Load cell output is proportional to excitation: a 2 mV/V cell at 10 V produces 20 mV at rated capacity. Doubling excitation doubles the signal, which is tempting when the indicator resolution is marginal. The cost is four times the heat in the bridge, because power rises with the square of voltage.
Self-heating and creep
Bridge power in a 350 ohm cell at 10 V is about 285 mW, spread over a small element. That is enough to raise the gauge temperature a few kelvin above ambient and to change the creep behaviour, which is why manufacturers specify a maximum excitation and sometimes a recommended value for continuous use. Running at 15 V when the datasheet says 10 V is a common cause of drift that looks like a bad cell.
Matching the indicator
Choose excitation so that rated output uses 50-80% of the indicator input range. Below that, resolution suffers; above it, you lose the ability to see overload. For a 2 mV/V cell and a ±20 mV input, 5-8 V is the sweet spot.
Pulsed excitation
Some instruments pulse the bridge to cut average power while keeping peak signal. It works well for fast weighing, but the settling time of long cables can make pulsed excitation read low on high-capacitance runs. Check the manufacturer settling specification before assuming pulsed is better.
