Connecting a capsule

[abbey road d enfer]

Anyways, I wonder why is it so much easier for some people to latch onto a side remark or bad phrasing of mine rather than try to talk about the main topic?

Because some of your remarks show that you have parcellar knowledge of electricity basics. If you think I commented on your phrasing, you are definitely wrong.

This far nobody has been able to explain it in a way I and my dirty fingers understand it. Instead there has been unbelievable amount of (deliberate?) misunderstanding and effort to lead the conversation off track. Love this forum regardless and I'm happy to be back here <3

I thought I had given a pretty good answer earlier, but it seems it's gone to another thread, so I'll give it again.
When the capsule's diaphragm is submitted to the bias voltage, a capacitor is mandatory, in order to prevent the high voltage to first ruin the FET's bias and second to blow it up.
When the backplate is receiving bias, a capacitor is not needed, because the diaphragm's voltage is determined by the FET bias.
In addition, when bias is applied to the diaphragm, it takes a high-Z resistor (typically 1G) that happens to be in parallels with the FET's bias resistor (usually another 1G). It results in halving the load on the capsule, which results in 3dB more noise.
As I wrote somewhere else, it's a case of parallels vs. series arrangement.
Of course, backplate and diaphragm could be interverted, but the higher intrinsic capacitance of the backplate makes it more sensitive to electrostatic interference.

 

[abbey road d enfer]

Doesn't make a big difference, only affects the low frequencies which is expected. So in a way @Khron is right, it's a filter cap.

Still, you definitely can't omit it, there has to be a path to ground (surprisingly though you get a really weak signal in simulation even if you leave the other side of the capsule floating; the mic wouldn't be usable though). So it's definitely in the signal path as I see it.

I Can't judge clarity from the simulation and have to try it in real life when I get back home (travelling atm) but I believe what is written above makes sense. Also if it was good practice to leave it out we would have seen mics implemented this way. So it's not JUST a filter cap imo.

(This is the circuit I used for the simulation, a really basic common source circuit built in a Behringer B1 using the stock capsule. C2 is the capacitor of interest. You can't omit it but you can short it to ground (with possible side effects). If you wonder about the 100 meg resistors and the low value coupling and bypass caps: it's meant to be hand held and used for really close miked vocals and it really seems to do that quite well.)

If you omit the cap, the circuit that feeds the amp is a RC series circuit with 50pF (the capsule) and 100Meg from a point that's AC-wise grounded.
With the cap, the source impedance is just 50pF. Of course it makes a difference.
I don't think anyone in their right mind would suggest to delete the cap. However, there may be different topologies where this wouldn't apply.

 

[pasarski]

Because some of your remarks show that you have parcellar knowledge of electricity basics. If you think I commented on your phrasing, you are definitely wrong.

I thought I had given a pretty good answer earlier, but it seems it's gone to another thread, so I'll give it again.
When the capsule's diaphragm is submitted to the bias voltage, a capacitor is mandatory, in order to prevent the high voltage to first ruin the FET's bias and second to blow it up.
When the backplate is receiving bias, a capacitor is not needed, because the diaphragm's voltage is determined by the FET bias.
In addition, when bias is applied to the diaphragm, it takes a high-Z resistor (typically 1G) that happens to be in parallels with the FET's bias resistor (usually another 1G). It results in halving the load on the capsule, which results in 3dB more noise.
As I wrote somewhere else, it's a case of parallels vs. series arrangement.
Of course, backplate and diaphragm could be interverted, but the higher intrinsic capacitance of the backplate makes it more sensitive to electrostatic interference.

Yes, but this is not what I'm interested in. You can swap the backplate and the membrane in both of the hookups, the capsule is symmetrical. So it's not about which one should be polarized, the membrane or the backplate.

(Edit: Never have had any issues with intrinsic capacitance of the backplate and never have heard this before, but it's good info Some say it makes a difference if you take the signal from membrane or the backplate and this would explain it, but I don't hear a difference. Also this is out of scope of my original question.)

And yes, my knowledge about electronics is not great, but you are misunderstanding the question I'm trying to ask. Maybe my last longer message clarifies it?

 

[pasarski]

If you omit the cap, the circuit that feeds the amp is a RC series circuit with 50pF (the capsule) and 100Meg from a point that's AC-wise grounded.
With the cap, the source impedance is just 50pF. Of course it makes a difference.
I don't think anyone in their right mind would suggest to delete the cap. However, there may be different topologies where this wouldn't apply.

Someone did, I wouldn't have brought it up otherwise.

 

[pasarski]

Here FD is connected via C15=1nF to the Gate of jFET Q7 and BP is polarized directly from the DC/DC converter with positive voltage.

Why is the Backplate BP not connected to ground via a capacitor?

How can it work?

The scheme belongs to the B1 a good microphone, similar to another Neumann scheme.

I think the schematic you posted might be incorrect (which makes my "analysis" of it also incorrect). In the one presented in this post there's a 22nF cap (C16) connecting the other side of the capsule to the ground.

https://groupdiy.com/threads/behringer-b1-mods.81332/

Also, in this picture you can see that there really is a cap named C16 which is missing from your schematic so yeah, it's probably incorrect and probably doesn't work properly like that.

 

[MagnetoSound]

C2 is the capacitor of interest. You can't omit it but you can short (edit: bypass?) it to ground (with possible side effects).

No, you can’t short the backplate cap to ground, because you will lose the polarising voltage.

 

[pasarski]

No, you can’t short the backplate cap to ground, because you will lose the polarising voltage.

Yes of course

 

[pasarski]

Okay, I got finally a explanation that satisfies me in a private convo with abbey. Apparently this originally got lost in the thread separation:

abbey:
"You should just consider the actual impedance of the source that feeds the amp.
In one case, you have a small capacitance (the capsule) in series with another small capacitance (the gate capacitor) and two resistors in parallels.
In the other case, you have the capsule in series with a large capacitor and only one resistor in parallels.*
Distortion in capacitors is proportional to the voltage across it.
Since the capacitor that feeds the backplate is of a very much larger value than the capsule, the AC voltage across it is extremely small, so distortion is also very small.
Now, one may replace the small gate capacitor by a larger one, which would solve the distortion problem. However, the large parasitic capacitance of this large capacitor would result in problematic loss of signal and susceptibility to electrostatic interference.

*Actually, I'm neglecting the parts that deliver the B+ voltage. They are also in the signal path, but for the sake of analyzing the circuit, B+ is considered a "perfect" voltage generator.
In practice it's generally close enough, however in some cases, the B+ circuit can be less than perfect and result in degradation of performance."


pasarski:
"Only explanation this far is that in the other scenario the cap is not in the signal path which doesn't seem to hold water."

abbey:

"Since the AC voltage across the capacitor is very small, thus its contribution to distortion is negligible, some consider it is just not there, which is wrong in theory, but right in practice."


So now I think it boils down to this(?):

The cap is in the signal path in both scenarios, but in the "direct connection" one there's less distortion.

I would guess the distortion is quite small on both cases though?