The Amazing Frisbee
modulated (PWM) by the acceleration in two axes. If it
senses zero-g, if that axis of the accelerometer (there are
two) is horizontal, or if the device is in free-fall, the duty
cycle of the output is 50%. The duty cycle increases by
12.5% for each "g" of acceleration, up to two in each direction. So, all the circuit does is measure the pulse lengths
on the two accelerometer axes, record those numbers, and
read them out afterwards. Simple!
Choice of Accelerometer
The best device to use is the Analog ADXL202. The
accelerometer itself is a tiny surface-mount device, but it is
conveniently sold on an evaluation board. This unit used to
be sold by Parallax (the makers of the BASIC Stamp), but
they have replaced it in their catalog with a MEMSIC unit,
which is a little smaller.
There are three big differences between these devices.
The Analog device uses a bending beam to sense acceleration and can be tuned to optimize bandwidth against
signal-to-noise (two 0.1 mF capacitors and a 120K resistor
on the board set the pulse width and sensing bandwidth
correctly for this application, the PWM period being about
1 ms), and draws only about 0.5 mA. The MEMSIC unit
uses a different sensing principle, measuring the "gravity-driven" convection of heated air. It, therefore, has an intrinsically slower response, and draws more current — about 4
mA. The evaluation board sold by Parallax has the PWM
period fixed at 10 ms — giving great accuracy, but a slow
response. For convenience, I give construction details
(Figure 1 A, B) for both kinds of accelerometer, but the
ADXL202 is better.
Most of the lift on the Frisbee comes from its top, so it
was safe to mount the circuit on the underside of the disc
where it doesn't greatly affect the airflow. I put clear plastic tape over much of the circuit after assembly to smooth
the airflow and minimize any drag effects.
I assembled my first circuit on a piece of stripboard for
sturdiness, but, since there are actually very few connections, it is easier (and lighter) to make individual wire connections to an IC socket. Complete with batteries and
switch, the circuit board version weighed around 28 g; the
bare-bones version was under 20 g (Figure 2). These
weights are minimal when compared with the 175 g weight
of the Frisbee — less than 20%.
I added an LED, just to be able to see what the program was doing. The program strobes it rapidly when it is
taking data, and goes on constantly afterwards when the
code is reading out the data. A dark LED is a sure sign that
something isn't working.
The other items in the circuit are the two cells connected in series (I gave them a few inches of wire, so that the
cells could be placed around the center of the Frisbee, balancing it) and a slide switch. Placing this near the rim of
the disc meant the circuit could be turned on just as I threw
the Frisbee. I mounted all the items on the underside of the
disc with silicone adhesive (Figure 3). It is important to
mount the accelerometer as close to the center as possible. One accelerometer axis should be along the spin axis
of the Frisbee.
One approach, if you have a BASIC Stamp breadboard, is to download the program to it, and transfer the
chip to the Frisbee set-up. However, if you want to tweak
the program, this can be tedious and tends to bend pins on
the Stamp. What I did was to make a separate cable to link
a nine-pin serial connector to pins 1-4 on the Stamp via a
small header (Figure 4.) Because the serial handling for
downloading the data from the unit after flight is easier, the
data output is on pin 5 (P0) — a two-wire header (or two pins
of a five-pin) connects pin 4 (ground) and pin 5 to a serial
conector. This connector is easily attached after the flight.
Figure 2. Arrangement of parts on the underside of the Frisbee (A) with a close-up of the
ADXL202 version, built on a small circuit board (B). The two circles are the lithium button cells.
Green chip is the BASIC Stamp, small green board is the ADXL202EB.