TMP36 Temperature Sensor Arduino Guide: Thermal Management & Heatsink Math

When most makers wire up a TMP36 temperature sensor to an Arduino, they stop at the basic voltage-to-Celsius conversion: 10mV per degree with a 500mV offset. But if you are building power-dense embedded systems—like motor controllers, LED drivers, or linear power supplies—the TMP36 is not just a way to read the room; it is a critical diagnostic tool for validating your thermal management strategy.

In this guide, we move beyond basic weather-station tutorials. We will analyze the TMP36’s own thermal limits, walk through the junction-to-ambient ($R_{\theta JA}$) thermal path math required to size a heatsink for an Arduino power stage, and show you how to use the sensor to validate your design against semiconductor derating curves.

TMP36 Thermal Characteristics and Self-Heating Errors

Before you can trust the TMP36 to measure a hot power component, you must understand its own thermal behavior. The TMP36 is a low-voltage, precision centigrade temperature sensor. While it draws very little current, it is still a semiconductor packaged in plastic, meaning it has its own thermal resistance and absolute maximum ratings.

Source: Analog Devices TMP36 Datasheet

How Hot is Too Hot for the TMP36?

The absolute maximum junction temperature is 150°C, but you should never operate the sensor near this limit. The specified accuracy of ±1°C is only guaranteed from -40°C to +100°C. Above 100°C, the internal bandgap reference begins to drift non-linearly, and the error can expand to ±2°C or more. For precision thermal management, keep the TMP36’s ambient environment below 85°C.

Calculating Self-Heating Error

Every sensor dissipates a tiny amount of heat. If you power the TMP36 from the Arduino’s 5V rail, the power dissipation is:

$P = V_S \times I_q = 5V \times 50\mu A = 0.25mW$ (or 0.00025W).

Using the TO-92 thermal resistance ($R_{\theta JA} = 152$ °C/W), the self-heating temperature rise is:

$\Delta T = P \times R_{\theta JA} = 0.00025W \times 152 \text{ °C/W} = 0.038\text{°C}$.

This 0.038°C error is negligible in open air. However, if you embed the TMP36 inside a sealed, unventilated enclosure alongside a hot stepper motor driver, the local ambient temperature ($T_A$) rises dramatically, and the sensor will accurately report that trapped heat, even if the sensor itself isn't the source.

Thermal Path Math: Sizing Heatsinks for Arduino Power Stages

The most common use for a TMP36 in a serious build is monitoring the case temperature of a power component, like a linear voltage regulator or a MOSFET. Let’s use the classic Texas Instruments LM7805 linear regulator as an example. Suppose your Arduino project uses an LM7805 to drop a 12V battery down to 5V to power a 0.5A servo and the microcontroller.

Step 1: Calculate Power Dissipation

$P_D = (V_{IN} - V_{OUT}) \times I_{LOAD}$

$P_D = (12V - 5V) \times 0.5A = 3.5W$.

Dissipating 3.5W through a bare TO-220 package without a heatsink will result in immediate thermal shutdown.

Step 2: Define the Thermal Path ($R_{\theta JA}$)

The thermal path from the silicon junction to the surrounding air consists of three resistances in series:

1. $R_{\theta JC}$ (Junction-to-Case): For a TO-220, this is typically 5.0 °C/W.
2. $R_{\theta CS}$ (Case-to-Sink): The thermal interface material (TIM). A standard silicone thermal pad or paste yields about 0.5 °C/W.
3. $R_{\theta SA}$ (Sink-to-Ambient): The heatsink itself. This is the value we need to solve for.

We want to keep the junction temperature ($T_J$) at or below 100°C for reliability, and we estimate the maximum ambient temperature ($T_A$) inside our project enclosure will reach 50°C on a hot day.

$R_{\theta JA(max)} = \frac{T_J - T_A}{P_D} = \frac{100\text{°C} - 50\text{°C}}{3.5W} = 14.28\text{ °C/W}$.

Step 3: Select the Heatsink

Now we isolate the heatsink requirement:

$R_{\theta SA} = R_{\theta JA(max)} - R_{\theta JC} - R_{\theta CS}$

$R_{\theta SA} = 14.28 - 5.0 - 0.5 = 8.78\text{ °C/W}$.

We need a heatsink with a thermal resistance of 8.78 °C/W or lower. Browsing the Aavid Thermalloy catalog, the Aavid 531202B02500G is a standard TO-220 extruded aluminum heatsink rated at approximately 7.5 °C/W in natural convection. This gives us a 1.28 °C/W safety margin.

Validating the Design and Interpreting Derating Curves

Once you have bolted the Aavid heatsink to the LM7805 and strapped the TMP36 directly to the metal tab of the TO-220 package using Kapton tape and a dab of thermal paste, you can use the Arduino to validate your math.

What Airflow and Enclosure Changes Buy You

If your thermal validation shows the heatsink is still running too hot, you have two levers to pull:

• Forced Airflow: Adding a small 40x40x10mm 5V DC fan (like a Noctua NF-A4x10) blowing across the Aavid 531202B02500G drops its effective $R_{\theta SA}$ from 7.5 °C/W down to roughly 2.5 °C/W. This buys you an extra 17.5°C of thermal headroom at 3.5W.
• Enclosure Venting: Sealed enclosures trap heat, raising $T_A$. Adding passive louvered vents at the bottom and top of the enclosure creates a chimney effect, lowering the internal ambient temperature by 5°C to 10°C without active fans.

Interpreting the Derating Curve

Semiconductor datasheets include a 'Power Dissipation Derating Curve'. For the LM7805, the curve shows that at 25°C ambient, it can safely dissipate up to 15W (assuming an infinite heatsink). However, the line slopes downward, hitting 0W at 150°C. If your internal enclosure ambient hits 80°C, the derating curve dictates that the absolute maximum power the silicon can handle drops to roughly 5.5W, even if your heatsink is keeping the case cool. The TMP36 allows you to monitor the local ambient and shut down the Arduino system before crossing this derated threshold.

Arduino Validation Code

This sketch uses the Arduino's internal 1.1V reference for higher ADC resolution, reads the TMP36 attached to the TO-220 case, and extrapolates the internal junction temperature to trigger a protective shutdown.

// TMP36 Thermal Validation & Junction Extrapolation
const int TMP36_PIN = A0;
const float V_REF = 1.1; // Internal 1.1V reference for precision
const float R_THETA_JC = 5.0; // TO-220 Junction-to-Case
const float POWER_DISSIPATION = 3.5; // Watts
const float MAX_JUNCTION_TEMP = 110.0; // Safety cutoff in °C

void setup() {
  Serial.begin(115200);
  analogReference(INTERNAL); // Use 1.1V internal reference
  // Allow reference to settle
  delay(200); 
  analogRead(TMP36_PIN); // Dummy read to switch MUX
  delay(50);
}

void loop() {
  // Average 16 reads to reduce noise
  long adcSum = 0;
  for(int i = 0; i < 16; i++) {
    adcSum += analogRead(TMP36_PIN);
    delay(2);
  }
  float adcAvg = adcSum / 16.0;
  
  // Convert ADC to Voltage, then to Celsius
  float voltage = (adcAvg / 1024.0) * V_REF;
  float caseTempC = (voltage - 0.5) * 100.0; // TMP36 formula
  
  // Extrapolate Junction Temp: Tj = Tc + (Pd * Rtheta_JC)
  float junctionTempC = caseTempC + (POWER_DISSIPATION * R_THETA_JC);
  
  Serial.print("Case Temp: ");
  Serial.print(caseTempC);
  Serial.print(" C | Extrapolated Junction Temp: ");
  Serial.print(junctionTempC);
  Serial.println(" C");
  
  if (junctionTempC > MAX_JUNCTION_TEMP) {
    Serial.println("CRITICAL: Thermal limit exceeded. Disabling load.");
    // Insert code to disable MOSFET gate or trigger relay here
    while(1); // Halt system
  }
  
  delay(1000);
}

Failure Signatures: How Hot is Too Hot and Thermal Stress

Thermal management isn't just about preventing immediate smoke; it's about preventing long-term degradation. When you use the TMP36 to monitor your system, watch for these specific failure signatures of thermal stress:

2. Electromigration and Bond Wire Lift-Off: If you consistently run a silicon junction above 125°C, the high current density combined with heat causes metal atoms in the microscopic traces to physically migrate. Over months, this increases on-resistance ($R_{DS(on)}$ for MOSFETs) and eventually causes the microscopic bond wires connecting the silicon die to the package pins to snap. The TMP36 will show normal temps, but the component will fail open-circuit.

3. Package Delamination: In high-humidity environments, rapid thermal cycling (heating up to 100°C under load, cooling to 25°C at idle) causes the plastic epoxy package and the metal leadframe to expand and contract at different rates. This breaks the hermetic seal, allowing moisture to reach the die. You will see this as erratic sensor readings or sudden short circuits.

4. TMP36 Bandgap Breakdown: If the TMP36 itself is subjected to temperatures exceeding 130°C, the internal bandgap voltage reference will permanently shift. The sensor won't necessarily melt, but it will output a permanently skewed voltage (e.g., reading 45°C when the room is 25°C). Always mount the TMP36 on a small breakout board with extended leads if you need to measure a heat source that exceeds the sensor's own 85°C precision limit.

By treating the TMP36 as a metrology instrument rather than just a basic input device, and by rigorously applying $R_{\theta JA}$ math to your power stages, you can build Arduino projects that survive the realities of continuous high-current operation without relying on guesswork.