16 - System Integration

Summary

This week the assignment was to design and document the system integration for my final project, Still Here.

Approach

For this week I want to move from separate tests and prototypes toward one complete system plan. Instead of only listing parts, I need to explain how the physical curtain, electronics, sensors, code, power and interaction logic fit together.

The goal of this page is to document the integration decisions before building the final version, so I can spot missing pieces, risky assumptions and dependencies between subsystems early. Mostly this week has shown me how it's decision time and I should really focus on getting the details clear ASAP.

Assignment

Design and document the system integration for your final project

Final project recap

Still Here is a curtain that nudges you to step away when you have been in the same space for too long. It is built around the experience of getting so absorbed in a project that taking a break gets pushed aside and self-care is neglected. Timers or notifications are often too easy to ignore, so this project turns the reminder into something physical: a doorway object you have to pass through.

Britney Spears, Robert Sebree 2000

The curtain first is calm and easy to dismiss, but over time it becomes more insistent. It shifts from subtle color changes to increasingly intense light patterns until it becomes hard to stay put. It takes the form of a 90s doorway bead curtain, using a familiar decorative object as the base for an interactive system. Passing through the curtain becomes the action that resolves the reminder.

The main system parts are:

Physical structure: a doorway curtain made from hanging strands and bead-like LED nodes
Input: sensors that detect presence, passage through the curtain and/or movement of the strands
Processing: a microcontroller that tracks state and translates sensor input into behavior
Output: addressable RGB LED strands that show calm, warning, distressed and reward states
Power: an external power supply for the LED strands and regulated power for the controller and sensors
Software: embedded code for reading sensors, controlling LEDs and managing the interaction logic

System diagram

All diagrams on this page are made by Codex

Subsystems

Physical structure

TODO: design top bar

Planned doorway size: 90x200
Number of strands: 10
Beads or LED nodes per strand: 20
Distance between LEDs: 10cm
Mounting method: top bar will be either laser or 3d printed based on the actual 90s bead curtain top bar but with spaces for pcb mounting, sensors and cable management
Cable routing: needs to be decided
Mechanical strain relief: 3d printed teeth holding the strands in place

Electronics control

TODO: decide whether a level shifter is needed

Microcontroller: Raspberry Pi Pico
Board version: Raspberry Pi Pico or Pico W, depending on availability
Number of available LED data pins: 10 pins needed, one for each strand
Sensor pins: 2 or 3 pins reserved for motion and curtain movement sensors
Communication/programming interface: USB, programmed with MicroPython
Important limitations: Pico GPIO is 3.3V while the LED strands run on 5V, so I should test whether the data signal is reliable and add a level shifter if needed.

Inputs

TODO: decide motion sensor
TODO: test magnetic field sensor

Input question	Sensor / method	Location	Status
Is someone on which sides of the curtain?	motion	top bar	final decision time
Are the strands moving?	magnetic field	top bar	needs testing

Outputs

LED type: 5V WS2812-compatible addressable RGB LEDs
Voltage: 5V
Number of strands: 10
LEDs per strand: 20
Total LED count: 200
Data line plan: one pin per strand
Expected maximum current: 12A if all 200 LEDs are on full white at 60mA each
Brightness limit in software: limit to 30-40 percent brightness for normal use

Power

LED voltage: 5V
Controller voltage: 5V
Sensor voltage: 5V
Maximum theoretical current: 200 LEDs x 60mA = 12A
Planned brightness/current limit: 30-40 percent, so around 3.6A to 4.8A for the LEDs during normal use
Selected power supply: 5V 15A minimum, 5V 20A preferred if available for extra margin
Ground connection strategy: connect LED ground, Pico ground and sensor ground together
Cable thickness / connector choice: use thicker wire for the main 5V and GND distribution, then split into lighter wires per strand

Power calculation:

Part	Amount	Current per part	Maximum current
LEDs	200	60mA	12A
Raspberry Pi Pico	1	about 100mA	0.1A
Sensors	2 or 3	estimate 50mA each	0.15A
Total theoretical maximum			about 12.25A
With margin			15A minimum

For one strand this means:

20 LEDs x 60mA = 1.2A maximum per strand

In normal use the LEDs should be much lower than this, because the software will limit brightness and the curtain will not run all LEDs at full white. Still, designing for the maximum helps prevent overheating, voltage drop and random resets.

Software architecture

The software will be running on MicroPython, either using an existing libraries or writing my own for the sensors and LEDs. One motion sensor and the LEDs code has been tested.

TODO: decide libraries and dependencies

Integration plan

Interfaces between subsystems

TODO: decide on connectors

From	To	Connection	Notes
Power supply	Power distribution	TODO	TODO
Power distribution	LED strands	TODO	TODO
Power distribution	Microcontroller	TODO	TODO
Microcontroller	LED strand x	TODO	TODO
Sensor 1	Microcontroller	TODO	TODO
Sensor 2	Microcontroller	TODO	TODO

Physical integration

TODO: design top bar
Where the control board will live: TODO:
How it will be mounted: TODO:
How wires are hidden or organized: TODO:
How the curtain remains flexible: TODO:
How the system can be opened for repair: TODO:
How the USB/programming connection remains accessible: TODO:

Risk list

Risk	Why it matters	Mitigation / test
LEDs draw too much current	Could overheat wires or crash the system	Use a 5V 15A minimum power supply, add a fuse and limit brightness in software
Voltage drop over long strands	LEDs may flicker or change color	Feed power through a thicker distribution line and test voltage at the end of the strands
Pico data signal is too low for the LEDs	The Pico outputs 3.3V logic while the LEDs run on 5V, which may make the LED data unreliable	Test one full strand from the Pico before final assembly and add a level shifter if the signal is unstable
Sensor readings are noisy	The curtain may trigger at the wrong moment or miss someone passing through	Log raw sensor values first, test both sides of the curtain and add thresholds, timing filters or averaging in software
Motion sensor cannot tell direction	The project needs to know whether someone is leaving or returning, not only that movement happened	Use two sensors or combine motion sensing with curtain movement sensing, then test real walk-through scenarios
Magnetic field sensor does not clearly detect strand movement	The sensor may not be sensitive enough or may need a very specific magnet position	Test magnet placement on one strand before designing the full top bar
Wires break from movement	Curtain strands will be touched, pulled and moved often	Add strain relief at the top bar, avoid solder joints in moving sections and test by repeatedly moving one prototype strand
Connectors become confusing during assembly	Ten similar LED strands and multiple sensors can easily be plugged into the wrong place	Label connectors, keep strand order consistent and document the pinout before final assembly
Code becomes hard to debug once assembled	Integrated systems hide errors, especially when power, sensors and LEDs fail at the same time	Test each subsystem separately first, keep serial debug output and add simple test modes for LEDs and sensors
Physical design blocks maintenance	Repairs become too difficult if the electronics are sealed inside the top bar	Make the control board and power connections accessible, and design the top bar so it can be opened without destroying it

Testing plan

Bench tests

Integration tests

Run LEDs and sensors at the same time
Test state transitions with simulated input
Test state transitions with the physical curtain
Test long-running behavior
Test recovery after reset or power loss

User / interaction tests

Does the curtain notice someone nearby?
Does it correctly detect someone passing through?
Does the escalation feel understandable?
Does the reward state feel clear?
Is any part annoying, unsafe or too fragile?

Decisions to make

Leftovers previous weeks

week 11 individual assignment
week 14 documentation
nueval feedback henk
nueval feedback take