When you first look at a circuit diagram of a Wheatstone bridge, it appears as a simple diamond of four resistors. But in precision measurement, this topology is the gold standard for extracting tiny resistance changes from strain gauges, RTDs, and thermistors. The direct answer to how it works is that it converts a minute change in resistance into a measurable differential voltage by comparing a known reference voltage divider against an unknown sensor divider.
Unlike a basic series-parallel network, the bridge operates on the principle of a null condition. When the ratio of the left arm matches the ratio of the right arm, the differential output is exactly zero. Any deviation in the sensor element unbalances the bridge, producing a voltage shift proportional to the physical change being measured. Below, we break down the exact node behavior, failure modes, and a real-world component design for a 10kΩ thermistor circuit.
The Wheatstone Bridge Topology: Nodes, Arms, and the Null Condition
To analyze the circuit, we must define the four nodes and the four arms. The excitation voltage (Vex) is applied across the top and bottom nodes, while the differential output (Vout) is read across the left and right midpoints.
• Node A (Top): Excitation Voltage (Vex+)
• Node B (Bottom): Ground / Excitation Return (Vex-)
• Node C (Left Midpoint): Vout+ (Between R1 and R2)
• Node D (Right Midpoint): Vout- (Between R3 and Rx)
Vout is always calculated as V_C - V_D.
In a balanced state, the voltage divider ratio on the left (R2 / (R1 + R2)) exactly equals the ratio on the right (Rx / (R3 + Rx)). If Vex is 3.3V and all four resistors are 10kΩ, both Node C and Node D sit at exactly 1.65V, yielding a Vout of 0.000V.
Failure Mode and Behavior Matrix
Understanding what breaks at the extremes is critical for troubleshooting sensor wiring on the bench. The table below maps the exact differential output shift (assuming a 3.3V Vex and nominal 10kΩ resistors) when a single element fails open or short.
| Arm / Resistor | Normal Role | Effect of Open Circuit | Effect of Short Circuit | Vout Shift (from 0V) |
|---|---|---|---|---|
| R1 (Top-Left) | Fixed Ratio | Node C pulls to GND (0V) | Node C pulls to Vex (3.3V) | Open: -1.65V Short: +1.65V |
| R2 (Bottom-Left) | Fixed Ratio | Node C pulls to Vex (3.3V) | Node C pulls to GND (0V) | Open: +1.65V Short: -1.65V |
| R3 (Top-Right) | Fixed Reference | Node D pulls to GND (0V) | Node D pulls to Vex (3.3V) | Open: +1.65V Short: -1.65V |
| Rx (Bottom-Right) | Sensor Element | Node D pulls to Vex (3.3V) | Node D pulls to GND (0V) | Open: -1.65V Short: +1.65V |
Note: An open sensor (Rx) and a shorted R1 produce the exact same output voltage. This is why physical inspection of wiring continuity is mandatory when a bridge pegs to the negative rail.
Why a Bridge? Comparing Topologies for Precision Measurement
Why use four resistors and a differential amplifier when you could just put the sensor in series with a single fixed resistor and read the voltage? The answer lies in common-mode rejection and sensitivity to micro-changes. According to All About Circuits, the bridge topology inherently cancels out excitation voltage drift, provided the ADC reads the differential output rather than a single-ended ground-referenced voltage.
| Criteria | Wheatstone Bridge | Simple Voltage Divider |
|---|---|---|
| Sensitivity to Small ΔR | High (measures deviation from a 0V null) | Low (measures absolute voltage near Vex/2) |
| Common-Mode Noise | Excellent (rejected by differential amp) | Poor (noise directly adds to ADC reading) |
| Excitation Drift | Cancels out (both arms scale equally) | Directly corrupts the final reading |
| Component Count | 4 Resistors + Instrumentation Amp | 2 Resistors + Single-Ended ADC |
Design Walkthrough: Sizing Components for a 10kΩ NTC Thermistor
Let’s design a bridge to measure ambient temperature using a Vishay NTCLE100E3103 (10kΩ NTC thermistor, 3977K Beta, ~$0.40 on DigiKey). We will power it from a 3.3V ESP32 GPIO pin and read it with a 12-bit differential ADC.
Component Selection
Using standard 5% carbon film resistors for the fixed arms will ruin your accuracy. The tolerance stack-up alone will prevent the bridge from ever reaching a true 0V null at 25°C. You must use precision thin-film resistors.
| Designator | Part Number / Spec | Value | Tolerance & Tempco |
|---|---|---|---|
| R1, R2, R3 | Susumu RG1608P-103-B-T5 | 10.0 kΩ | 0.1%, 25 ppm/°C |
| Rx | Vishay NTCLE100E3103 | 10.0 kΩ @ 25°C | 1%, Beta 3977K |
| Vex | ESP32 3V3 Pin | 3.3V DC | ±2% (USB dependent) |
The Math: Null vs. Shift
At exactly 25°C, Rx = 10,000Ω. Node C is 1.650V, Node D is 1.650V. Vout = 0.000V.
Now, let the ambient temperature rise to 35°C. Using the Beta parameter equation, the NTC resistance drops to approximately 6,490Ω. Node C remains locked at 1.650V (since R1 and R2 haven't changed). Node D shifts:
V_D = 3.3V × [ 6,490 / (10,000 + 6,490) ] = 3.3V × 0.3935 = 1.298V
Vout = V_C - V_D = 1.650V - 1.298V = +0.352V (352mV)
A 10°C shift yields a massive 352mV differential swing. A standard 12-bit ADC with a 3.3V reference has a resolution of ~0.8mV per step, meaning this bridge can resolve temperature changes down to roughly 0.02°C, limited only by the ADC noise floor and the self-heating of the thermistor.
Breadboard Testing: Step-by-Step Verification
Before committing this to a PCB, verify the null condition on a breadboard. You will need a digital multimeter (DMM) capable of reading millivolts DC.
- Prep the Rails: Connect your ESP32 or bench supply 3.3V to the red rail and GND to the black rail. Verify with your DMM that the rail voltage is between 3.25V and 3.35V.
- Place the Fixed Arm: Insert R1 and R2 in series from the red rail to the black rail. The midpoint is Node C.
- Place the Sensor Arm: Insert R3 and the NTC thermistor (Rx) in series parallel to the first arm. The midpoint is Node D.
- Route the Probes: Use short jumper wires to bring Node C and Node D to isolated, adjacent breadboard rows to give your DMM probes a stable contact point.
- Measure the Null: Set your DMM to DC millivolts. Place the red probe on Node C and the black probe on Node D. At room temperature (~22-25°C), you should read between -15mV and +15mV. (It won't be exactly 0.000V due to breadboard contact resistance and the 1% tolerance of the NTC bead).
- Induce a Shift: Pinch the glass bead of the NTC thermistor between your fingers. The body heat will drop the resistance. Watch the DMM; the positive millivolt reading should climb steadily past +100mV within seconds.
Real-World Gotchas: Lead Resistance and Thermal Drift
A standard solderless breadboard introduces 0.5Ω to 2.0Ω of contact resistance per jumper. For our 10kΩ thermistor design, 1Ω of error is 0.01%—completely negligible. However, if you adapt this exact circuit diagram for a 120Ω strain gauge or a Pt100 RTD, that same 1Ω of breadboard resistance will introduce nearly a 1% measurement error, entirely destroying your calibration. For low-impedance sensors, you must solder the bridge directly to a PCB or use a 4-wire Kelvin connection topology.
Another critical factor is self-heating. In our 3.3V design, the current through the sensor arm is roughly 165µA (3.3V / 20kΩ). The power dissipated by the NTC is I²R, which equals about 0.27mW. A standard glass-bead NTC has a dissipation constant of ~2mW/°C, meaning the bridge circuit is heating the sensor by 0.13°C just by being turned on. If you increase Vex to 5V or 12V to "get a bigger signal," you will exponentially increase self-heating and introduce severe thermal drift errors.
For load cells and micro-strain measurements where the ΔR is measured in fractions of an ohm, discrete op-amps often lack the common-mode rejection required. In those cases, bypass the discrete bridge amplifier and route the Vout+ and Vout- nodes directly into a dedicated 24-bit bridge ADC like the HX711 or the TI ADS1232, which feature integrated programmable gain amplifiers (PGAs) specifically designed to digitize the microvolt null shifts of a Wheatstone bridge.
