Addie Pinney · 3rd grade science fair · Due Feb 1, 2027
MS-2000
The Mosquito Shooter. A laser turret that learns what a mosquito looks like, follows it as it swings on a fishing line, and tags it. The mosquito lights up and buzzes when it is hit, and every hit is saved as data.
1: Brainstorm ✓2: Requirements ✓3: Pick a design ✓4: Lock it in ✓5: Parts and order6: Build7: Practice8: Experiments9: Board
The turret must do all of these. Every design choice gets checked against this list. Agreed Sep 15.
Its name is the MS-2000, the Mosquito Shooter.The name goes on the turret, the board, and every data sheet.
It shoots a laser.Real mosquito turrets use lasers, so ours does too.
It makes sound effects.A laser "pew" or a cannon "boom" every time it fires, picked with a button. A beep when it locks on. On a hit, the wand plays a splat and the turret plays a victory sound. A small speaker with a volume knob makes it loud enough for a noisy gym.
It learns a fake mosquito and finds it by itself.Addie trains the camera on her mosquito by holding the learn button and turning the mosquito so the camera sees every side. After that the turret aims with no hands on the controls.
The mosquito hangs from a fishing line and can be held and moved by a person.A "mosquito wand": a short rod, a thin line, a mosquito on the end. The turret has to follow it up, down, left, and right.
A hit is proven, not guessed.A light sensor inside the mosquito feels the laser. The mosquito's eyes flash, the wand buzzes and shows an X, and the turret adds a point. Every hit is counted by the machine, which is what the experiment needs.
It records its own data.The wand micro:bit saves every run (test, setting, run number, hits, time to first hit) in its memory. Plug it into a computer and the data opens as a table and a graph, ready for a spreadsheet.
It fits a 3 foot fair table.The laser is lined up for 3 feet, every test runs at 3 feet, and at the fair the mosquito hangs 2 to 3 feet from the camera.
The test can be repeated the same way every time.For the experiment, the wand clips onto a hook so the mosquito swings like a pendulum from a marked starting angle.
Class 2 laser, under 1 milliwatt, with the maker's label on it.No stronger laser, even though a stronger one is what it would take to kill a real mosquito. See the next section for the math.
Mounted, never handheld, and the beam always ends on the black backdrop.The mosquito swings in front of a black foam board backdrop. The turret's tilt has a printed stop, and the code limits how far it can turn left, right, and up, so it cannot aim past the backdrop. The person holding the wand stands to the side, behind the turret line.
The laser only turns on when it is locked on the mosquito, for a short flash, and only with the arm switch on.
Everything sits in 3D printed housings.Turret base, pan and tilt head, laser and camera mount, the mosquito's sensor pod, wand handle, backdrop feet and clip, pendulum pivot, and cable clips.
The best moment happens in the first 30 seconds of the demo.
Addie does the training, coding, mosquitoes, and testing.MakeCode blocks. Dad does the printing and wiring.
Budget is flexible.Final parts list: about $224 from four stores, about $254 with AAA batteries and craft supplies.
Can a laser kill a mosquito?
Yes, but not a safe one. Scientists measured exactly how much laser energy it takes, and the answer is a great fact for the board.
7 J/cm²
Energy per spot that killed 90% of mosquitoes (blue or green laser, 1/40 of a second flash).
~14 watts
Laser power that takes, for a 2.5 mm spot in that 1/40 of a second.
0.001 watts
Our Class 2 laser.
14,000×
How much weaker ours is than a mosquito-killing flash.
Why we don't use a stronger laser
A 14 watt laser is Class 4, the most dangerous class. It can blind an eye instantly, even from a reflection off a shiny surface, and it can start fires.
Science fair rules (ISEF) allow only Class 1, 2, 3A, and 3R. Class 3B and Class 4 are not on the list.
Going up to the strongest allowed class (5 mW) still leaves us about 2,800 times too weak. It adds risk and does not kill anything.
Holding our 1 mW beam on one spot does not add up either: the scientists found longer, weaker flashes kill less, because the heat leaks away.
How the real Photonic Fence stays safe
It is built by grown-up laser scientists and works 30 meters away.
It will not fire if a person or animal is in the zone.
The newer systems use a laser color that is safer for the retina.
Addie's turret does the same finding, tracking, and hit-proving as the real one. The only thing it leaves out is the power, on purpose. That choice goes on the board as part of the science.
Math: 7.1 J/cm² × area of a 2.5 mm spot (0.049 cm²) = 0.35 J. Delivered in 0.025 s = 14 W. 14 W / 0.001 W = 14,000. Source: Keller et al., Scientific Reports, 2020.
The problem
Houston has a lot of mosquitoes. They bite, they itch, and some carry diseases. Scientists have built machines that shoot them out of the air with lasers.
Under 0.1 s
How fast the Photonic Fence finds a mosquito and fires.
Wingbeats
It tells mosquitoes from other bugs by how fast their wings flap, plus size and flight.
1 to 1.5 mph
How fast a mosquito flies. Slower than you walk.
Only girls
Only female mosquitoes bite. Males drink flower nectar.
35 meters
How far away a mosquito can smell the carbon dioxide you breathe out.
174
Kinds of mosquitoes in the United States. About 2,700 in the world.
Pick a design
Picked Sep 15: Design A, Pan and Tilt Camera Turret. B stays the fallback (lock the tilt). C is out.
Addie's pick
DESIGN A
Pan and Tilt Camera Turret
Turns left, right, up, and down to follow the mosquito wherever the wand moves it. The camera and laser point the same way, so when the mosquito is in the middle of the camera, the laser is on it.
At the fair
A judge swings the mosquito. The turret chases it, flashes, and the mosquito lights up.
The science
What makes it miss: speed or target size?
How hard
Medium-hard
Cost
About $254 (final list)
DESIGN B
Pan-Only Camera Turret
Only turns left and right. The mosquito wand has to keep the mosquito at one height, marked by a line on the backdrop.
At the fair
Simpler and very reliable, but judges must keep the mosquito at the line.
The science
Same speed test.
How hard
Medium
Cost
About $156
DESIGN C
Sensor Turret
No camera. Two distance sensors say "more left" or "more right". Cannot be trained on the mosquito.
At the fair
Works, but misses more, and it fails the "learns the mosquito" requirement.
The science
How close does the mosquito need to be?
How hard
Medium
Cost
About $131
Addie's pick: A, Pan and Tilt because
A and B use the same parts except for one servo, so we can build A and fall back to B by locking the tilt if it gives us trouble.
How Design A works
Side view. The mosquito swings in front of a black foam board backdrop. The beam can only end on the backdrop.
Train. Addie points the camera at the mosquito, holds the learn button, and slowly turns the mosquito so the camera learns it from every side and distance. Then she saves it.
Track. The camera reports where the mosquito is on its screen (320 by 240 dots). Too far left, the pan servo turns left. Too high, the tilt servo tips up, but never past the printed stop.
Lock on. When the mosquito is inside a small box in the middle of the screen, the micro:bit shows a target and beeps.
Fire. If the arm switch is on, the laser flashes for a quarter of a second and the speaker plays the firing sound: laser "pew" or cannon "boom", whichever is picked by touching the gold logo on the wand.
Prove it. The laser dot lands on the white pod sewn to the mosquito, and the whole pod glows. The light sensor inside jumps. The wand micro:bit flashes the mosquito's red eyes, plays a splat, shows an X, and radios "HIT". The turret adds a point and plays a victory sound.
Addie's jobs
Sew the sensor pod onto the stuffed mosquito
Train the camera on the mosquito
Code the turret: track, lock, fire, score
Design the sound effects in MakeCode's sound editor: the laser pew, the cannon boom, the lock-on beep, the hit splat
Code the wand: feel the hit, flash, buzz, send "HIT"
Name the turret and decorate the backdrop
Run every test, write down the numbers, make the graph
Dad's jobs
Print every housing (list below)
Wire the laser, servos, sensor, and eye LEDs
Line up the laser with the center of the camera
Check every safety requirement before the laser is switched on
Help with the tricky code
What we 3D print
Every part of the equipment sits in something printed.
Print
What it holds
Notes
Turret base
micro:bit on the Xia mi board, USB-C power bank in a cradle, side power switch, USB-C charge port, laser arm switch, speaker
Window so the micro:bit lights show, speaker grille on the back, hole for the volume knob. Charge port and switch window on the right side. Heavy infill so it does not tip.
Pan and tilt head
Two servos, the tilt axle
Tilt stop built in: the head physically cannot point above the backdrop.
Sensor mount
HuskyLens on top, laser underneath
Laser in an adjustable clamp so it can be lined up with the camera's center.
Servo covers
Servo gears and horns
Covers pinch points (fair rule).
Sensor pod (P6)
Light sensor in the cap; two red LED eyes on the outside
White PLA cup with a 40 mm face that glows wherever the dot lands. Sewn onto the stuffed mosquito. Line tab on top.
Wand handle
Second micro:bit and a 2xAAA pack
Rod socket on one end. The micro:bit's speaker does the buzz.
Backdrop feet, clip, and pendulum pivot (P7)
Black foam board and a 1/4 inch dowel
Pivot holds the line for repeatable swings, with notches for 10, 20, and 30 degrees.
Cable clips (P8)
Fishing line and the sensor ribbon
Keeps the wires along the line and the rod.
Experiments and data
The build answers the engineering goal. The experiment answers the science question. Two tests, 30 runs, all recorded by the machine itself, done over winter break.
Engineering goal: Can I build a robot that finds, tracks, and hits a flying mosquito by itself?
Science question: What makes the MS-2000 miss more: a faster mosquito or a smaller one?
Hits are how accuracy is measured. Each test changes one thing and keeps the rest the same, so the test that drops the hits the most is the biggest reason the MS-2000 misses.
Hypothesis
If the mosquito is (faster or smaller), then the MS-2000 will get hits, because . One for each test.
I think will make it miss the most, because .
Variables
Change (one per test): swing angle (farther back means faster) or the size of the target hole
Measure: hits in 30 seconds, and time to the first hit
Keep the same: same turret, mosquito, line length, room lights, camera training, a charged power bank in the turret, and fresh batteries in the wand. Every run is 3 feet from the camera, the distance the laser is lined up for. When one thing changes, the other stays at 10° or the 40 mm hole.
Test
Question
What changes
Runs
1. Speed
Does a faster mosquito get hit less?
Swing angle: 10°, 20°, 30° (3 ft, 40 mm)
3 × 5 = 15
2. Target size
Does a smaller target get hit less?
Black paper cover over the pod face with a 40, 20, or 10 mm hole. 10 mm is about the size of a real mosquito. (10°, 3 ft)
3 × 5 = 15
How one run works
On the wand, pick the test and setting with button A, and the run number with button B.
Hang the wand on the hook, pull the mosquito to the angle, let go.
Press start. The MS-2000 hunts for 30 seconds by itself.
The wand beeps at the end and saves one row of data.
Write the hits on the paper data sheet too, as a backup.
What gets saved for every run
Test (speed or size)
Setting (for example 20°)
Run number (1 to 5)
Hits in 30 seconds
Time to the first hit
Room light level, to check the room stayed the same
The micro:bit V2 keeps this data even when the batteries come out. Plugged into a computer, the file called MY_DATA opens as a table with graphs and can be downloaded for a spreadsheet.
Test 1: hits in 30 s
Run 1
Run 2
Run 3
Run 4
Run 5
Average
10° swing (slow)
__
__
__
__
__
__
20° swing (medium)
__
__
__
__
__
__
30° swing (fast)
__
__
__
__
__
__
Data rules
Every run counts. Bad runs stay in the data.
Five runs for every setting, never fewer.
If something breaks, write down what happened and redo that run.
Do each test on one day so the room stays the same.
Take a photo of every setup.
Turning data into answers
Average: add the 5 runs and divide by 5.
One bar graph per test, averages on the bars.
Compare: which setting got the most hits, and by how much?
Answer the question: which test dropped the hits the most? That is the biggest reason the MS-2000 misses.
Conclusion: "My hypothesis was (right / wrong) because the data showed ___."
What I would change next time.
Start this fall, no parts needed: hang a paper mosquito on a string and watch how the swing changes with the starting angle. Cut the three black paper covers (40, 20, and 10 mm holes) for the size test; all three covers are the same size outside so the camera sees the same mosquito. Addie can also write the wand code and try the datalogger blocks in MakeCode on the website.
The fair demo
The best moment has to happen in the first 30 seconds.
The table
Trifold board behind
Fits a 3 foot table: turret at one end, backdrop at the other, mosquito 2 to 3 feet from the camera, beam running sideways along the table, not toward the judges. See the table layout.
"Class 2 laser" sign, maker's label visible
Camera screen turned so judges can see the box it draws
Spare AAA batteries, a phone charger for the power bank, and a spare mosquito
What Addie says
"Scientists built a machine that shoots mosquitoes with lasers. I built a safe one. It's called the MS-2000."
"Hold this mosquito and try to get away." (hands the judge the wand, judge stands beside the turret)
The MS-2000 chases it, flash, the eyes blink, the wand buzzes. "HIT!"
"It knows it's a mosquito because I trained it." (camera screen)
"A laser that kills mosquitoes would need to be 14,000 times stronger. That's too dangerous for school."
"The MS-2000 recorded 30 runs by itself. Here's what happened when the mosquito went faster and got smaller. The thing that made it miss most was ___." (graphs)
Parts list (final)
Locked Sep 16, 2026. Every part has a primary and a backup, with links checked Sep 15 and 16. The primary order uses four stores. Changed Sep 16 by change order MOC-004: a rechargeable USB-C power bank replaced the 4xAA pack.
The turret board is now the DFRobot Xia mi: it has a 5 V HuskyLens port, 5 V ports for the servos and speaker, and a relay for the laser. The micro:Driver is the backup.
Clutch servos that slip instead of breaking if someone grabs the turret.
A separate laser arm switch that cuts power to the laser only.
Sep 16, MOC-004: a rechargeable USB-C power bank stays inside the base and charges through a port on the side, like a phone. A switch on the right wall is the main power. The bank gives a steady 5 V, so the servos move at the same speed on run 1 and run 30. One charge runs about 4 to 6 hours. The wand still uses AAA batteries.
M2 screws for every printed part; the HuskyLens keeps its own M3 kit.
Check on arrival
With both servos sweeping and the laser on, the Xia mi 5V pins read at least 4.6 V. The bank gives 5 V, the low end of the DC jack's 5 to 12 V range. If it reads low, power the board through its own USB 5 V input with a USB cable that has an in-line switch.
Measure the power bank, the in-line switch, and the charge port with calipers before printing the base.
Whether the board powers the micro:bit (a spare 2xAAA pack is on the list in case it does not).
The Class II label on the laser before it is ever switched on.
Laser safety
These are the rules the big international fair (ISEF) uses, which many school fairs copy. Our school may be stricter, so we ask the teacher before ordering.
What ISEF says
Class 1, 2, 3A, and 3R lasers are allowed "to be used responsibly".
Lasers must carry the maker's label showing their power.
Handheld lasers are not allowed.
No beams through magnifiers or telescopes.
Moving parts with pinch points must be covered.
Unsafe lasers get taken away with no warning.
What we do
Class 2, the lowest visible class, under 1 mW, labeled
Bolted into the printed head, never held
Printed tilt stop plus code limits: it cannot aim past the backdrop
Black foam backdrop behind the mosquito stops every shot; no mirrors or shiny parts near the beam
Short flashes only when locked on, only with the arm switch on
The wand holder stands beside the turret, never between the turret and the backdrop
Servo gears covered
The power bank is a sealed lithium battery: never crush, pierce, or drop it. If it gets hot, swells, or smells, a grown-up takes it out and puts it somewhere safe. Never charge it while the turret runs.
The plan
St. Rose 3rd grade: final project due Monday, February 1, 2027. Exhibition Wednesday, February 3 at 2:00 PM in the Parish Hall; winners go to the Archdiocesan fair on February 25. School due dates and the rubric are on the fair guide. Changes since the design lock, the project history, and lessons learned are on the changes page.
1
Brainstorm
Done Sep 15
Air cannon, vacuum, and light beam ideas. The family picked lasers.
2
Requirements
Done Sep 15
The list at the top, including the name: MS-2000.
3
Pick a design
Done Sep 15
Addie picked Design A, the Pan and Tilt Camera Turret.
4
Lock in the design
Done Sep 16
Design A locked. Still to do by hand, in the journal: why Addie picked A and a labeled drawing. Ask Miss Taggart about laser rules.
5
Final parts list and order
List done Sep 16; order by Sep 30
Order the primary parts from DFRobot and DigiKey using the parts list. Pick up the local items. Start the cursive journal.
6
Print and build
Oct 1 to Nov 13
Mosquito and wand first (prove a hit lights the eyes). Then the pan and tilt head. Then the backdrop and pendulum. Then the code and sound effects.
7
Camera training and practice runs
Nov 14 to Dec 11
Teach the camera, set the hit threshold, make sure the data saves. At school: question due Nov 18, research Dec 2, hypothesis Dec 9.
8
Experiments
Dec 12 to Jan 15 (due Jan 19)
Test 1 speed, Test 2 target size. 30 runs, all at 3 feet, one test per day, over winter break.
9
Data and conclusion
Due Jan 25
Averages, two bar graphs, and the conclusion.
10
Board and practice
Due Feb 1; exhibition Feb 3
Question, hypothesis, variables, materials, steps, results, conclusion, Catholic connection, sources, and the "14,000 times" fact. Practice the 30-second demo on the family.