Presented are both lighted costumes and animated porch displays built over the years for Halloween.
|Lighted Costumes||Animated Displays|
Have you ever seen the Spectromagic night parade at Disneyworld in Florida? One of the types of characters in the parade are butterflies and faeries: actors with large wings lit with many lights. That, I suppose, was inspiration for a lineage of costumes: lighted wings.
This year (2011), Jennifer was certainly not going out for Halloween, however she still wanted a costume since she was working at the Howell Family Pumpkin Farm. Here, she is working at the main gate in her costume which includes twinkling wings and lit "antennae"
The wings feature many channels of lighting including six channels around the perimeter (allowing "chase" effects), four channels in the center (allowing "random twinkle" effects), and two channels for the antennae. The actual chase sequence is quite complex beginning with a "chase" sequence where the perimeter light chase in a clockwise direction, then counter-clockwise, then up from the center towards the bottom, then from the bottom upwards, followed by a random twinkle of all lights on the wings (including the perimeter). The GIF does not do this justice at all!
It was found in previous years that LEDs did not give a nice "twinkle" effect like incandescent lamps do (just look at the brake lights of vehicles using LEDs versus those using incandescent lamps), however the energy savings (translated into battery life and battery weight) are enormous and so LEDs were used however each lighting channel was PWMed allowing the lights to be "ramped-up" and "ramped-down" in intensity. The PWM algorithm is accomplished is software (since no chip features a twelve-channel PWM) using a high-speed interrupt (the PIC 18F252 microcontroller was running at 2MIPS giving plenty of clock cycles to accomplish the task).
The wings , normally covered with black nylon stretched over the frames, are seen here during construction. The outer frame was constructed of thin-walled (yet strong) steel tubing bent into the required shape and MIG welded to form the wing shape. The frames were bolted to a wood base which also holds the battery pack and the required circuitry.
A close-up of the circuit (to the right) shows the basic construction on a piece of perfboard. Aside from the 18F252 microcontroller (seen here with an RJ12 plug permanently wired allowing easy upgrades), the circuitry includes two ULN2003A darlington driver chips which drive the actual LEDs (since the total current drawen by the combined channels exceeds the drive capabilities of the PIC device). Also visible are resistors used to limit current through the LEDs. LEDs are wired as strings of four in series and powered directly from the 12V battery supply (the use of the ULN2003 driver allows LEDs to be driven from a higher voltage than the 5V supply used to power the microcontroller).
The Technical Details ...
The basic controller is based on a PIC18F252 microcontroller running at 8MHz. Twelve general-purpose I/O lines are used as digital I/O to drive LEDs via two ULN2003A drivers. These drivers allow direct TTL-level drive and so no extra components are required (the I/O line connects directly to the chip input). Darlington drivers, these devices saturate easily and so are perfect for PWM usage. Each chip features seven channels and so two chips are used.
Since no chip offers twelve independent PWM channels (although a dsPIC33FJ128MC802 was considered which offers six channels), these are implemented in software. Twelve variables "PWMChx" set the duty cycle of each channel and hence the the intensity of the LEDs. An ISR is triggered by Timer0 at 16.7KHz and a counter run continually. When the counter equals zero, each channel is turned ON and when the counter equals a "PWMChx" variable, the counter is turned OFF so a value of 1 represents a very low intensity (duty cycle of 1/255) and a value of 0xFF represents DC. The rest of the code is a very simple loop which ramps-up LED intensity when an LED it turned ON (via a loop which increments the "PWMChx" values) and ramps-down the value when an LED is turned OFF. These variables are simply set in the main code, and read in the ISR (and translated to a duty cycle).
A simple piece of code, and certainly not the "cleanest" I've ever written (it's not exactly "safety critical" code and so there were no marks for neatness here :), it can be Downloaded Here. It may be assembled under MPLAB version 8.x.
Previous Versions ...
For 2007's 'Tinkerbell' costume a set of small wings was created which were made to twinkle by employing a PIC 16F876 microcontroller to generate random patterns on four sets of lamps, arranged so that the entire set runs off a 14.4 volt cordless drill battery. The photo does not do justice to the wings as it does not show the 'twinkling' effect - the effect is best with incandescent lamps as well, LEDs failed to show the gentle twinkling effect caused by the thermal inertia of the filament (solved in 2011 by using PWM to "ramp-up" and "ramp-down" light levels). The wings themselves were made from thick steel wire bent in the form of a wing and fastened to a piece of wood attached to straps from an old school backpack. Lamps were strung in sets of four across the frames and the outside of the wing covered with fabric sewn as a slip-cover: the slip-cover is pulled back on the lower right corner of a wing here to show the lamps underneath. 64 lamps were used in all. In 2008, the wings remained the same however a lithium-ion laptop battery was used to reduce weight: lithium batteries at 4.7Ah provide considerably more output than NiCads at about 2.2Ah and weight much less.
An once again Jennifer had a lit costume - this time an angel's wings with a lit halo. In 2006 we went energy efficient though - blue LED lights were used to surround wings made of thick steel wire with power supplied by a 12V rechargeable cordless drill battery. To supply the required 120 VDC for the light string, the 12V battery fed to a tiny inverter circuit previously used to power Nixie tube displays (this is a tiny inverter which runs off 12V and produces 150VDC at 10mA). Overall the wings weighed about half of the previous years' wings which featured a large lead-acid battery.
Our first set of wings from 2005. In the style of Disney's Spectromagic Parade, these wings are about 4 feet across and feature 140 bright, sequenced, lamps. Powering 140 lamps requires a large battery, in this case a 7Ah, 12V sealed lead-acid battery which runs a small inverter providing 120 volts for the light string.
The battery and inverter are mounted between the wings (which are made of a steel tubing frame) on a piece of plywood. Shoulder straps from an old knapsack serve as a mount.
These wings were incredibly bright, so much so that in the fog she could be seen on the other side of the block as a lighed haze over rooftops, but unfortunately the battery (which only lasts a few hours) was quite heavy, especially for a nine year old! Later versions employed more energy-conscious lighting (i.e. LEDs) and lighter lithium batteries.
For over twenty years we have built automated (animatronic) displays for the front porch to scare the trick and treaters at Halloween. Starting simple, with a ghost which makes a 'boo' sound when approached, these displays have become more elaborate over the years and now incorporate laser sensors, microcontrollers, and pneumatic actuators. Within the past ten years, displays have included witches who move, stirring boiling cauldrons, and skeletons which pop out of coffins. Our 2008 to 2010 displays, pictured here, featured a skeleton in an electric chair. When a kid approaches the porch the system triggers and the skeleton rises out of the chair accompanied by fog, a strobe light, and the loud sounds of an electrical arc. The front window featured a rear-projection video of a large tesla coil operating with large arcs streaming everywhere to complete the 'mad scientist' appearance (also completed by yours truly wearing a lab coat and a wig of "shocked" white hair).
The prototype for the 2008 skeleton is seen here in the workshop. The electric chair has a light rope for the "wires" connecting the skeleton to the chair. Upon triggering, fog is released by a fog machine behind the chair, illuminated by eerie green light and a strobe light.
The skeleton has taken various forms and each year it seems to get a bit more elaborate. In previous years, such as this display from 2005, the skeleton popped-out of a wooden coffin which, when triggered, is opened via a pneumatic cylinder. Once opened, and after a small delay, the skeleton suddenly pops right out, driven by a second cylinder.
The coffin itself was built from eighth-inch board with a 1-by-2 frame. Screen door closers operate as pneumatic cylinders - the adjustment screw was removed and a 1/8-27 pipe thread tapped into the end. A flexible line on compression fittings connects the cylinder to a 24V solenoid air control valve. Two cylinders and two valves are employed. When a laser trigger (described below) sets the display in motion, the first cylinder, which opens the coffin door, in activated. An RC delay circuit then provides a delay of about one second before the second valve is opened rapidly firing the skeleton out of the coffin. A commercial fog machine provides spooky effect by filling the coffin with fog while it is closed.
The cauldron and bones from a previous year's display sits on the right side of the porch.
In 2006, the display combined the witch from 2003 and the coffin from 2005. Upon triggering the sensor (visible in the lower left of the photo) the coffin opened, and following a short time delay, the witch turned to greet the trick-or-treaters! Turning the witch was accomplished by mounting her on a piece of wood with a large bolt as a pivot. A pneumatic cylinder actuates the mount.
In 2004, the display featured a boiling cauldron 'stirred' by a witch. A hand, operated by a pneumatic cylinder, comes out of the cauldron when triggered. The display was improved two years later to feature the witch on a turntable allowing it to move quickly to face oncoming trick-and-treaters. Click Here to see a movie showing the cauldron in action
Mechanical Details ...
For the witch and cauldron display, the cauldron was made from paper-mache (I used strips of paper soaked in drywall compound) and is 22 inches in diameter. It features a moving hand which comes out of the cauldron when kids cross the laser-sensor (details below) by pneumatic power. The cauldron also features a moving stirring-spoon driven by a windshield-wiper motor inside - when attached to the witch it appears that the witch is stirring the brew. Pumped full of fog from a fog machine underneath, the addition of a red spotlight and a load of plastic bones on top gives the illusion of some sinister brew of bones!
2003's display was a plastic skeleton which rises from a coffin as kids
approach the porch. The coffin itself is built from old pieces of 1/4" plywood
with a 1*2 frame inside so it can easily be dismantled. An old storm-door closer
acts as a single-acting pneumatic cylinder to open and close the door. A 3-way
solenoid air valve operates the door on demand. This was the basis for 2005's improved
An updated version, used in 2007, features a lift inside the coffin so that when triggered the coffin first opens then the skeleton 'sits up' abruptly to an upright position. Again, two pneumatic cylinders are employed - one to open the coffin and a second to prop the skeleton up - the action is seen here during prototyping in the workshop.
A fog machine fills the coffin full of spooky fog which is released when opened and a green floodlight inside the coffin draws attention to it.
The insides of the coffin reveal the skeleton (on a hinged board which lifts via a cylinder), a green floodlight, and a small fog machine. Previously, a larger fog machine was used which was mounted below the coffin and plumbed-in using ABS pipe.
The cylinder which opens the coffin is also visible in this photo (it is painted black to match the interior).
The electric chair prop for 2008 consisted of a simple wooden chair built from low-grade, rough, wood. The skeleton is supported by two pieces of 1-by-2 wood hinged in the middle allowing the skeleton to move in a more natural way when lifting from the chair. The skeleton moves first in a bent-over position, when maximum extension is approached a chair pulls the support behind the skeleton's back causing the skeleton to move to a prone position. Two actuators could have been used however a single cylinder and two hinges proved to be a simpler solution. A close-up photo of the mechanism shot from the rear shows the arrangement of the cylinder, two support arms, and the chain which pulls-back on the arm to erect the skeleton. Wire ties hold the skeleton to the arms.
This shot of the rear of the chair was taken before it is covered with black cloth to hide the mechanism. A small fog machine was mounted on a bracket immediately behind the chair - when actuated it releases a programmable length blast of fog (the length set by a digital timer module). Beside the fog machine is a green floodlight which, when scattered by the fog, casts an eerie green glow all around and on the top is a strobe light (flashing associated with electricity). Also seen in the bottom of the photo is the solenoid air valve, timers for the fog machine, two power supplies, the electronics package, and set of computer speakers. The speakers are fed continuously from a laptop playing an audio file of "electricity" noises (loud arcing from a Tesla coil) and are simply connected to the same AC supply as the spotlight: when the display is activated the spotlight, strobe, and sound are all activated.
An old screen door closer, used as a pneumatic cylinders, is seen in the left photo - in this case the cylinder lifts a 1-by-2 inside the coffin (or on the chair) onto which the skeleton is mounted. The middle photo shows the two solenoid air control valves which direct compressed air into each cylinder (some displays used two cylinders, like the "skeleton popping out of the coffin" display while others like the "electrocuted skeleton" used only one). Both valves are connected to a simple manifold and a quick-connect fitting for connection to the compressed-air hose. Finally, the electrical controls in the right photo are seen here on the workbench. The laser sensor (held down here with a roll of duct tape during testing) sits atop one of two power supplies for the system. Beside that are two electronic timers used to control the fog machine (when used with the coffin, one injects fog into the coffin at periodic intervals and the second injects a burst of fog when the coffin opens. When used with the chair, only a burst is required when the display is activated). Finally, two relays on a board take the 'beam tripped' signal from the sensor and drive all DC (control solenoids) and AC (floodlight) loads. The electronics were designed to be generic enough to be used with various displays. The first relay (a 4PDT) activates immediately when the sensor is tripped while the second relay closes about one second later. Usually, the first relay activates fog and lights (and possibly the cylinder which opens the coffin for that display) and the second relay activates the skeleton motion.
The Laser Sensor ...
A key element of the display is an accurate sensor to trigger the display when a trick-or-treater approaches the porch. We had tried everything from PIR sensors to ultrasonics but these never proved reliable (the PIR, for example is quite unpredictable in cold weather when the kids are wearing thick jackets under their costumes). The laser sensor seen here proved to be the only reliable method of triggering. In addition to providing a simple trigger, the sensor was expanded to include two beams so that it can sense direction.
The display is triggered by a sensor which uses two infrared (780 nm) laser beams
to detect which direction the 'victim' is travelling by determining which beam is broken
first. A PIC16C84 microcontroller performs the necessary logic. The sensor then
drives a 12V relay which activates the pneumatic air solenoid, a green spotlight
in the coffin illuminating the skeleton, and sound effects from an inexpensive doorbell
which screams when pressed (a commercially-available halloween prop obtained from a dollar-store
for about $2.50).
The laser diode which provides the beam was scavenged from an old handheld barcode
scanner. The laser includes a matching adjustable lens which proved quite
useful in collimating the beam. The entire laser is housed in an old
whiteboard marker along with an LM317T acting as a current regulator (The laser
diode runs well at 55mA). Along with the laser diode the scanners also provided
PIN detectors with matching dielectric filters to exclude all light except the
780nm laser's. The laser beam was collimated to cover both PIN diodes at the target distance.
An inside view of the sensor shows all optical and electronic components mounted
on perf board inside an LED clock case. The PIN photodiodes along with matching filters
are mounted behind a red bezel on the case front. The diodes are connected with 2M2 pull-up resistors to the input of two CMOS 4049 inverters before entering the PIC - by trial-and-error it was determined that the 2M2 resistance was optimal to provide a TTL-level signal (+5V when the beam was broken, 0V when the beam was complete). A PIC16C84 microcontroller runs a simple program to detect the direction of the motion based on which beam is broken first. Five DIP switches allow configuration changes such as direction (forward only or fwd and reverse, time for output pulse, retriggerable mode, etc.). Two DIP relays are triggered in response to the trigger event. To drive the entire display the output from the small DIP relay usually drives a much larger relay providing the current needed for the solenoid valves and spotlight.