Reverse Engineering the Message Mate
FAN
Reverse engineering
is the art of deconstructing and studying products for the purpose of
discovering their underlying technological principles. The process is related to a cost
and teardown analysis. The end
result is a collection of documents and drawings that resembles the original
manufactures engineering documentation.
Essentially you could replicate the product if you wanted to. Chinese knockoffs are a common example of
this practice. I’m going to
turn-the-tables in this article because there is much to be learned from the
design of Chinese electronic toys. The
low profit margin nature
of the products leads to an extreme attention to cost reduction. This leads to elegance in design and
manufacturing practices not typically found in traditional Western engineered
products.
You can buy
a Message Mate from a variety of online wholesalers and retailers. The example above is from one of the
The Message
Mate is a handheld battery operated fan that displays five 16 character
messages with a single 9 element LED array attached to one of the fan
blades. The message can be entered
(programmed) by the user and consists of letters, numbers, and symbols. The messages are edited using only three
pushbuttons located on the side of the handle.
It is manufactured by Dongguan PinGuan Electronic Co.,
Ltd. in
The company
holds a Chinese patent # 2731156
on the illuminated fan concept invented by Li Tao. Below is an exploded diagram from the patent
showing most of the basic elements of the design. The patent covers blinking LEDs [14] in a
pattern while the fan [3] is turning.
The electronics are located on a round printed circuit board [13]. Power is transmitted to the electronics
through the metal shaft [41] of the motor [4] and a brush [21] with a rotating
contact [15]. However, the more
elaborate functions of the programmable Message Mate are not covered by this
patent.
The product
produces the message by rapidly turning on and off 9 LEDs while the fan
rotates. You can see the flexible
circuit board on which the LEDs are mounted in the photograph below. The fan rotates very quickly and the
phenomenon of persistence
of vision makes you see the message as though it was floating in mid
air. One of the biggest engineering
challenges of the device is communicating with the LEDs while they are
spinning. I’ll get to that detail later.
Close
examination of a single character reveals that it is 9 pixels high by 7 pixels
wide with a 1 pixel space between characters.
The matrix below shows how the ampersand (@) is produced. The 9 pixel height is surprising since you
would expect a nice microprocessor friendly power of two value like 8. This leads me to believe that there must be
some specialized circuitry employed to actually send out the 9 bit output
values.
I measured
the fan speed to get a feel for the timing requirements of the
electronics. A simple light detector can
be built with a Cadmium
Sulfide (CdS) photoresistor, 10K ohm resistor, and 9V battery. The resistance of the photoresistor drops
with an increase in light and can be used to detect the light emitted by the
LEDs. The schematic of the circuit is
shown below along with the connection to an oscilloscope. A single solid character message was programmed
into the Message Mate and then the photoresistor was held close to the spinning
fan where the character was being displayed to pick up the light from it. This resulted in a waveform on the
oscilloscope from which the time of rotation could be determined. With new batteries the speed was measured at
about 55 revolutions per second or about 3,000 RPM. This is a very reasonable speed for a small
DC motor with a light load. That works
out to 18ms per revolution, 1.1ms per character, or 142us per pixel. That is a pixel rate of about 7KHz which is
easily in the range of most microprocessors.
Another
important fact about the timing can be observed by simply watching the
display. The electronics must “know” the
location of the fan blade relative to the handle. For example, while editing a message, the
location of a character remains fairly static.
If the display was not synchronized, there would be some drift or
movement over time. So, as we
disassemble the device, we should find some kind position sensor.
Let’s start
the disassembly process by taking the back of the unit off. It held together with three self-tapping
Philips screws. You can see the backend
of the DC motor on the right and the programming buttons to the top. The power switch is grey and to the bottom of
the photograph shown below. The body of
the handle is made from polystyrene
plastic. The battery terminals were
first fitted into slots and then anchored by simply melting the plastic to hold
them in place. It is a little surprising
that the length of the interconnect wires is excessive, but it might have made
the assembly faster.
We are also
able to see how the battery gets connected to the rotating shaft of the
motor. A copper brush is soldered to the
metal top of the motor and has a sliding contact with the motor shaft as seen
in the photograph below. This brush,
through the metal body of the motor, is eventually connected to the positive
terminal of the battery. This style
connection has a fairly limited life expectancy, but the product is only
realistically used for a few hours until the novelty wears off anyway. Generally, planned obsolescence
like this is actually a sign of quality in that the customer doesn’t pay for
more life than they actually needed.
The plastic
nose cone of the fan is easily pried off and the clear polyvinyl chloride fan
blades removed to reveal the flexible circuit board with the array of
LEDs. The LEDs are directly wire bonded to the circuit
board and then some clear epoxy added to protect the connections. In addition to being the least expensive
method, it also reduces the weight of the assembly. If the discrete LEDs had been purchased in
surface mount packages and then soldered, it would have not only been more
expensive, but also have been considerably heavier. An electrolytic capacitor can be seen through
the oval hole on the left.
With the
cover removed, the control board with the electronic capacitor can be seen in
detail. You can see the shaft of the
motor electrically connected at the center of the board. Although it looks like the electrolytic
capacitor is glued to a blob of black epoxy, that epoxy is actually covering
the integrated circuit that controls the Message Mate. It was also wire bonded to the circuit board
to reduce cost and weight. Additionally,
there are two chip resistors (22 and 51 ohms) and two capacitors (both 0.01uF)
on the board. I don’t have time to
determine their function, but I suspect they provide some short time
delays. The implied RC time constants
are around 0.5 and 0.22us for 2 and 4.5MHz.
Maybe they are parts to generate internal clocks.
From the
side we can make out some of the moving contacts that pass signals and power to
the control board. The brush on the right is soldered on the underside of the
control board and sweeps around the outer part of the lower board. The brush on the left is soldered on the lower
board and sweeps around the inner part of the bottom of the control board. Additionally, there is a spring that is
mounted to the lower board and presses against the bottom of the control board
making a forth connection. Remember, the
shaft is also a connection.
After
pulling the control board off the motor shaft, you can better see the spring
and the brush soldered to the lower board.
Both of these parts are eventually connected to the negative terminal of
the battery. On the bottom of the
control board you can see that the spring continuously presses against a large
conductive region while the brush must periodically pass over a small separate
conductive area. This must be the way
the control board senses the position of the handle.
The brush
on the bottom of the control board is sliding against the conductive ring on
the lower board which is connected to the programming button circuit board in
the handle. The button board has 4
resistors and one capacitor. The circuit
is a simple voltage divider designed so that pressing each button creates a
unique voltage. This voltage is how the
integrated circuit on the control board knows which, if any, of the buttons is
being pressed. No buttons pressed = 0V
while Button 1 = 1.1V, Button 2 = 2.3V, and Button 3 = 3.6V.
The
schematic below at least characterizes the circuitry of the Message Mate. Most of the mystery of the device should be
gone now. As promised, the design is
quite simple and economical. I’d
question whether the timing and spring contacts could be combined to simplify
it further, but there may be other power tradeoffs that make this slight
redundancy necessary.
