Load cell: low excitation voltage?

[Emil]

I'm making a battery powered device with a load cell in it. I would like to run it from a CR2032 cell, so not much more than 2V from a nearly empty cell after an LDO drop.

Most load cell datasheets recomend an exitation voltage of 10V and while they specify a maximum voltage, I have not seen any specifications for minimum voltage. As a loadcell is just a Wheatstone bridge, I can't see any reason why it wouldn't work with an arbitrarily low voltage(?)

The full load output voltage will drop linearly with the excitation voltage, so a higher gain amplifier would be required and noise could be more of a problem. Other than that I can't see any real problem with a low excitation voltage. Am I missing something?

[daqq]

You are correct, a standard load cell is just a few resistors in a bridge. The reason for wanting a higher excitation voltage is that the signal output is VERY low out of them. Generally it's in the order of 1mV @ full load for every 1V you put in as the excitation voltage. So it is desirable to have a high excitation voltage to get more mV per full load.

[retrolefty]

 Lower bridge excitation voltage will generate less output voltage as you stated. Also sensor self-heating will be lower, a good thing for some sensor types. If accuracy is an issue, a bigger problem will be dealing with falling battery voltage effecting measurement accuracy unless regulated. Bridge voltage is effectively your measurement voltage reference and needs to be as stable to your accuracy requirements are.

[daqq]

If accuracy is an issue, a bigger problem will be dealing with falling battery voltage effecting measurement accuracy unless regulated.

You can get around this if you use the excitation voltage as the reference voltage for your ADC.

[T3sl4co1l]

Remember to use pulsed measurements.  No sense in leaving it on all the time if you only need a measurement every second or so!

Duty cycle is then determined by settling time of the rest of the circuit (including startup time of the amps, if you're disabling them as well, which isn't a bad idea).

Tim

[NivagSwerdna]

The commercial scale disassembled here... https://aardvarklabs.wordpress.com/2013/02/03/disassembling-the-scale-for-the-first-time/ uses a single CR2032.  During sleep it intermittently checked for load by applying battery voltage to see if there is someone on the scale, when awake it boost the excitation voltage to 3v76.

[Emil]

Remember to use pulsed measurements.  No sense in leaving it on all the time if you only need a measurement every second or so!

Duty cycle is then determined by settling time of the rest of the circuit (including startup time of the amps, if you're disabling them as well, which isn't a bad idea).

I only need a measurment every 30 to 60 minutes, so I will definetly leave it off between measurements. The reason I'm interested in low voltage is not to reduce power consumption of the load cell when it's on, but rather to avoid the need for a higher voltage battery (or a boost converter).

[Emil]

If accuracy is an issue, a bigger problem will be dealing with falling battery voltage effecting measurement accuracy unless regulated.

You can get around this if you use the excitation voltage as the reference voltage for your ADC.

I'm planning to use a regulated supply for the excitation voltage. I was planning to also use the excitation voltage as the reference voltage, but the STM32 I plan to use can not have VDDA < VDD, so I would need to add some components to keep it at the same level as VDD when the regulator for the excitation voltage is disabled.