Hi. Please forgive my total ignorance here. My daughter is working on a
project and I'm trying to put her on the right track. The project is:
Build a mechanical "mirror". The mirror is composed of maybe 1000 small
square or rectangular pieces of something. Each of these pieces is a pixel
in the image that is produced on this mirror. Each piece must be controlled
independently (on or off - tilted or not - in or out - etc.) by a computer.
The computer will be receiving the incoming image from a video camera. So a
person stands in front of the camera and the mechanical mirror responds by
turning each pixel on or off (or tilting or not). So...I'm thinking that
she might start by looking at the world of servos. Any thoughts as to a
good place to check out servos that could, let's say, push a wooden tile up
and then release the tile. (Thus the tile would be able to look "appear" two
different ways.) I'm thinking that she needs a servo that can do linear
motion. Or, in other words, how can we translate the circular motion of
servos into a linear motion that could push the tiles forward or backward.
She needs to just get one tile going to test. She has limited access to
"build anything from scratch" tools, so we want to get a hold of something
that is pre-made to do the linear thing. Like a servo, a gear thingo and a
rod attached to the gear thingo. Any "getting started" help on this would
be most appreciated. Thanks, Rich the Dad.
Servos are commonly used for such tasks - like pushing an airplane flap up
and down or a wheel left and right - that's what they were designed for.
You simply use a rod as a linkage from the horn on the servo to the hinged
flap. Just look at the inside of any RC model and you will see how they
are used for the task. The servo would allow you to control the amount of
tilt on each mirror - which sounds like overkill for you application.
However, if you actually intend to do this with 1000 mirrors - it will cost
a fortune (many thousands of dollars).
A better way would be to do something like those mechanical road signs use.
Make the tile you are trying to control magnetic, and put a coil under it
to attract or repel it. The road signs mechanically lock into one of two
positions by using something like a spring so the coil only needs to be
energized one way or the other to make it flip - it doesn't have the be
held on constantly so it saves power. That allows the system to be driven
a column at a time making the drive electronics simpler. But it's slow to
flip to a new configuration. If you actually want to drive it as video
speeds, you will need a lot of power and a lot of drive electronics.
Or, you can just by the DLP chip that does the same thing on a very small
but high resolution scale and not build anything. :)
That's cool. I thought it was made out of mirror tiles at first since that
is what the first poster talked about, then I realized it was nothing but
pieces of wood being moved. What a cool idea. I'd like to have one of
those in my home.... hum.....
Hi. Thanks for the responses. The mirrors that were built by Daniel Rozin
is actually the inspiration behind the project that my daughter is working
on. You're right in that it is expensive. The team she is on has about
$20,000 to build the "mirror". They looked into a device known as a linear
actuator the other day and found them to be very expensive (when the cost of
each one is multiplied by 1000).
I wish I could come up with a way that a small charge (or whatever you would
call the electricity that moves from a computer to a chip) could be applied
to something that is stationary and would change its appearance quickly.
Video monitors. :) You could create one of his "software mirrors" that
His wooden mirror seemed to actually tilt the block at different angles to
represent the different shades of gray. That also means you need the
correct overhead lighting in a dark room for the effect to work best. But
you would need a real servo to make that work.
$20K is probably enough of a budget to do it. The cheapest servos can be
found for about $10 and in the volume you are talking you might find an
even better deal.
You will need someone with a good background in electronics and computer
programming knowledge however to design and build the circuits to allow the
computer to drive 1000 servos and to write the software to translate the
video data into commands to move the servos. It's going to consume a good
little bit of power as well.
If you want to create a device that only has two positions for each element
(light and dark), then you can just use a solenoid or simple magnetic coil
of some type. That could be put together for less money but probably
requires more electronics experience to create the driver for it. You
might even be able to make your own coils for less than you can buy them.
You can create a similar low resolution effect simply by using lights
(LEDs?) in a large grid. That's how the large outdoor video displays work.
Thanks again for your added responses. One of the fundamental goals of the
project is to produce a structure that is attractive. Another goal is to
use an approach that is "different". Different to, that is, the approaches
taken already by Daniel Rozin. Another goal is to come up with an approach
that the team can handle. The team is composed of a few undergrad students
in the computer department. They have limited access to "building things"
people, so they need to come up with something simple, elegant, and unique.
Of course asking Dads to help is fine.
On Sat, 22 Jul 2006 12:33:57 GMT, "Richard Greenberg"
If you have a few hundred computer under grads, then why don't
they make a computer controlled geek "mirror" section for
football games. They could use the flash cards to spell out
words or make images. Each could have an RF receiver with an LED
or vibrator motor that when activated would have them flip the
cards. They could scroll things like "geeks like beer too" or
have somebody roaming around with a cam taking shots of the babes
for the computer science "jumbotron". Even the jocks would be
Here's what you need:
Each of these meters has a moving-coil electromagnetic movement that
requires only a tiny amount of power to activate. Remove the plastic
cover and glue a piece of paper or aluminum foil to the indicator
needle and you're off. You'll probably have to operate them with the
meter and paper facing downward or you'll have trouble with stability
because the weight of the paper will unbalance the meter movement.
Price: I believe they're under a dollar for the small ones. Downside:
you have to order from China. Try requesting a few samples. You can
also find these things in surplus stores and on eBay.
Hi again. I spoke to my daugther today and passed on your
comments/suggestions etc. Anyway, we started talking about the idea of
doing a spinning thing. If there is a variable speed motor (or even a one
or two speed motor) attached to a spinning somethingorother with a pattern
of some sort on it, then there could be a number of "appearances" for each
of the round "pixels". Any thoughts as to what type of motor would be most
suitable to such an application? Each motor of course would have to be able
to be individually addressed by the computer. Thanks again. Rich the DAD.
Well, if you want it to rotate to different fixed positions, then again,
you need a servo. This is, if you want the computer to tell the motor to
rotate the card to the 90 deg position, or to the 48 degree position etc.
I think you could do some cool things with that by having a dark half-disk
covering a white half-disk background. Different positions would cause
different amounts of the white background to show through. There are
multiple configurations like that which could work.
But, using the modified meter technique could work as well. They are very
low power but they are in effect a motor attached to a spring. Their
position will be a function of how much power you push through them. They
are very delicate, but they might be strong enough to cause a very light
weight paper flag to move.
If, on the other hand, you are talking about something spinning constantly
at different speeds, then you can use cheap toy motors that can be found
for about a dollar each. Or I guess, you could just put a bar on the shaft
that would hit stops when it tried to spin. You could then make the
computer spin it forward or backwards to force it to hit the clockwise or
counter clockwise stop. Trying to position it between the two stops would
not be very reliable.
The complexity is that the computer can tell a motor to spin, but it can't
know how far the motor has rotated without some type of feedback device
sending a signal back to the computer telling the computer the position of
the shaft. This is what a servo has - a motor combined with gears, a pot,
and electronics that makes the motor spin until it's reached the correct
position and then it stops the motor. This allows the computer to simply
tell the servo what position it wants the motor moved to and the servo gets
the job done. A motor alone doesn't work that way.
But could a simple motor be told to spin or not to spin from a command from
a computer? I would think so. What we're thinking is that the motor would
spin and that would be one state, and stop and that would be the other
state. The design on the disk attached to the motor would determine the
appearance of the 2 states.
Or....the motor could spin spow, spin fast, not spin and then we would have
3 states of appearance. The simple spinning motor idea has the advantage (I
think) of cost and possible simpler deployment.
Or.. the motor could be variably spun, producing many subtle states of
The problem I see with the meter approach is that they would have to attach
something very light weight, like paper, to the needle. They really need to
make something more substantial in appearance I believe. It is meant to be
displayed (interacted with) eventually in a lobby.
That might be cool. As I said, you can buy small cheap motors for a dollar
or less. If you buy 1000 of them you can probably do even better. If you
want them to spin in only one direction, the electronics are cheaper and
simple as well.
Yes, that can be done as well without too much cost. I'm not sure how much
the eye will be able to detect different spinning rates however. You will
just have to experiment.
Yeah, with the meters, you would have to be careful transporting it to make
sure you didn't bend or damage the needles. Small motors would be far more
I like this idea a lot. Use centrifical force to move "something"
farther as the motor spins faster, and a spring to retract the
"something" as the motor slows down. The "something" could open
and close around a ball possibly containing the motor, it could open
a "fan" like item like a peacocks tail feathers, it could make something
larger or smaller, or many other types of changes in shape and color.
You might need to feed the very small computer signal to a "transistor"
for each motor to control the higher current for each motor, but the
cost is almost negligable. I quoted "transistor" because many
different semiconductor devices could be used, not just transistors.
Rich the Dad,
One thing to consider (if this is the "right track" to lead the
daughter...;), is that spinning objects have angular momentum - which may
not be an issue when rendering still images. With video, there should be a
minimum speed and this should be factored into the mechanical design to
minimise inertia and into electronics - spinning down fast is a little
harder than spinning up fast.
My vote would be to use servos (like other posters mentioned) which cost
under $10/ea. Forget linkages. Keep the output shaft stationary and use the
body of the servo to mount your aesthetically pleasing material of choice.
Keep the modules small and self contained like Legos. Use a controller like
the one offered by Lynxmotion - which controls 32 at a time (and allows for
synchronized group moves..) and costs ~$40/ea viz a little over a dollar per
Hope you have a fun and memorable time.
I'd like to suggest a completely different way of doing it.
Drill a hole through each tile horizontally, so it can spin
on a wire axle. Cut the tiles fairly thick, and drill a
shallow 3mm hole through the axis from the front to the
back. Mount a small rare earth magnet, like a 3mm cylinder
and 5mm long, in both front and back of each tile. Then you
can mount a coil behind each tile, the kind of coil that
would be used in reed relays.
Then a +ve pulse will spin the tile to face forward, and
a -ve pulse will make it face back. The coils get wired
between a horizontal and a vertical drive wire, so that
you can do row/column multiplexing. Then all you need is
a half-bridge for each drive wire, and you have a complete
grid where you can spin every tile individually.
Thanks for your ideas and thoughts, I'm trying to comprehend it in its
entirety. I understand that you're suggesting that a pulse will create a
magnetic force and spin the tile. Do you think that this force would/could
be enough to spin a tile?
Of course. The only question is how much current it will take. Depending
on the design, it might take a lot of current.
What he suggested is not that different in concept to how a compass works.
You run current through the coil to make the top end the N pole, or run the
current the other direction to make it the S pole. The magnet and tile
would flip one way or the other just like a compass needle would point one
direction or the other in response to the magnetic field created by the
The problem is that if you turn the current off, and the tile is still
moving from the flip, it might simply keep spinning instead of stopping.
If you keep the current on, it will hold the tile, but it might oscillate
back and forth for some period of time. How long it will take to flip, and
how long it will take to stabilize before you can turn the current off will
be a function of how much drag the tile's pivot has. To try and make this
display moving video images, you would have to be flipping the tiles very
quickly. And if you want to you the cheaper row/column multiplexing he
suggested, it means you can only flip the tiles on one row or one column at
a time. So it will take some time to flip the tiles in the entire array.
If you have to apply current and wait 1/4 second for the tile to flip and
stabilize before you can turn the current off and move on to the next row,
and you have 30 rows, then it will take 7.5 seconds to change all the tiles
just once. That's not much of "video" effect.
But, you could multiplex the coils very quickly scanning all of them in a
fraction of a second. But then each coil will only be on for 1/30th of the
time. That can work if you use enough current while it's own to produce a
strong enough field. But to move the tiles quick enough to create any sort
of workable video effect, you might have to use a lot of current with some
very high power drivers. It's all a function of how light weight you can
make the tiles. If their are actually mirror tiles, they are going to be
heavy. If you can use some other light weight material like plastic or a
thin sheet of metal then you can use less current to make it flip.
This is similar the idea was was getting at in one of my earlier posts.
It's how the electromechanical road signs work. Instead of a spinning
tile, they use a hinged tile flap. Basically, the areas you want to turn
black or yellow (as used on those signs) has a hinged flap with the hinge
in the middle of each pixel. One side of the flap is yellow, the other is
black. The background is also yellow on one side and black on the other.
When flipped one way, it's hiding the yellow, and showing all black, when
it's flipped the other way, it's hiding the black, and showing the yellow.
The "flap" is magnetic, and the coil underneath it causes it to flip one
way or the other just like the spinning tile. But because it's a flap, it
can't keep spinning. It just flips one way or the other. You could make
this flap using the sheets of magnetic material they use for refrigerator
magnets (like the magnetic business cards). With a metal board to mound
this on, the flap would just naturally stick once the coil was turned off.
This is something like how those changeable road signs work.
This design is fine for creating signs that change slowly. To create a
video-like effect, would harder. You don't get any sort of shades of grey,
you just get black/white.
Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here.
All logos and trade names are the property of their respective owners.