Signal Processing

This article details the “final” preamplifier design and the process involved in its development.

Outcome

The following is the design milestone that was reached at the end of the internship. This is by no means a perfect design but it can be used as a starting point for future development.

In this design, the first stage is an impedance buffer, which is used to convert the high impedance output of the piezo transducer into a low impedance signal. The second stage is composed of an amplifier configured with a negative feedback loop that amplifies the signal and sculpts the frequency response of the whole circuit. Lastly, the design includes some filters just before the output.

Both the LM4562 and the OPA1656 work with a split rail supply, with the LM4562 working within a range of ±2.5 V to ±17 V and the OPA1656 working within a range of ±2.25 V to ±18 V. In the prototype, the amps are powered with a split rail supply derived from a virtual ground circuit based on this guide from Warren Young. To meet the operating voltage of both chips, the circuit was powered with 4 AA batteries to achieve a minimum of ±3 V supply voltage.

The process

Defining the requirements

As mentioned in the overview, the stated goal of the project was to design a low-cost, high performance microphone preamplifier to be used for soil ecoacoustics research. In practice, this goal has a few implications.

Firstly, the preamplifier needs to work with contact microphones, which are the optimal way to record sound that travels through a surface instead of traveling in the air. In fact, as seen in a paper by Robinson et al., the practice involves the use of a metal rod placed into the soil that transmits the vibrations of the soil into a contact microphone (2024, p. 3). In addition, these microphones are characterised by their high impedance output. The preamplifier design therefore needs to take this into account.

Secondly, given the intended user scenario the preamplifier needs to be as transparent as possible. Any sort of distortion or interference needs to be reduced to the minimum, even when distortion might otherwise sound pleasant in a musical sense. Misrepresenting the sound of the soil might prevent researchers from finding relevant information about the soil’s health.

Thirdly, the device needs to use parts that are relatively cheap, easy to acquire and easy to work with for whoever intends to follow the guide to build their own design. Making soil ecoacoustics accessible means that more researchers can get involved into monitoring the environment with the goal of preserving its integrity. An open-design and its low cost also means that more of these devices can be deployed to more accurately map the health of an environment.

Building the microphone

Before I even started building a circuit, I needed to get my hands on a contact microphones. While buying one was unfortunately not an option due to budget restrictions, I did have access to a few piezoelectric transducers. After following a guide from Zach Poff I successfully created my own contact microphone.

If you plan on building one yourself, make sure to not move the soldered leads excessively. The best way to do this is to support the leads with a solid structure that keeps the leads still. When I was using one of the microphones I unintentionally tugged one of the leads by pulling one of the hook up wires with excessive force. After that I spent a few minutes trying to see what had broken in the circuit before I eventually decided to check the microphone.

Prototype One (24I12_0102A)

The preamplifier circuit is the result of experimentation with two different base designs. The first experimentation was done by building this preamplifier design by Martin Nawrath on a breadboard. Martin’s design works with power from two AAA batteries, it has a built in buffer for impedance conversion and it also has a potentiometer to adjust the gain. My implementation of this circuit design includes a few small changes and a few mistakes due to my lack of experience with building circuits.

The original design uses a TLC272, which works with supply voltage of up to ±18V. When I started the project I had a few LM4562 at my disposal. Compared to the TLC272, the LM4562 requires a split rail voltage supply. The circuit was modified accordingly. As shown in the pictures below, I built the circuit on a protoboard that gave me access to a split rail supply voltage of ±12V. Another small change was using two polar capacitors in series as coupling capacitors instead of the non polar ones found in the original design. The reason for this was purely down to the fact that among the components I had at my disposal there were not non-polar capacitors with the right capacitance value. Lastly, I removed the potentiometer used for gain adjustment because I wanted to make sure I could take reliable measurements without having to worry about keeping track of the potentiometer position.

A small modification of this prototype was done to add a balanced output that I could use with an XLR cable. I did this because initially I only had access to a Zoom H6 that I could use to record the output, which was the only way I could assess the distortion amount. After some tinkering I eventually decided to build an entirely different circuit. I decided to do this because during further testing I got disappointing results. Specifically, I was either seeing a very distorted output or I didn’t get any sort of output. At this point I also figured out that the jumper cables were not the best way to test the circuit as they contributed to loose connections that worsened the performance of the circuit.

Prototype Two (24I12_0201A)

This set of prototypes is based on the RIAA schematic shown on page 30 of the LM4562 data sheet. In the first iteration I simply recreated the circuit while trying to match all capacitance and resistance values as much as possible. In later iterations I then added an impedance buffer like seen in Nawrath’s design.

While experimenting with this design I occasionally used polarised capacitors where the original schematic calls for a non-polarised one. When doing so, I made sure to place the two capacitors in series but in opposing polarity (+ - | - +). I did this to match the capacitance value of the original design while using the resources I had at my disposal. However, this probably contributed to some unwanted noise in the preamplifier output. In addition to this, I soon found out that C1 (a 33µF capacitor) contributed to a rather long startup time.

In a second iteration (24I12_0202A) replaced what I assume was a paper-in-oil capacitor for a film capacitor with the same capacitance value. By replacing this component I fixed the long startup time. This prototype

input settings	in V	in RMS	out V	out RMS	dB Gain
500 Hz 20 mV	Vpp ≈ 60 mV	15.35 mV	Vpp = 3.76 V	1.30 V	38.55 dB r o/i: 84.69
1000 Hz 20 mV	Vpp ≈ 60 mV	15.40 mV	Vpp = 2.80 V	953 mV	35.85 dB r o/i: 62.08
2000 Hz 20 mV	Vpp ≈ 63 mV	15.30 mV	Vpp = 2.04 V	692 mV	33.07 dB r o/i: 45.08
5000 Hz 20 mV	Vpp ≈ 61 mV	15.30 mV	Vpp = 1.04 V	352 mV	27.20 dB r o/i: 22.93
10 kHz 20 mV	Vpp ≈ 60 mV	15.25 mV	Vpp = 580 mV	193 mV	21.94 dB r o/i: 12.50

Prototype Three (24K21_0301A)

This version combines the buffer and the RIAA design from the previous two iterations. This series of prototypes is also designed to work with a split rail supply of ±6 V instead of ±12 V. The split rail supply is derived from a virtual ground circuit based on this guide from Warren Young.