Updated 1st December 2019
Updated 10 May 2018
Updated 1 June 2018
December 2019. Time moves on and the PF5102 has been discontinued by On-Semiconductor and is nearly unobtainium. It looks like the J111/J112/J113 are the only TO-92 JFETS suitable for our purposes. We are currently sampling the J113. Some Piezo Preamps will be shipped with J113 JFETs. We are considering making a last-call purchase of PF5102s, but there seems to be little to choose between the two. With the J113, we adjust the value of R10 to 470 Ohms to give a bias current of about 2mA (for a J113 with Idss about 10mA).
May 2018. After two and a half years and thousands of Phantom Piezo Preamps sold, we finally ran out of 2SK596S-C JFETs (which have been discontinued for some time by On-Semi). Going forward, the best option is the Fairchild PF5102 (now also an On-Semi product). We have slightly redesigned the Phantom Piezo Preamp PCB to allow for the use of the ‘5102 (which has a different pinout to the ‘596) and also we have incorporated a pair of zener diodes across the input to protect the circuit (and your loudspeakers) from large voltage spikes (which will be generated by a piezo transducer if you give it a significant physical shock). As a result, the redesigned PCB is 1mm longer than the old one.
Here is the revised schematic:
Note that R10 sets the bias current in Q1 and Q2. You should aim for about 2mA current flowing in R10. If you use Q3 with a random Idss, you will need to adjust R10 so that the voltage across it divided by its resistance is about 2mA. In the photograph above, R10 is the resistor right above where it says “V2”.
Note that a small number of modules have been shipped with a higher value for R10 (4k7). We have identified that some preamps may have slightly elevated distortion and somewhat lower gain than the nominal 12dB as a result of the value of R10 being a little high for the Idss value of Q3 installed. Thanks to Luc for bringing this to my attention. In almost all cases, this effect will be inaudible and those of you affected will actually benefit from increased headroom. However, if you want to restore the missing gain, please reduce the value of R10. We suggest that you do not try and replace R10 unless you are expert at soldering and we recommend that you solder a 220 Ohm resistor across R10 (i.e. in parallel) on the top side or the underside of the PCB. All preamps in future will ship with the 220 Ohm resistor for R10.
Luc also mentioned that the schematic is inverting as it is shown. This is true. If you want the preamp to be non-inverting, you need to connect pin 3 HOT and pin 2 COLD at the input.
Finally, I ordered some 2SK170 JFETs from China. I built some preamps with 2SK170 JFETs and some with 2N5457 JFETs and compared them with preamps made with PF5102 JFETs. Somewhat unsurprisingly, there is very little to choose between them. In fact, variations between different examples of preamps with the same type of JFET were more significant than the overall differences between the different types of JFET. When I get some more time, I will do a full roundup and post the screenshots.
The new preamps are available in the shop here and a version with the XLR not soldered is also available for those who want to hard-wire the output.