Update: Please see also Phantom Piezo Preamp revisited
Update: Phantom Piezo Preamp modules are available for purchase in the shop.
A few years ago, Alex Rice put a design up on the internet for a phantom-powered JFET piezo preamp. Here’s the schematic of the original design as reported by Zach Poff:
So far so good. Q1 and Q2 are cited by Alex as a matched pair of N-channel JFETs with Idss 1-2mA and that Q3 should sink 1mA, giving a 0.5mA bias to each of Q1 and Q2. R5/R1 and R6/R2 bias their respective JFETs. R5 and R6 provide a little negative feedback and R4/C1 forms a Zobel network to reduce high-frequency ringing (oscillation) in the cable. Alex Rice claims that the final design evolved as a result of “a lot of fiddling around with circuit simulators and breadboarded prototypes.”
Now I’m no expert at analog simulation, but I have played fairly extensively with LTspiceIV and the above circuit does not appear to work. It seems obvious that the value of C2/C3 is really rather low and therefore the low-frequency response of the above circuit is not going to be very good. I simulated the design in LTspice IV like this:
Q3 is replaced by a 1mA constant current source; The input is an ac voltage source; R5, R6 and V2 represent the phantom powering circuitry in the mixer; and C3, C4 and R7 represent the early part of the input stage of the mixer. We’ve left the Zobel network out because we want to see how the circuit simulates without it. We take our output from the differential voltage OUT+~OUT- which is approximately what the high-impedance differential-input-stage of the mixer-preamp will see.
Note: Please read Phantom Piezo Preamp revisited for updates and corrections.
The simulation command is .ac dec 20 1 20Meg to give a small signal response from 1Hz to 20MHz.
The above circuit simulates with a lower -3dB point at 2.2kHz and the output is down by 41dB at 20Hz. This directly contradicts Alex Rice’s assertion that the frequency response is down by only 3dB at 20Hz. And this is as you would expect. Each side of the input signal is going to see a low pass filter formed by the input capacitor and the input resistors in parallel. If we lump all the input resistances together we get (R1+R3) || (R2+R4) to give 3MOhms a.c. input resistance and 220pF in series with 220pF to give 110pF input capacitance. Therefore, the input signal sees a high-pass filter with a corner frequency of 1/(2 x pi x R x C) = 482Hz, which is considerably higher than 20Hz.
If we increase the value of C1 and C2 to 22nF and increase R1~R4 to 3.3 MOhms, the -3dB point falls to 22Hz, which is more like what we would want.
Now, the high frequency response of this design is theoretically 3dB down at 25MHz so we definitely want to incorporate some roll off. Alex also mentions adding source resistance to Q1 and Q2 to reduce the gain and series resistors in the output lines to further reduce the likelihood of oscillation in the cable between the preamp and the mixer. This gives the following topology:
This simulates with a -3dB frequency response of 18Hz to 21kHz. Note that the capacitor in the Zobel network (C3 above) has been increased from 220pF to 680pF. If you further increase C3 to 1nF, the upper -3dB point will reduce to about 14kHz. The gain of the circuit is somewhat dependent upon the input impedance of the mixer and with R7 at 47k, gain is about 15.5dB. I simulated this design with various JFET models and they all give broadly similar results. The major effect is differences in nominal transconductance affecting the gain.
The point of the exercise is to find a use for the 2SK596S JFETs I just bought and characterised (and indeed put on sale here). The ‘596 features comparatively low Idss. Unfortunately, there doesn’t seem to be a spice model kicking around, so we can’t simulate it in the above circuit topology. However, we can characterise the Idss and figure out a suitable value for the current source:
We finally settled on and built this design on a small PCB with a Neutrik NC3MAAH-1 pcb-mount XLR connector:
Initial tests give good results. The next plan is to build a wooden stomp box and give the design a proper workout. We had plenty of PCBs made so we will be putting a kit together. Let me know if you’re interested in the mean time.
So what can this design be used for? Well, it can be built into all kinds of instruments that have a piezo pickup element, whether that is a commercial pickup or a home-brew piezo disc glued or taped onto a resonant surface. Also, you could install it into a guitar or bass (the PCB envelope without the XLR and input sockets fitted is 37mm x 29mm x 9mm high) and give your instrument a balanced active output.
Please see also Phantom Piezo Preamp revisited for updates and corrections.
As the unit must be phantom powered (and there is no battery option), the design is suitable for semi-pro or professional use where there will always be a mixing desk with phantom power available and a sound guy who’s not afraid to use it.