10V External vs. 3.3V Internal Excitation Voltage
APPLICATIONS > FORCE MEASUREMENT >

There are two primary excitation options when working with Strain Gage or Load Cell sensors. One is the internal 3.3V reference voltage provided by the i42x/i43x cards; and the other is an external 10V reference supply. We will compare these and focus on four primary considerations, described below.

All i42x/i43x/i60x devices provide a 3.3V reference voltage for sensors. This is read back by the instruNet system to an accuracy of approximately 0.01%, and this value is used to calculate the sensor value.

If one uses an external reference, then it is NOT read back by the instruNet system. Instead, the end user tells the instruNet software what voltage is being applied; and one wants it to not change from this reported value. If instruNet's understanding of the reference voltage is off by 2%, for example, then the measured sensor value would be off by 2% as well.

There is a device called a "power supply" that outputs a voltage typically accurate to 5%, which is not very accurate. Also, if one measures it accurately with a volt meter; then the question is how stable is this over time, temperature and load. And how much noise is added to this voltage by the power supply machine itself. The answer for most power supplies is not too stable and too much noise.

There is another device called a "precision voltage reference". This provides a very accurate and stable voltage, yet often does not support high loads.

If you do supply an external voltage reference then make sure it is stable over time, temperature, and load; has low noise; and supports the load required by your sensors.

The easy solution is to use the 3.3V reference provided by the instruNet hardware.

Energy is pumped into a sensor when a voltage is applied across its terminals. This energy causes the device to heat up; and this is referred to as "self heating". More self heating occurs with higher voltages and smaller resistances across the device.

Many devices change when their internal temperature changes. A load cell, for example, will report a different weight, to some extent, if the load cell device itself is heated. This change is specified in the device datasheet under "thermal drift".

If a strain gage is affixed to a beam, and the gage heats, then it will cause the beam to heat as well. And this will cause the properties of the beam to change slightly.

One often minimizes sensor and material-under-test self heating with lower excitation voltages (e.g. 3.3V instead of 10V) and higher device resistances (e.g. 350Ω instead of 120Ω). Another way to minimize device heating is to affix it to a large thermal load (e.g. large chunk of metal). The opposite of this is a device in air with no heat sink that becomes so hot that it incurs permanent damage (i.e. it burns up).

The power that is pumped into a sensor is calculated with the below equation where "Voltage" refers to the number of Volts across the sensor and "Resistance" refers to the resistance across the sensor in units of ohms.

Power (watts) = Voltage * Voltage / Resistance

Strain Gages and Load Cells have four internal resistors of approximately the same resistance in a Bridge configuration. The resistance across the four resistors (i.e. top of bridge to bottom of bridge) is the same as the resistance across one of the four resistors (i.e. remove one resistor from bridge and measure resistance across it).

An example would be to place 10Volts across a 350Ω strain gage or load cell. This would pump 10*10/350 = 285mWatts into the 4 bridge resistors (87mW into each resistor). This is a lot of energy and will cause the device to become hotter. 3.3V is more gentle and would result in 3.3*3.3/350 = 31mW of heating, which is 1/10th as much as what we see with 10V. 120Ω gages see approximately 3 times more heat with 833mW at 10V and 90mW at 3.3V. Note that power increases with the square of voltage and linearly with resistance.

In summary, 3.3V is preferred over 10V when taking into consideration device self heating.

If you increase the excitation voltage by a factor of three, the output signal (i.e. voltage across bridge) increases by a factor of three as well. If the noise added to the signal is constant, then the signal to noise ratio decreases three-fold when you increase the excitation voltage by three. For this reason, 10V excitation is better than 3.3V when taking into consideration noise. However, in most cases, instruNet filters noise with Integration and/or analog low pass filtering; and is kept to an acceptable level with 3.3V excitation. Yet there are scenarios where noise is a dominant source of error and 10V external excitation is favored over 3.3V.

In many cases, the device datasheet specifies an excitation voltage of 10Volts. This is often misleading, since it implies one must run the device at 10Volts. Yet the device determines the ratio of the input voltage to the output voltage, and therefore works fine at lower excitation voltages, provided the signal-to-noise ratio is acceptable.

Conclusion

The accuracy of a measurement is determined by several factors, described above, and in most cases, the internal 3.3V reference provides a more accurate and reliable measurement than an external 10V reference.