Skip to content
Parth Sharma
  • Work
  • LinkedIn
  • parthsharma4591@gmail.comEmail
  • 510-458-7035Call
The black two-part controller enclosure with a blue 16 by 2 LCD in its front window and angled vent slots on the right side.
EXTERNAL RETROFITSG90 servo0°35°70°Arduino UnoI²C LCD, press-fit windowBreadboard + LCD contrast potBridging-safe ventsPress-fit pegs
The SolidWorks model of the controller: a black two-part enclosure with a blue LCD window and angled side vents.The black controller mounted on a painted wall, its blue LCD reading the target temperature and humidity, a USB cable running out of the side.
The CAD (left) and the finished controller installed over the existing switch (right).
The black controller mounted on a painted wall, its blue LCD reading the target temperature and humidity, a USB cable running out of the side.
Installed over the existing switch.
The black controller mounted on a painted wall, its blue LCD reading the target temperature and humidity, a USB cable running out of the side.
The controller where the old switch lives, reading target temperature and humidity.
  • DHT22Room temperature and humidity, read through the ventsNot modeled
  • 16×2 I²C LCDSystem state at the wall; held by the press-fit windowIn the CAD
  • HC-05Bluetooth stream of temperature, AC state and timingNot modeled
  1. 1MeasureDHT22 reads the room
  2. 2DecideArduino applies temperature and time logic
  3. 3CommandSG90 moves 0° → 35° → 70°
  4. 4RespondAC switch toggles; the room temperature follows
The servo is the bridge between the software and a legacy AC system that was never designed to be automated.
MATLAB bar chart of total weekly energy cost: always on is highest, manual use is lower, and the Smart AC V1 controller is lowest, with error bars.
Weekly energy cost from the logged data: always on, manual use and the controller (MATLAB). Bluetooth logging every 10 s.
All work
The black two-part controller enclosure with its blue LCD and angled side vents (CAD render).

Independent project

Smart Climate Control MonitorHow do you add intelligent control to a 20-year-old AC system without modifying its internal wiring?

A wall-mounted retrofit that physically flips the existing switch, designed in SolidWorks and built to cut energy use, not just add convenience.

lower observed energy cost in the tested setup
≈ 25%
Measured

Scroll to take it apart

The external retrofit

The solution had to sense the room, physically operate the existing switch, and add intelligence without opening the wall.

Enter the cross-section

The enclosure had to be designed around motion, clearance, sensing, mounting, feedback, and manufacturing.

Servo becomes the interface

The servo converts digital commands into repeatable mechanical motion at the existing rocker switch.

Calibrated positions from the build. The rocker switch itself is not part of the CAD, so the arc shows the horn travel only.

Reveal the controller

The Arduino coordinates sensing, timing, and actuation inside a mechanically constrained package.

Its place sets the enclosure volume, the wiring access and the assembly order, which is why it is here, not because the electronics are the subject.

Sense and communicate

The DHT22 measures the local environment. The LCD gives the user immediate system-state feedback. Bluetooth makes the behavior measurable.

The DHT22 and HC-05 are not modeled in the CAD; the LCD, its contrast potentiometer and the breadboard are.

Design the enclosure

The enclosure was built around the hardware envelopes and the assembly sequence, not around an arbitrary rectangular box.

PETG layers, three perimeters
0.2 mm
Print settings
clearance holes for #8 drywall screws
4.8 mm
Design

Press-fit LCD window, self-threading M3 bosses, bridging-safe vents and minimal supports.

Close the control loop

Measure temperature. Decide when cooling is needed. Command the servo. Observe the room respond.

Return to the wall

The design only matters if the mechanism works where the old switch actually lives.

Measure performance

The controller did not just move the switch. It logged temperature, AC state, actuation timing, and energy-cost data every 10 seconds.

Result and limitation

In the tested setup, the controller reduced observed energy cost by approximately 25% and reduced the daily temperature swing from 4.2°C to 3.3°C.

lower observed energy cost
≈ 25%≈ 25%
Measured
daily temperature swing
4.2 → 3.3 °C
Measured

Results from the tested setup. Retrofit without modifying internal wiring.

Next

Alef battery enclosure