2SK117 replacement for a Curiously Low Noise Amplifier

[mycroft]

Folks,

I want to build a couple of these Curiously Low Noise Amplifier from Charles  Wenzel site (http://techlib.com/electronics/audioamps.html#curiously%20low%20noise) but the input JFET is no longer available. I intend to use them to measure power supply ripple and noise and some low noise sensors. Any suggestions on a suitable replacement?

Thanks.

[andy3055]

Here is a source: https://www.digipart.com/part/2SK117

[Kleinstein]

The SK209 is relatively similar, just in SOT23 case.  To92 or similar case JFETs are becoming more and more difficult to get.
Anyway also keep an eye on the sub-group (e.g. color code) for Ids.

[David Hess]

If the LSK189 from Linear Systems is not suitable, then I am sure they have a part which is.  Or a dual part like the LSK389 could be used in parallel.  Interfet should have something suitable.

[Kleinstein]

Usually the choice of JFETs from a more local supplier is relatively limited and especially TO92 parts are often NOS or similar - so it really depends on the source where you get the parts. One may still get the 2SK117 (at least in Germany). Otherwise check the usually limited list that is available.
Modern JFETs usually come in SOT23 or similar.

[MasterT]

My thoughts:
1. input cap 0.01u has impedance 15.9k at 1kHz.
2. input bias res. 47 meg, certainly is not noise free part.
Q: where noise generated by res. would flow?
This is definitely not "low noise" design, but rather exactly opposite.

[Kleinstein]

The 47 M resistor only contributes significant to noise at frequencies below the lower cross over, which is at some 0.3 Hz.
At higher frequencies the noise from the resistor is shorted by the source (unless that is higher than 47 M). One would hear the noise with an open input.
The capacitor impedance does not mean it would produce the corresponding amount of noise. The capacitor does produce a little noise but this is only from the loss part, which is something like 0.1 - 1 % of the impedance.
So the design can be low noise, but I would not expect it to be low distortion.

[MasterT]

...
At higher frequencies the noise from the resistor is shorted by the source (unless that is higher than 47 M). One would hear the noise with an open input.

How come it would be shorted, capacitor is essentially 15k resistor in between,  so doesn't matter how low output impedance of the signal source,  amplifier sees noise out of 47 meg loaded by 15k (at 1 kHz, input shorted to ground for simplicity).

[Kleinstein]

47 M loaded with the 15 K impedance is only 1/3000 the noise of the 47 M resistor. This is considerably less than the FET noise. As the FET has 1/f noise this is also true for the lower frequencies.

[MasterT]

According to this page https://www.daycounter.com/Calculators/Thermal-Noise-Calculator.phtml
at 25C 47meg res. has 27.8118 uV just for 1k bandwidth. Even it goes down by 3000, 9nV is still pretty high value