Closed-Loop Temperature Control (Peltier + TMP117)

Overview

Designed and implemented a closed-loop temperature regulation system for a custom acrylic enclosure using Peltier thermoelectric modules and a discrete-time controller derived from continuous-time control theory. The controller was discretized using the Tustin transform and deployed on an embedded microcontroller, enabling stable real-time regulation without relying on a standard PID formulation.

Key contributions

• Modeled enclosure thermal dynamics using a lumped parameter approximation
• Designed a continuous-time compensator and discretized it for embedded execution
• Implemented a real-time discrete-time controller on Arduino Mega
• Integrated PWM-controlled Peltier heating with closed-loop feedback

Physical System

Thermal chamber

• 30 × 30 × 30 cm acrylic enclosure

Actuation

• 4 × Peltier (TEC) modules
• Hot side facing inward, cold side outward
• External heatsinks for heat rejection
• PWM control in heating mode

Sensing

• TMP117 precision temperature sensor suspended at the geometric center (mid-air) to reduce boundary bias

Controller hardware

• Arduino Mega executing the control loop and multiplexing two temperature displays

Control System Architecture

Closed-loop feedback structure:

Control loop

1. TMP117 measures enclosure air temperature
2. Controller computes control effort at each sample
3. PWM drives the Peltier heating modules
4. Thermal dynamics produce next temperature state

Thermal System Modeling

Approximated the enclosure as a lumped thermal system with effective resistance and capacitance:

This abstraction captures dominant thermal dynamics while remaining tractable for controller design and embedded implementation.

Controller Design and Discretization

Continuous-time compensator

Discretization (Tustin)

Controller discretized using a bilinear transform with a 0.1 s design interval:

Resulting implemented difference equation:

Embedded Implementation

Controller deployed on Arduino Mega with real-time execution:

• Persistent state for previous control effort and error terms
• PWM saturation to maintain safe actuator bounds
• Potentiometer-adjustable 50–80 °F setpoint
• Dual multiplexed displays for setpoint and measured temperature

Validation and Analysis

Validated expected closed-loop behavior using frequency and time-domain analysis:

Engineering Challenges and Solutions

Thermal latency and slow dynamics

• Required stability-focused design and smooth control effort
• Addressed via compensator design rather than aggressive gain

Continuous-to-discrete fidelity

• Needed discrete implementation to preserve continuous-time behavior
• Tustin discretization helped maintain stability and response shape

Measurement bias

• Wall and surface temperature bias avoided by suspending sensor at enclosure midpoint

Actuation constraints

• PWM saturation and switching logic implemented to keep control stable and safe

Results

• Deployed a theory-driven controller on embedded hardware
• Demonstrated stable closed-loop regulation architecture using PWM actuation
• Raised the enclosure temperature from approximately 70 °F toward a 77 °F setpoint during experimental validation
• Established an end-to-end workflow from modeling → controller design → discretization → embedded deployment

Technical Stack

• Continuous and discrete-time control systems
• Tustin transform discretization
• Arduino (embedded C++)
• PWM Peltier heating
• Thermal system modeling
• TMP117 sensor integration

Future Improvements

• Add multi-point sensing to measure gradients and validate lumped-model assumptions
• Automate logging/plotting for repeatable evaluation across setpoints
• Benchmark against PID baseline under identical constraints
• Add actuator thermal monitoring and protection
• Add active cooling to reduce overshoot and support bidirectional regulation