1. The Fundamental Trade-Off

Selecting the excitation current for an RTD is a balancing act between two opposing errors:

• Self-Heating (High Current): Sending more current increases the signal voltage ($V = IR$), which is good for noise immunity. However, power dissipates as heat ($P = I^2R$). In stagnant air, even 1mA can heat a Pt1000 sensor by 0.5°C, ruining the accuracy.
• Resolution & Noise (Low Current): Reducing current eliminates self-heating. But if the current is too low (e.g., 10µA), the signal voltage becomes tiny ($10\mu A \times 100 \Omega = 1 mV$). If your ADC has a resolution of 0.1mV, you can only resolve temperature in 0.25°C steps. The signal gets buried in the noise floor.

2. Pt100 vs Pt1000 Strategy

The resistance value changes the rules significantly.

• Pt100: Low resistance. Needs higher current (typically 1mA) to generate a readable voltage. Because resistance is low, $I^2R$ heating is manageable.
• Pt1000: High resistance. Generates 10x the voltage for the same current. But also generates 10x the heat. Standard practice is to reduce current to 100µA or 200µA. Using 1mA on a Pt1000 in air is a common design error.

3. The Digital Floor (ADC)

An Analog-to-Digital Converter (ADC) chops the voltage range into discrete steps. For a 16-bit ADC with a 2.5V reference, the step size (LSB) is $2.5 / 2^{16} \approx 38 \mu V$.

If you use a Pt100 with 100µA current, the sensitivity is only ~38µV/°C. This means your fancy PLC can only detect 1°C changes! You must match the current to the digitizer's capability. High-precision metrology uses 24-bit ADCs, allowing lower currents without resolution loss.

4. Environmental Impact ($P_D$)

The self-heating error depends entirely on where the sensor is.
Water ($P_D \approx 50 mW/^\circ C$): You can blast 5mA through a sensor in flowing water with zero error. The water strips the heat away.
Still Air ($P_D \approx 1-2 mW/^\circ C$): The heat has nowhere to go. Limits are very strict.