Bridge Sensor Excitation Circuit Design

A bridge sensor turns a physical quantity into a small differential voltage, and that voltage is proportional to the excitation applied to the bridge. The excitation circuit therefore sets both the sensitivity of the measurement and the error caused by self heating.

Why the Bridge Needs a Defined Excitation

A bridge sensor is four resistances arranged so that the quantity being measured unbalances them. The output is the difference between two dividers, and it scales with the voltage applied across the bridge.

Two bridges of the same type with different excitation voltages produce different outputs for the same input. That is why the excitation is part of the specification, and why a change in it appears as a scale error rather than as an offset.

The excitation also dissipates power in the bridge elements. A strain gauge bridge with a hundred and twenty ohm arms at five volts dissipates about fifty milliwatts, which heats the sensing element and shifts its zero.

excitation voltage and Self Heating

Lowering the excitation voltage reduces the self heating but also reduces the signal. The optimum is the highest voltage at which the self heating error stays inside the budget, and the datasheet usually suggests one.

A pulsed excitation with a low duty cycle removes the trade, because the measurement is made before the element has time to heat. The circuit is more complex and the measurement has to be synchronised with the pulse.

Where the sensor is in still air, the self heating is much larger than the same sensor in a moving liquid. The dissipation constant of the installation, not of the sensor alone, sets the error.

ratiometric measurement Explained

In a ratiometric measurement the analog to digital converter uses the excitation voltage as its reference. The reading is then the ratio of the bridge output to the excitation, and any change in the excitation cancels.

That removes the requirement for a precise excitation source and replaces it with a requirement for a stable ratio. A supply that varies with temperature no longer matters, provided the reference follows it.

The technique requires the converter reference to be taken from the actual bridge excitation rather than from a separate rail, and that the two are connected with the Kelvin connection so that the lead drop does not appear in the ratio.

Kelvin connection to the Bridge

A Kelvin connection brings the excitation to the bridge on one pair of conductors and senses the voltage at the bridge on another. The current carrying pair drops voltage in its resistance, and the sense pair does not carry current, so the measurement excludes that drop.

For a six wire connection the same principle is applied to the output, so that the lead resistance in the output path is excluded as well. Six wire connections are used on sensors with several metres of cable.

The sense pair has to be connected at the bridge terminals rather than at the instrument. Connected at the instrument, the arrangement measures the lead drop and gains nothing.

Constant Current Excitation

Driving the bridge with a constant current rather than a constant voltage has an advantage: the output depends on the change in resistance rather than on its absolute value, and the lead resistance no longer divides the excitation.

The current source has to be accurate and stable, and it dissipates its own power. Where the bridge is at the end of a long cable, the constant current approach removes the sensitivity to the cable resistance.

The bridge output with a constant current is twice as large for the same power, which is a useful improvement in a low level measurement. The trade is the complexity of the current source.

Interaction with sensor drift

sensor drift is the slow change in the zero and the span of the bridge. Part of it comes from the elements themselves and part from the mechanical assembly, and the excitation affects the second.

A bridge held at a constant temperature drifts less than one that is heated and cooled by the excitation. The excitation circuit and the thermal design are therefore part of the same problem.

Where the drift is dominated by temperature, a temperature sensor on the bridge body allows the firmware to correct it. The correction is worth making only when the drift has been measured rather than assumed.

Noise on the Excitation

Noise on the excitation appears in the output in the same proportion as the signal, so the excitation has to be as quiet as the precision of the measurement requires. A switching supply directly across a bridge is a poor combination.

A linear regulator or an RC filter after the switcher gives a quiet excitation at the cost of some efficiency. The filter capacitor should be placed at the bridge when the cable is long, so that the noise is removed where it matters.

The excitation current also produces a magnetic field around the conductors, and a twisted pair for the excitation and for the output keeps the resulting coupling small.

Cable Length and Compensation

The cable between the bridge and the instrument carries both the excitation and the output. Its resistance is small compared with the bridge arms, and its temperature coefficient is far worse, which is why the Kelvin connection matters as soon as the cable is longer than about a metre.

A four wire or six wire connection adds conductors to the cable, and the cost of that is small compared with the error it removes. The conductors in each pair must be identical, because a difference between them appears directly in the measurement.

Where the cable runs through a region of changing temperature, the compensation is only as good as the matching of the conductors. A screened cable with a separate drain wire is convenient, and its two signal conductors are often not identical, which is worth checking before it is used for a precision channel.

Verification and Faults

Verify the excitation at the bridge terminals rather than at the instrument, using the sense pair as the measurement point. A value lower than expected indicates lead drop, which the Kelvin connection should have excluded.

Check the bridge output with no load applied and at a known load, and confirm that the ratio between the two matches the datasheet. A ratio error indicates an excitation problem, while a constant error indicates an offset.

A reading that changes when the cable is moved is a connection problem, and one that changes slowly with the ambient temperature is a drift problem. The release checks that keep such a board consistent are collected in our PCB design release checklist, the assembly points in judging PCB quality, and the layout measures in our guide to mixed signal board design.

FAQ

Should I use constant voltage or constant current excitation? Constant current removes the cable resistance and doubles the output for the same power, at the cost of a more complex source.

What is a ratiometric measurement? Using the excitation as the converter reference so that a change in the excitation cancels in the ratio.

Why does my bridge reading drift? Self heating and the thermal environment. Reduce the excitation, use a pulsed drive or correct for the measured temperature.