Well folks, don't
hold your breath waiting for Acoustic Antenna: Part 2! Its
taken me 2 years to get this far already. If you are not into
computers, the following may not be to your taste, but do
give it a read.
If you have taken an interest in using Doppler
to get the airspeed and RPM of your model, then this is for
you. I know the F2A guys are using this, and really its essential
to optimise performance and even see what the other guys are
doing. Have a look at my other articles for more info.
The sound from a model airplane engine is comprised
of a set of tones known as harmonics. The tones have an apparent
higher pitch when the model is approaching you, and a lower
apparent pitch when it is going away from you. This is a large
effect, easily measurable, and was discovered and named after
Herr Doppler. By measuring the pitch change, one can determine
the speed of the model and the RPM of the engine using some
very simple equations.
You need a tape recorder to store the sound,
a computer with a sound card and Richard Horne's wonderful
free program called Spectrogram to extract the tones. That's
all, and it works a treat. You can even get lap times, acceleration
and heat times, even with more than one model up. Good for
F2C, you get speed and RPM for all 3 models.
Now after doing this for a while, I wanted
more. Surely with all that racket going on, there is more
to be had from a sound analysis. Would it not be desirable
also to find the airplane trajectory? In R/C pylon race, a
tightly flown course can win races. But if you go too far
passed the pylons, you lose out badly, but may not even know
you are doing it. Replaying the course you have flown on a
computer screen would let you know your mistakes. So
what we need is a bunch of microphones stuck up in the air
and configured in such a way as to reveal the position in
space of the model as it flies along. I call this device an
"acoustic antenna". It is going to need at least 4 microphones
grouped in a bunch to find the direction of the model in 3
dimensions, so right away we have a hardware problem. Our
previous Doppler method only needed one mike, but now we need
4! How do we get the sounds into the computer? Thought you
might ask that!
Well it turns out that, not only can you connect
a mike to the mic-in port on the sound card, but with
the right circuitry you can put 2 more into the line-in port.
Not only that, you can install a second sound card into the
computer, giving another 3 mikes! That ought to be enough!
It may also be possible to add 2 more mikes to the sound card
CD port, and even 2 more to the AUX port! So what about software?
Clearly Spectrogram won't handle this lot, you are going to
have to write your own code. The programming language BASIC
is widely accessible to those with a computer bent, and the
easiest of all languages to learn. I use a version called
Power Basic, available from www.PowerBasic.com.
You have to pay for this version, and its harder to use than
say Quick Basic, but its very fast and is a fully compiled
language. This means you need to be able to write code that
will read and control your sound card. Here is an example
of code that will read a sound value from the microphone port
of just about any sound card in any PC.
OUT &H220 + &HC, &H20
do loop until INP(&H220 + 14) and &H80
value = INP(&H220 + &HA)
There, wasn't that fun! Similar commands are
available to read the other ports and control the sound card.
Now, what do we do with the values from all those mikes we
have stuck up in the air? Well, the idea is this. Because
the mikes are separated from each other, a given sound wave
from the model will reach the different mikes at different
times. Since sound is really quite slow at 340 m/s, these
times are easily measured. In fact, the mikes really only
need to be about 6" apart, so our acoustic antenna will be
quite small, about the size of a
loaf of bread.
The quantity to be measured is not really time,
but is called phase. If 2 mikes are square on to the model,
then the sound will reach both mikes at the same time, and
the phase is zero. If the 2 mikes are pointing at the model,
then the phase is at a maximum, as the sound reaches one mike
much sooner than the other. This means we can get an indication
of the position of the model from the phase difference. With
2 more mikes arranged in a square, we can in fact find the
exact direction in which the model lies. Determining its trajectory
is then just a heartbeat away. The phase difference is obtained
by determining the Fourier transform of the signals read from
the mikes. These transforms yield the tones as seen in the
Spectrogram program, and also the phase values.
This really works. Sitting here tonight in my
computer lab I have a tone generator (built from a Jaycar
kit, price $50) in the corner of the room. I have 2 mikes,
about 6" apart, going into 2 sound cards in my 486 DX4-100.
Moving the mikes around each other, I can watch the phase
change on my computer screen.
So that is Part 1 of this development program;
the feasibility study. It all works and cost me $40 for the
extra sound card, $5 for the extra mike, $50 for the tone
generator, and nothing for the computer, as Big Norm Kirton,
the super-fit bus driver and scramble flier gave it to me!