Phantom Piezo Preamp Modules are available from the shop. Click here.
Sometimes, back of the envelope calculations do not tell the whole story and neither do simulations. This article is a follow-up to the original article Phantom Piezo Preamp and addresses some errors and misconceptions that have arisen.
The Phantom piezo preamp module (available here in the shop) has been very popular and has sold well. Occasionally though, we have had questions – both in the comments and privately – which I have addressed as best I can. However, confidence in my answers has not been as high as it should be, so I invested in a Focusrite Scarlett 2i2 and downloaded ARTA audio measurement and analysis software and set to work taking a new look at the performance of the Phantom Piezo Preamp.
Low frequency response and input impedance
Here’s a reminder of the original design schematic:
In the original article, we suggested that the 220pF coupling capacitors (C2 and C3) were too low in value to give good low-frequency coupling; we suggested an increase to 22nF along with an increase in value of R1, R2, R5 and R6 to 3M3 (along with some other changes – see the final schematic below). We suggested that the -3dB lower corner frequency improves to 22Hz with these values. Our back-of-an-envelope calculations suggested that the effective input impedance is 3M3 and the lower -3dB corner frequency should therefore be 1/(2*Pi*R*C) where R equals 3M3 and C equals 11nF (two 22nF in series) which equates to 4.4Hz. However, we cite a -3dB corner frequency of 22Hz – which we understood from the LTspice simulation. I did not originally do the calculation which gives 4.4Hz, so I didn’t notice the discrepancy.
If we plot the actual frequency response of the above design using ARTA, we get this (click for a larger image):
As you can see, the lower -3dB point is at about 8.5Hz. This suggests that neither our back-of-an-envelope assessment nor the simulation is correct. Assuming the input capacitance is 11nF, we can calculate (from the 1/(2*Pi*R*C) formula) the input resistance as 1.7 M Ohms. This suggests that we should not have included the values of R4 and R7 in our initial assessment and the input resistance is set by R5 || R9.
Note that the actual system low-frequency response you will see is a function of the phantom preamp and the characteristics of the input stage of the following microphone preamp.
Finally, we should note that the system low-frequency response depends upon the input impedance of the device the Phantom preamp is connected to.
High frequency response
The high frequency response is moderated by the Zobel network formed by C1 and R2 in the above schematic. To increase the high frequency response we would need to lower the value of C1 as follows:
Changing the input impedance
There has been some interest in increasing the input impedance of the design and – based on the original article – there is a natural tendency to increase the values of R4, R5, R7 and R9 to a higher value. However, we note above that the input impedance is set by R5 || R9 only. If we replace R5 and R9 with 10M Ohm, the input impedance will increase to 5M Ohm. An alternative solution would be to connect the piezo transducer with series resistors. You could just use one resistor, but two equal value resistors, one each connected at pins 2 and 3 on the input would be more balanced. So, if you wanted to increase the input resistance to about 10M Ohm, (say) you would add a 3.9M Ohm resistor in series with each of the piezo-transducer connections to give an input impedance of 3.9 + 3.9 + 1.7 = 9.5 M Ohm. Note that this would reduce the input sensitivity by 15dB as the added series resistors would make a potential divider with R5 and R9. Also, note that very high resistance values may exhibit significant self-noise (Johnson noise), so there is a trade-off between input impedance and noise.
Finally, be aware that if you change the value of R5/R9 (without changing R4/R7) you will change the bias point of Q1 and Q2. This will affect input headroom. E.g. if you increase R5/R9 to 10MOhm, the d.c. operating point at the gate of Q1/Q2 will increase from ~22V to ~33V.
I hope this clarifies the situation further and welcome any comments either below or by email.