Output Devices

For my final project, I designed a custom PCB in KiCad to illuminate the beehive entrance for camera viewing. The board controls 7 LEDs wired in parallel, switched by a MOSFET, and controlled by a single GPIO pin from the Raspberry Pi 5.

Why Two Different LED Colors?

Each half of the hive entrance has a Raspberry Pi Camera Module 3 Wide mounted to observe bee activity. To see the bees clearly, the entrance needs lighting — but there's a catch: I need to observe the bees at night without disturbing them.

This is where bee vision comes in. Honeybees see light differently than humans. Their visual spectrum ranges from about 300nm (ultraviolet) to 650nm (orange), which means they can see UV light that we can't, but they cannot see red light. Red light has a wavelength of approximately 620–750nm, which falls entirely outside the bee's visible range. To a bee, a red LED might as well be off — they simply don't have the photoreceptors to detect it.

Bees have three types of photoreceptors: one peaking in the UV range (~340nm), one in blue (~430nm), and one in green (~535nm). Notice there's no receptor anywhere near the red end of the spectrum. This is why beekeepers have long used red headlamps when inspecting hives at night — the bees remain calm because they literally cannot see the light.

White LEDs, on the other hand, emit a broad spectrum that includes blue and green wavelengths — both of which bees can see clearly. White light would agitate the bees at night, disrupting their rest cycle and potentially triggering defensive behavior.

So the solution is two boards per entrance half:

• White LED board: Used during daytime viewing when the bees are already active and ambient light is present. The white LEDs supplement natural light for clearer camera footage.
• Red LED board: Used during nighttime viewing. The red light illuminates the entrance for the camera (which can see red light just fine) without disturbing the bees at all.

The Raspberry Pi determines whether it's day or night and activates the appropriate board when the cameras turn on. In total, I need 4 boards — 2 white and 2 red (one of each per entrance half).

Circuit Design

I designed the board in KiCad. The circuit is simple but effective:

• 7× 3mm LEDs wired in parallel, each with its own current-limiting resistor
• N-channel MOSFET on the low side to switch all 7 LEDs on/off
• 10kΩ pull-down resistor from the MOSFET gate to ground — this ensures the LEDs stay OFF when the GPIO pin is not actively driving the gate (e.g., during Pi boot-up or if the GPIO is floating). Without the pull-down, the gate could pick up stray voltage and turn the LEDs on unexpectedly.
• Gate resistor between the GPIO pin and the MOSFET gate to limit inrush current to the gate capacitance

The entire board is controlled by just 3 wires:

1. Power (VCC): 3.3V or 5V from the Pi (depending on LED forward voltage requirements)
2. Ground (GND): Common ground with the Pi
3. GPIO signal: A single GPIO pin from the Pi drives the MOSFET gate — HIGH turns the LEDs on, LOW turns them off

Power Consumption Calculations

White LED Board (×2):

• Typical 3mm white LED: forward voltage (Vf) ≈ 3.0–3.2V, forward current (If) = 20mA
• Running from 5V supply with a current-limiting resistor per LED
• Resistor value: R = (5V − 3.1V) / 20mA = 95Ω → use 100Ω resistor
• Actual current per LED: (5V − 3.1V) / 100Ω = 19mA per LED
• Power per LED: 5V × 19mA = 95mW per LED
• Per board (7 LEDs): 7 × 19mA = 133mA, 7 × 95mW = 665mW per board
• Both white boards: 266mA total, 1.33W total

Red LED Board (×2):

• Typical 3mm red LED: forward voltage (Vf) ≈ 1.8–2.0V, forward current (If) = 20mA
• Running from 3.3V supply with a current-limiting resistor per LED
• Resistor value: R = (3.3V − 1.9V) / 20mA = 70Ω → use 68Ω resistor (standard value)
• Actual current per LED: (3.3V − 1.9V) / 68Ω = 20.6mA per LED
• Power per LED: 3.3V × 20.6mA = 68mW per LED
• Per board (7 LEDs): 7 × 20.6mA = 144mA, 7 × 68mW = 476mW per board
• Both red boards: 288mA total, 0.95W total

Summary: At worst case (all 4 boards on simultaneously, which wouldn't happen in normal operation), total draw would be about 554mA / 2.28W. In practice, only 2 boards run at a time (either both white or both red), so actual draw is 133–144mA per board pair — very manageable for the Pi's GPIO and power supply.

Raspberry Pi Control Logic

The Pi 5 will determine day/night using one of these methods:

• Time-based: Use sunrise/sunset times for the hive's GPS coordinates (calculated with a Python library like astral)
• Light sensor: An ambient light sensor could provide real-time readings, but adds complexity

When the camera is activated for on-demand viewing:

• If daytime → turn on white LED boards via GPIO
• If nighttime → turn on red LED boards via GPIO
• When camera stream ends → turn off all LED boards to save power

Design Files

Download the fabrication files:

Milling the Boards — Carvera CNC

Since I need 2 boards for each half of the hive (one white LED, one red LED — 4 boards total), I reformatted the layout to fit 3 boards side by side on a single panel. I generated the G-code in Carvera's software and milled it on the Carvera CNC. I ran the job twice, giving me 6 boards total — more than enough for the 4 I need, with spares in case of soldering mistakes.

Download the G-code file for the 3-up board layout:

Testing the Board — LED On

After soldering the components, I tested the board by connecting 3.3V power, ground, and driving the MOSFET gate with a GPIO pin from the Raspberry Pi. When the GPIO goes HIGH, the MOSFET switches on and all 7 LEDs light up.

Raspberry Pi 5 as the Microcontroller

The Raspberry Pi 5 acts as the microcontroller for the output device. It uses its GPIO pins to directly control the custom LED PCB I designed — driving the MOSFET gate HIGH or LOW to switch the LEDs on and off. The Pi handles the logic (determining day vs night, responding to camera activation commands) and interfaces with the hardware through its GPIO, just like any microcontroller would. The custom LED board I designed is the output device that the Pi controls.

One of the 7 LEDs didn't light up during testing. I checked the solder joints and they were fine — it was just a bad LED. Since the LEDs are wired in parallel, the other 6 still work perfectly and provide plenty of light for the camera. I'll swap the dead one out when I do the final assembly.

Control Code — Python on Raspberry Pi 5

The LED boards are controlled from the Pi using the gpiod library (the modern replacement for RPi.GPIO on Pi 5). A single GPIO pin drives the MOSFET gate — HIGH turns the LEDs on, LOW turns them off.

# led_control.py
# Controls the custom LED PCB via MOSFET gate on GPIO 17
# Board wiring: 3.3V → LEDs → MOSFET drain, GPIO 17 → gate, GND → source

import gpiod
import time

# GPIO 17 controls the MOSFET gate for the LED board
LED_PIN = 17

# Open the GPIO chip (Pi 5 uses gpiochip4)
chip = gpiod.Chip('gpiochip4')
led_line = chip.get_line(LED_PIN)
led_line.request(consumer="led-board", type=gpiod.LINE_REQ_DIR_OUT)

try:
    # Turn LEDs ON
    led_line.set_value(1)
    print("LEDs ON")
    time.sleep(5)

    # Turn LEDs OFF
    led_line.set_value(0)
    print("LEDs OFF")

finally:
    led_line.release()
    chip.close()

How the Code Works

• GPIO 17: Connected to the MOSFET gate on the LED board. When driven HIGH (3.3V), the MOSFET conducts and completes the circuit from VCC through the LEDs to ground.
• gpiod library: The Pi 5 uses a new GPIO chip (gpiochip4) and the legacy RPi.GPIO library doesn't work. gpiod is the modern Linux GPIO interface.
• set_value(1): Drives the pin HIGH → MOSFET gate goes to 3.3V → MOSFET turns on → LEDs light up.
• set_value(0): Drives the pin LOW → MOSFET gate goes to 0V → MOSFET turns off → LEDs go dark. The 10kΩ pull-down resistor ensures the gate stays LOW when not driven.

In the final system, this same logic runs inside the hive-monitor agent — the Pi checks whether it's day or night and activates the appropriate board (white or red) when the cameras turn on.

For the week 10 group assignment, our team measured and tested output devices. Full group documentation: Charlotte Lab — Week 10 Group Assignment.

My Contributions — LED Batch Testing with Multimeter

I tested a batch of 20 through-hole LEDs (3mm, mixed colors) using a digital multimeter in diode mode to verify forward voltage (Vf) and identify any defective units before using them in a project.

Test Method

Each LED was tested individually by placing the red probe on the anode (longer leg) and the black probe on the cathode (shorter leg). In diode mode, the multimeter supplies a small test current — a healthy LED produces a forward voltage (Vf) reading and faintly illuminates. LEDs that showed OL (open loop) or abnormally low Vf were flagged as failed or marginal.

Expected Vf Ranges

LED Color	Expected Vf Range
Red / Yellow	1.8 – 2.2 V
Green	2.8 – 3.1 V
Blue / White	3.0 – 3.4 V

Results — 16 Pass, 2 Marginal, 2 Fail

#	Color	Vf (V)	Lit?	Result	Note
1	Red	1.84	Yes	Pass	Normal
2	Red	1.91	Yes	Pass	Normal
3	Red	1.88	Yes	Pass	Normal
4	Red	1.76	Dim	Marginal	Below spec
5	Red	OL	No	Fail	Open circuit — dead
6	Yellow	2.05	Yes	Pass	Normal
7–9	Yellow	1.98–2.11	Yes	Pass	Normal
10–13	Green	2.88–3.09	Yes	Pass	Normal
14–15	Blue	3.17–3.21	Yes	Pass	Normal
16	Blue	3.38	Yes	Marginal	High end of range
17–18, 20	White	3.08–3.14	Yes	Pass	Normal
19	White	0.43	No	Fail	Shorted junction

What I Learned

• Diode mode is the right tool: Supplies just enough current to forward-bias the LED without damage, and the Vf reading tells you immediately if the junction is healthy.
• 10% failure rate from bulk bags is normal: Always test before soldering — desoldering a dead LED from a finished PCB wastes time.
• Vf correlates with color: Forward voltage relates to the semiconductor bandgap energy which determines wavelength. Red ~1.9 eV (lower Vf), blue/white ~3.1 eV (higher Vf).