FET choice for Mic front end ?

[Val_r]

I was intrigued in using a 2SK170.
In some article I read there are better choices, since the 170, even if has low noise, lacks in correct transient reproduction.
Any suggestion more than welcome.

 

[Gus]

What?

What article are you posting about?

It is all about the designers knowledge.

 

[Val_r]

Came across this site, which offers affordable mic mods.
At the end of the page, he states that the K170, because of it's high input capacitance, does not offer a correct transient response.

I just built a front end electronic using those K170, and not experiencing those problems.
I can get a 30V/uS slew rate and low distortion.

I was planning to try the 2N4416A

Any opinion?

 

[Gus]

Maybe he does not understand how a jfet in some microphone circuits work?

 

[rodabod]

I wouldn't think you'd have to worry so much about slew rate at the levels we are talking about here.

Don't forget to use "Evidence Audio LYRIC HG solid core copper wire".  :-\

 

[Jonte Knif]

High input capacitance simply acts like a pad. It is in parallel with the capsule C. There is no "slew rate", so "transient reproduction" can not suffer. Neither can freq response.

But I _can_ imagine reasons not to use K170. I wouldn't like to deal with the fairly big nonlinear part of the Cin, it causes some distortion.

It will probably work just fine, but check other j-fets too. I would look after somewhat lower Gm and input C and transfer C. (Gm and C:s go hand in hand)

But this all depends on the topology. If the K170 is cascaded, or is used as a follower then you have less problems. In a U87-like circuit you get the most problems.

 

[Samuel Groner]

If the K170 is cascaded.

You probably mean cascoded.

Samuel

 

[Jonte Knif]

It was a typo, it was a typo, it was....a mistake.

 

[burdij]

2N3819 - you can get a big bag of them for $13.

 

[Val_r]

Hey,

Thanks for the replies guys!

High input capacitance simply acts like a pad. It is in parallel with the capsule C. There is no "slew rate", so "transient reproduction" can not suffer. Neither can freq response.

But I _can_ imagine reasons not to use K170. I wouldn't like to deal with the fairly big nonlinear part of the Cin, it causes some distortion.

It will probably work just fine, but check other j-fets too. I would look after somewhat lower Gm and input C and transfer C. (Gm and C:s go hand in hand)

But this all depends on the topology. If the K170 is cascaded, or is used as a follower then you have less problems. In a U87-like circuit you get the most problems.

The topology:
capsule -> K170 as impedance converter (source follower) -> K170 diff. input discrete op amps, 1 in non-inverting and 1 in inverting -> output.
The ht for the capsule comes from an external psu.

The front K170 has another K170 used as a ccs, feeding it's source terminal.

2N3819 - you can get a big bag of them for $13.

It has a higher input capacitance: (ref. 4416) datasheet says around 3pF. Yfs in the range (3 to 6.5 mS). Even though better than the typical 30pF Ciss of K170 !

The Siliconix J304 does a little better: Ciss=2.2p, Gfs between 4.5 and 7.5 mS.
Mouser stocks the J305. I hope they give the same result. Datasheet says they are the TO-92 version of the 2N4416.

Comment!

 

[Marik]

Tricky to use, but with all its flaws I like the 2SK170 in the mics. I don't see its higher Ciss as a drawback, as it helps reducing nonlinearity of the capsule' capacitance/voltage transfer and reduce distortions. Really, it is all about where to make compromises.

Best, M

 

[Sonic]

As variant BF245: Cis=4pF, Yfs = 3 to 6.5 mS

 

[tk@halmi]

Mouser stocks the J305. I hope they give the same result. Datasheet says they are the TO-92 version of the 2N4416.
Comment!

I use the J305 in KM-84 type circuits with good results. I had better luck finding low Idss high headroom J305s than 2N3819s using Fairchild parts. Your experience will vary from batch to batch. It is almost fun to measure and grade 1000 transistors while you watch TV.

The J305s actually need a gate-drain capacitor for feedback to reign in high frequency crunch. The 2SK170 has it built right in there. But the two require different circuits to work well. As Mark said the high input capacitance is really not a problem when paired up with the right capsule and circuit.

There were some really awesome high gain 2SK parts that made easier to implement mic amp circuits than the 170, but Toshiba has discontinued all of them a good while ago.  

 

[Val_r]

The capsule I am using has a 58 pF capacitance.
The 2SK170 works fine, the following circuit gives plenty of gain, so no sensitivity issues there.

Mic has a consistent output level even with the mic preamp set at 20dB of gain!

What are those Toshiba parts you were referring to?

 

[Sonic]

J201 Itself did not try in microphones, unlike BF245, but on datasheet it is more interesting , than J305.

 

[Jonte Knif]

tk,

I'm no expert, so could you please explain a bit.

The J305s actually need a gate-drain capacitor for feedback to reign in high frequency crunch. The 2SK170 has it built right in there. But the two require different circuits to work well. As Mark said the high input capacitance is really not a problem when paired up with the right capsule and circuit.

I don't get it. A condenser capsule is not a resistive source and any C placed in parallel just pads it. What is the high freq crunch you are referring to? As far as I understand the nonlinearity of the fet capacitance causes high freq distortion with resistive sources but in this case it is not frequency dependent because the source is capacitive..

If you increase the capacitance with external cap (in this case with miller effect I guess, but still) you will make the j-fet nonlinear capacitance smaller in proportion. Is it this you are referring to or general padding? Or subjective result? I guess I don't understand all the effects of feedback in this case.

-Jonte