My first project for Jeff’s research was a two-channel position transducer. The purpose was to convert the side-to-side movements of a locust’s abdomen into an electrical signal, which could be recorded and analyzed later. Alternatively, two independant channels could be used either to record two axes of movement of a single wing, or to record one axis of each of two wings. During the experiment, the locust is suspended from a rigid tether, to which it is glued with a blob of wax. A wind-tunnel is used to trick the locust into thinking it is actually flying.
The design was based on a methods paper [Sandeman, D. C. (1968). A sensitive position measuring device for biological systems. Comp. Biochem. Physiol. 24, 635-638]. His design was for a single channel. I trivially extended the design for two channels, and updated it with some more modern technology (an integrated op-amp chip instead of vacuum tubes).
I’m sure I wrote a user’s manual for this thing at the time, but I can’t find it now.
The theory of operation is based on a radio technique called a synchronous detector. The detector part of a radio receiver is responsible for demoduating the received signal back into the original analog signal. An FM (frequency-modulation) detector must convert the frequency of the received signal into a voltage. An AM (amplitude-modulation) detector must convert the amplitude (ie, strength) of the received signal into a voltage There are many ways to accomplish this conversion..
The position transducer generates a low-frequency radio signal on the terminal labelled “Wand”. This signal is fed into a very thin (thinner than a hair) piece of wire (called magnet wire, because it’s usually used to wind coils), which is secured along the length of the locust wing or abdomen with droplets of glue or wax. The magnet wire is so light and flexible that it does not significantly affect the movement.
An input channel of the transducer consists of two inputs, labelled “+” and “-“. Each of these is connected to an antenna (basically just another piece of wire). The two antennae are mounted in parallel, wiith the locust abdomen or wing in between them. The orientation of the antennae relative to the wand determines which axis of motion will be measured. If the two antennae are positioned vertically on either side of the abdomen, then the device will measure the left/right deflection (ruddering motion) of the abdomen. If the two antennae are above and below the wing, then the channel will measure the vertical tilt. If they are positioned in front and behind, then it will measure the forward/backward sweep angle of the wing.
Jeff made his antennae by simply stripping a coax cable, leaving a few inches of the centre conductor exposed. The other end of the coax cable plugged directly onto the BNC jacks on the front-panel of the transducer. The antennae were fixed to a plexiglass frame, the locust tethered inside.
The signal transmitted by the “wand” will be received by the two antennae, with strength (amplitude) proportional to the proximity of the wand to each antenna. The signal picked up by the antennae will be very small, because our transmitter is not strong. And the source impedance of the antennae will be very high. So, our + and – inputs must have very, very high input impedance. We will also pick up truckloads of other signals too. Lots of 60Hz hum from the AC power and lighting in the room can be expected. Our two antennae are relatively close together, and in parallel, so most noise that is picked up will be common to both, and can be filtered out using a differential amplifier with lots of common-mode rejection.
A differential-amplifier with lots of common-mode rejection, and enormous input impedance… this says “op-amp” in letters 100 feet high. For the ultimate in common-mode rejection and high input-impedance, it also says “instrumentation amplifier”. An instrumentation amplifier is a circuit using three matched op-amps to achieve huge amounts of common-mode rejection and input-impedance. They usually don’t have all that much gain, though.
Instrumentation amplifiers are used extensively to process weak signals from high-impedance sources that are likely to pick up a lot of noise. Familiar applications include electroencephalography (EEG) and electrocardiography (ECG).
It is possible to build your own out of op-amp chips, but you’re much better off buying one in a single package. I used the AD620 from Analog Devices. The same chip would turn up again in my 16-channel EMG amplifier and the Myoelectric Locust.
Each antenna input is connected to one input of the instrumentation amplifier. The output is a copy of the original signal, but it’s amplitude is proportional to the wing position. If the wing is closer to the “+” antenna, than the amplitude will be strongly positive. If the wing is close to the ‘-‘ antenna, then the amplitude will be strongly negative. With the wing centred between the antennae, the amplitude will be approximately zero.
Converting this varying amplitude into a DC voltage is done by an AM detector. This design uses a synchronous detector. A synchronous detector uses a clock that is sychronized to the carrier frequency of the received signal. On each pulse of the clock, it samples the instantaneous voltage of the input signal. Since the clock is at the same frequency as the input, these samples will always be taken at the same point in the cycle of the received waveform. Ideally, this point should be the peak, to generate the largest output voltage. It doesn’t matter if the samples lead or lag the peak a little bit, as long as they always fall at the same point.
Normally, generating this synchronous clock is tricky, requiring a phase-locked loop. But in this case, the transmitter and reciever are both in the same box, so we have the synchronous clock already.
TODO: Find the schematic I’m sure I already made, and include it.
The output of the synchronous detector is a staircase waveform where the voltage is the amplitude of the received signal. A simple low-pass filter smoothes out the jagginess.
The output of the position transducer does not directly represent the wing angle. There is a non-linear relation between the voltage and the wing angle. Rather than try to figure out that relation mathematically, it was easier to just measure it. Jeff used a fake wing that could be set to specific angles using a protracter. He measured the voltage corresponding to a number of specific angles, and produced a reasonable angle-vs-voltage calibration curve.
TODO: Provide an image of a wing-angle recording.
Here’s the “Principle of Operation” document I included with the transducer. I attached this document, together with a schematic diagram, in an envelope inside the chassis. I believe that all electronic products should include such detailed information. A manufacturer’s right to protect their “IP” is secondary to a customer’s right to repair or even modify the products they buy. Too many electronic products today are unrepairable, basically disposable. But that’s a topic for another rant.
Here are some of the papers that have been written based on data gathered with this device.
- Hammond Manufacturing, makers of fine electronic project boxes and transformers. And they’re located just down the road in Guelph, Ontario.
- Analog Devices, makers of fine analog ICs such as the AD620 Instrumentation Amplifier used in this project.
- Here’s an interesting article I found on the subject of synchronous detectors. Actually, this site has quite a lot of good stuff on it.