Please consider this schematic of an electret microphone capsule: Electret Mic Capsule Schematic (Source: Wikipedia)

While considering something similar for signal conditioning of a 20 mVpp, 3k impedance output from a piezo bender, I found that discrete JFETs seem to have declined in availability and popularity, and increased in price.

This leads to a suspicion that some other discrete component that works with low signal voltage, high impedance, bipolar signals has replaced the JFET for such purposes, where anything more than a single discrete component would be overkill. A MOSFET would require biasing beyond Vgs(th), that is my fall-back.


  • What is the modern discrete alternative to the cheap-and-ubiquitous JFET?

I have considered the ALD110900 N-channel zero threshold MOSFET, but it is not cheap, not ubiquitous, and doesn't support gate voltage dropping below source.


The attraction of the JFET in applications such as the electret mic above is that it is a 3-terminal device, no regulated Vcc is required, zero-bias input signal is fed in, input modulates resistance regulates current, and it all works!

With integrated circuits like op-amps, that simplicity of implementation is lost.


Please note that the context below is not the question, it only indicates where the thought process started.

A high impedance (3 kOhm) bare piezoelectric sensor (piezo bender) is located in an underwater, hermetically sealed hydrophone at the remote end of a 5 to 15 to meter long single-core shielded cable. The 20mV to 70 mV peak-to-peak signal is getting swamped along the line due to the location being extremely EMI-rich (industrial environment). At the operator end is a hand-held, battery operated readout device that is attached to the sensor cable when required. The signal needs reinforcement in some way, for getting it back to the hand-held unit.

Some minimal circuitry can be added at the sensor end if needed, and the handheld end circuitry can be modified as required. Modification of enclosures or cabling is not an option. Volumes are 10-20 units, so a 1ku order on parts is not an option.


  • No option for adding power supply wire to sensor enclosure.
  • Limited power: 4xAA battery operated readout device.
  • Sensor enclosure too small for AA cell or much circuitry.
  • Sensor enclosure cannot be retrieved more than once a year or so for battery replacement etc.
  • \$\begingroup\$ What is the threshold for "cheap"? JFETs on Digikey for 50 cents a pop. I could see that being a problem in high volumes, but for a run of 10-20 units? \$\endgroup\$ – dext0rb Jul 15 '13 at 21:29
  • \$\begingroup\$ @dext0rb That's not the question really - The question is, if, as it appears, JFETs are becoming less popular, what do people use these days for applications like the schematic shown? Regarding your comment: 10-20 units now, but production cost becomes a factor later if the concept flies. Target prices: Under 25 cents in singles, 10 cents in 1ku, and small-volume availability on free shipping sites (eBay, dx.com, etc) for a part that can do 20mA Ids. Digikey is off the table for small orders to India just for the $40 minimum shipping fees. \$\endgroup\$ – Anindo Ghosh Jul 15 '13 at 21:43

We have switched from JFETs to depletion mode NFETs such as a BSS169.

  • \$\begingroup\$ Terrific! That's one of the options I was examining. Any specific caveats or differences in availability / performance you can share? +1. \$\endgroup\$ – Anindo Ghosh Jul 18 '13 at 18:46
  • \$\begingroup\$ Not too much to be aware of - performance wise they overlap quite a bit. Our application was originally specified for a JFET, but for the specific parameters we need (low Rds_on, low Vgs) it is just easier to find a depletion NFET now. \$\endgroup\$ – bjkopp Jul 18 '13 at 20:14
  • \$\begingroup\$ To be clear, this refers to a depletion-mode, N-channel MOSFET, right? \$\endgroup\$ – Kaz Jul 19 '13 at 1:52
  • \$\begingroup\$ @bjkopp How do they compare with regard to noise? \$\endgroup\$ – Kaz Jul 19 '13 at 1:56
  • \$\begingroup\$ @bjkopp Also, how do they compare to the JFET's feature of "Vgs 0 Volts = Ids 12 mA" (say, for a given JFET), allowing current modulation with no extra biasing or components? \$\endgroup\$ – Anindo Ghosh Jul 19 '13 at 3:29

Given that I read the cable cannot change then I must change my answer. Maybe : -

enter image description here

The beauty of the JFET is that it is simple so the potential beauty of this circuit is that it should perform better. Ultimately gain is dependent on the 1k in the hand-held equipment but if this is a current source (maybe a JFET!) then gain can be a lot higher than 6.66. I believe that because this amp is less saturated than the JFET amp, there will be more gain as well.

To make it simpler and more akin to the JFET operation the 150R resistor can be shorted out and maybe get rid of the 10uF and short the 10k. Those last two components prevent ac feedback to the circuit and should keep the input impedance at 100k but running without them may also work.


Opamps and instrumentation amps generally.

For this application I'd recommend using an instrumentation amplifier on the receive side of the cable. This will cancel out most of the common mode noise you are picking up in the cable, and lead to much cleaner output. If there is still too much noise, then you will have to amplify before the cable, as well.

If it's still too much noise, you'll probably need to convert to differential signalling at the source, and convert back to single ended on the receive side.

  • 1
    \$\begingroup\$ Marko, so instead of "some other discrete component that ... has replaced the JFET for such purposes", the only modern alternative is an integrated device, or several? My question's focus is on that elusive discrete part, a JFET alternative. The specific application stated is just for context of where my question arose from. \$\endgroup\$ – Anindo Ghosh Jul 16 '13 at 4:10

JFETs aren't so hard to find...they'll be cheaper in large volumes through direct channels, but the electret mic application should keep them cheap for a few more years, until MEMS takes over.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.