule is a 'non-linear capacitor' w.r.t. membranedisplacement.
('inversely proportional and a shift')

Potentially dumb question ahead: is this a indeed a significant/noticable
contributor to microphone performance flaws ?

Any tricks existing to tackle this (eventual) problem ?

And where could I perhaps read more ?
I saw various AES-articles about models etc, might be time for a subscr.


Thanks !

  Peter

Click to expand...

This is indeed a known fact since the beginnings of condenser microphones. This distortion is more or less compensated by the compressionb between diaphragm and backplate, which is also non symmetrical. AFAIK, the subject has been the subject of several white papers, but no magis solution has appeared. The abject truth is here: we all have been lured into thinking that Frank Sinatra had natural third-order distortion!

Hi,

Thanks for responding, much appreciated. Any links to those whitepapers perhaps that come to mind ?

And surprising there isn't a definitive solution yet, or is it just OK like this ?
As in: it actuall sounds 'better' with it.

Can't imagine no-one tried to eliminate this imperfection.

Regards,

  Peter

Somehow I do think to remember that the DC bias of the capsule reduces the distortion. I don't see it right now though (at least with the available 30 s to think about it) so perhaps I'm off.

Samuel

http://recforums.prosoundweb.com/index.php/m/137141/0/?srch=capsule#msg_137141

Samuel Groner said:

Somehow I do think to remember that the DC bias of the capsule reduces the distortion. I don't see it right now though (at least with the available 30 s to think about it) so perhaps I'm off.

Samuel

Click to expand...

Hi & thanks for chiming in.

Maybe because of the  DC >> AC ? (and the AC being related to the shifts on the curve)
If it's indeed because of this then higher DC-values & identical AC would mean less non-lin.
- but at the same time perhaps a higher required plate distance to cope with a high DC, so perhas
no net effect...

Bye,

  Peter

Gus said:

http://recforums.prosoundweb.com/index.php/m/137141/0/?srch=capsule#msg_137141

Click to expand...

Hi,

Thanks Gus, I recall having seen that pdf he references to a while ago,
but didn't remember it today.
Great, thanks for bringing it up again !

Bye,

  Peter 

Gus said:

http://recforums.prosoundweb.com/index.php/m/137141/0/?srch=capsule#msg_137141

Click to expand...

But some second thoughts now on the problem spotted by Mr. Lavry:

Q = C*V

&

C = ( eps_r * eps_0 * Area ) / ( platedistance - membranedisplacement )

so

Q =  ( eps_r * eps_0 * Area ) * V / ( platedistance - membranedisplacement )

so with Q & Area being constant (and let's assume the same for eps),
the relation between the generated signal voltage V & ( platedistance - membranedisplacement )
IS linear.
OK, there is an additional term in the denominator, but we're not talking about inversely proportional anymore.

What am I doing wrong ?
I'd say that the actual capacitance-value could be considered to be just an intermediate variable,
why worry that it's inversely proportional to the generated voltage ? (actually vv: delta-C first, then delta-V, but OK).


Note that I assume that the relation between the incoming soundwaves & the membranedisplacement is linear,
maybe that's not completely valid ?

Anyone ?


Regards,

  Peter

Anyone ?

I've also posted to the micbuilders-list, let's see what that gives.

Bye,

  Peter

Hi,


Got a response from fellow-GDIYer ZAPNSPARK that confirms & completes the above, thanks!

As I understand it then, a proportional relation (iso inv.prop.) between diaphragm-displacement and C (which can be done) is actually unwanted, since it doesn't 'repair'/compensates the inv. prop. between C & V any longer...


Regards,

  Peter

clintrubber said:

Please allow me to start with:
C = ( eps_r * eps_0 * Area ) / ( platedistance - membranedisplacement )
So as known the capsule is a 'non-linear capacitor' w.r.t. membranedisplacement.
('inversely proportional and a shift')

Click to expand...

This is too crude simplification to
describe nonlinearity in capacitor mic. When diaphragm
go to much close to backplate, capacity goes to infinity-which causes
near zero output voltage, but force goes also to infinity, but nonlinearly.
This process linearise ideal condenser transducer in the way, that
idealizing of the real capacitor microphone (nulling parasitic capacitance via pure electric way-Frederiksen)
makes system distortion lower.

Better way is to assume microphone as conservative systhem, like here:

http://radio.feld.cvut.cz/~vlk/my/poster08paper.pdf

Hi Michal,

Thanks for the response, much appreciated. I'll be reading your paper!

Bye,

  Peter

some comments to paper:
nonlinear VCCSs are realised as SPICE directives (that two rows),
and grounding of the sources must be done via small resistance. It
is needed because property of SPICE discussed here some months ago,
that SPICE analyses only some subgroup of circuit.
(But all normally produced ICs belongs to this group, but
models of transducers like that in paper and some special low noise topologies
not.) It is very frustrating to think up where parasitics must be added
only for doing some simullation. And this idle work {which consumes much more time than
circuit model development} I must do only to satisfy SPICE's
perversity.