350ohm strain gauge measurement questions

[cmcraeslo]

Hello all,

I'm designing a 350ohm strain gauge measurement (2 half bridges)

This is what I have so far:
- I have noisy environment and it's not important to output the absolute value of the measurement, so I was thinking of ratiometric conversion, rather than absolute
- Planning to use INA3333 (https://www.ti.com/lit/gpn/INA333)
- Supply voltage for the ADC's REF voltage will be 3.3V (standard LDO, filtered)
- I'm planning to split the 3.3V supply to 1.65V and feed that to a INA's VREF AND E+ (split will be done using a opamp so low output impedance into the REF pin). Plan is to get 1.65V +/- 1.5V into the ADC.
- According to Figure 35 of the INA3333, they show "proper common mode voltage" resistor of 150 ohm (R1) connected in series to ground. Am I right to assume that I need to do 350/2 resistor here? I've seen some designs using 1.27k ohms here. Not sure what this resistor is for?
- I will not be able to calibrate it on power up, only when installed in the product for the first time, so some temperature compensation will be needed. Was thinking of adding a NTC to the uC board and just do error compensations/measurements/calibrations vs temperature once I have the circuit running. NTC will be in a place that will 100% correlate to the temperature of the strain gauges and wires. Am I over simplifying things here?

Would love your comments before I start.

Thanks

[Kleinstein]

Are the two half bridges coupled to form a full bridge, or are they separate to measure 2 independent loads ?
For 2 independent measurements one would need a reference divider of some sort (e.g. 2 precision resistors or alternatively use a high resolution (e.g. 24 bit) ADC with no extra amplification and use a digital reference level).

The resistor at the OPA333 is there to set the gain.

[cmcraeslo]

It will form a full bridge.

@Kleinstein See attached schematics for resistor I had in mind.

[Kleinstein]

The resistor R1 is there to reduce the current through the bridge and maybe adjust the common mode voltage. Usually the point half way should be OK.

It depends on the thermal properties if the bridge can use a full 3.3 V. It may still run a little to warm and thus may need the added resistor. If the ADC supports differential reference inputs one could use the reference from only the bridge and this way reduce the effect of R1 on the scale factor.

The ADC should probably run in differential mode, with one input to the virtual ground.

[cmcraeslo]

I plan to supply the bridge (E+) with 1.65V and E- to ground.

[Mecanix]

Found out the 120ohms SG are far more stable and drift much less, depending on the sensitivity of your project ofc.
The AD623 with a DAC, also ;-)

[KT88]

The CN0107 from Analog Devices shows a handy way to do your application. It allows to take voltage from the bridge supply as the reference. You get a perfecly ratiometric measurement that way. The input to the ADC is fully differential which inherently suppresses common-mode noise and allows effective filtering. Converting the bridge output into single ended with a discrete amp might create more problems than it solves...


Cheers

Andreas

[cmcraeslo]

KT88, I forgot to mention, that I need fast sampling. If I'm reading this correctly, AD you mentioned has 16hz update rate? I need atleast 10k.

[Kleinstein]

If one needs a fast update rate, noise usually gets very important as the bandwidth is high. There are a few ADC that can read fast with high resolution, but a separate amplifier can be lower noise.
The INA333 is however is low power and more like high in noise. Depending on the ADC used and accuracy requirements, one could get away with just a single low noise OP for the amplification. 2 OPs would be the better and more standard solution. 

[cmcraeslo]

Do you have an example with 2 opamps as "better" solution?

[Kleinstein]

An example circuit is found in the data-sheet to ADA4522. The capacitors may need adjustments depending on the ADC and speed used.