Microphone FET selection

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

kingkorg

Well-known member
Joined
Apr 15, 2017
Messages
1,788
Location
Norway
I believe all that cgs capacitance thing is blown out of proportion. LDC capsules have passive (constant and non variable) capacitance of 50-100pf. So even 20pf cgs doesn't change things much.

Sapsule swings maybe ~10pf at quite decent (LF!) spl level. Speech at some distance swings in in maybe 1, 2pf range. All that passive capacitance is in parallel the whole time.

Which is why i was interested in the negative capacitance concept, but can't still wrap my mind around it.
 
Last edited:

ccaudle

Well-known member
Joined
Mar 18, 2010
Messages
493
Location
Houston
that LSK170 have rather high input capacitance (Cgs), about 2pF.
...
2N4416 has only about 2-3pF Cgs

Typo? I just checked the data sheet, I think LSK170 has about 20pF typical input capacitance. The Linear Systems version of 2N4416 has less than 1pF maximum.

So I guess it depends on how you are using the JFET. I was thinking of source follower, the gate to source capacitance doesn't really matter there, and the LSK170 has low voltage noise.
The 2N4416 is probably low noise, but the Linear Systems datasheet only qualifies it at high frequencies, so hard to be sure.
 

abbey road d enfer

Well-known member
Staff member
GDIY Supporter
Moderator
Joined
Jan 22, 2008
Messages
14,996
Location
Marcelland
Yes.
I just checked the data sheet, I think LSK170 has about 20pF typical input capacitance. The Linear Systems version of 2N4416 has less than 1pF maximum.

So I guess it depends on how you are using the JFET.
Correct. That's why I mentioned bootstrapping (which source follower does). However the penalty is that the noise of the subsequent stages add directly to the input noise, instead of being more or less swamped when the input stage has gain.
Actually, the best topology is probably cascode, which not many DIY'ers embrace.
 
Last edited:

abbey road d enfer

Well-known member
Staff member
GDIY Supporter
Moderator
Joined
Jan 22, 2008
Messages
14,996
Location
Marcelland
I believe all that cgs capacitance thing is blown out of proportion. LDC capsules have passive (constant and non variable) capacitance of 50-100pf. So even 20pf cgs doesn't change things much.
Do you think it applies to an SDC? 20pF is pretty close (even larger) than many SDC capsules. That's an instant 6dB loss of signal.
Which is why i was interested in the negative capacitance concept, but can't still wrap my mind around it.
Negative capacitance involves active circuitry, which will probably be as noisy as the head amp. Not sure because I haven't really dug into it. I think if it was a working concept, one of the big names would have adopted it.
The closest thing is a topology I've seen in some Sony mics, where the FET is bootstrapped two ways, once on the source, and once on the drain. I don't have the schemo at the ready, will have a look.
 

kingkorg

Well-known member
Joined
Apr 15, 2017
Messages
1,788
Location
Norway
Do you think it applies to an SDC? 20pF is pretty close (even larger) than many SDC capsules. That's an instant 6dB loss of signal.
Sure, in case of sdc, cgs would play much bigger role. Small capsules swing even less in capacitance. Something like Rode NT1 circuit with increased polarization voltage (100v if capsule can take it) might be a solution. If i understand well, Nt1's cgs is bootstraped.
 

ccaudle

Well-known member
Joined
Mar 18, 2010
Messages
493
Location
Houston
The closest thing is a topology I've seen in some Sony mics, where the FET is bootstrapped two ways, once on the source, and once on the drain
If you find it please post the model number. Sounds like some of the AKG circuits (AKG460 uses an op-amp as part of the boostrap circuit).
 

abbey road d enfer

Well-known member
Staff member
GDIY Supporter
Moderator
Joined
Jan 22, 2008
Messages
14,996
Location
Marcelland
I haven't found this schemo. I remember that it used the signal at the FET source to modulate the supply voltage going to the drain. The explanation was that it gave better linearity by preventing modulation of Cgs, at the cost of a loss of gain.
 

abbey road d enfer

Well-known member
Staff member
GDIY Supporter
Moderator
Joined
Jan 22, 2008
Messages
14,996
Location
Marcelland
A strange question for a person who understands microphone circuitry.
Not so strange. I understand microphone circuitry, amongst many other, but I can't remember all the Neumann circuits and their variations. Quite often members mention a piece of gear, expecting that the schematic is known to every other member; it's not the case. When referring to a particular circuit, it is good to make sure the schemo is posted in the thread.
 

MicUlli

Active member
Joined
Mar 18, 2022
Messages
28
Location
Germany
I would recommend the following simplest circuit for first experiments. JFET J111 is one of the cheapest fets you get on the market and works quite well.
 

Attachments

  • simple_electret_P48.JPG
    simple_electret_P48.JPG
    59 KB · Views: 47

sellch

Member
Joined
Dec 21, 2020
Messages
9
I would recommend the following simplest circuit for first experiments. JFET J111 is one of the cheapest fets you get on the market and works quite well.
good morning
have a good Sunday!
could this circuit work on a ribbon microphone?
do you know how much noise and gain it has?
Thank you.
 
Last edited:

abbey road d enfer

Well-known member
Staff member
GDIY Supporter
Moderator
Joined
Jan 22, 2008
Messages
14,996
Location
Marcelland
good morning
thank you!
could this circuit fit into a ribbon microphone?
It may, but you would need a high-ratio transformer in between. FET's prefer high-impedance sources, which a ribbon definitely isn't.
do you know how much noise and gain it has?
Gain is about unity.
Regarding noise, simulation with LTspice gives very pessimistic results. I think the spice model for J111 is not correct for noise.
Simulating with other FET's show that the noise level is roughly that of the FET itself.
For example a 2N3819 has a noise voltage density of 3nV/sqrtHz at 3kHz. I choose 3kHz because it's where audition is most sensitive to noise.
The noise level for a bandwidth of 20kHz is then 3nV. sqrt(20 000) or 423nV=>0.423uV or -125dBu. Take it as a ballpark figure.
Anyway, the acoustic impedance noise of the capsule is probably several dB above that.
 
Last edited:

sellch

Member
Joined
Dec 21, 2020
Messages
9
It may, but you would need a high-ratio transformer in between. FET's prefer high-impedance sources, which a ribbon definitely isn't.

Gain is about unity.
Regarding noise, simulation with LTspice gives very pessimistic results. I think the spice model for J111 is not correct for noise.
Simulating with other FET's show that the noise level is roughly that of the FET itself.
For example a 2N3819 has a noise voltage density of 3nV/sqrtHz at 3kHz. I choose 3kHz because it's where audition is most sensitive to noise.
The noise level for a bandwidth of 20kHz is then 3nV. sqrt(20 000) or 423nV=>0.423uV or -125dBu. Take it as a ballpark figure.
Anyway, the acoustic impedance noise of the capsule is probably several dB above that.
thank you very much Abbey!
Did you see in the LTspice program what output impedance?
 

abbey road d enfer

Well-known member
Staff member
GDIY Supporter
Moderator
Joined
Jan 22, 2008
Messages
14,996
Location
Marcelland
thank you very much Abbey!
Did you see in the LTspice program what output impedance?
Yes. About 10 ohms. That's the benefit of the emitter follower. It doesn't mean you can drive a speaker, though. :)
Note that the impedance increases at LF due to the two 22uF capacitors that are in series AC-wise.
 
Top