Strain Gauges and Display

[geo_leeman]

Hello all.  A bit of a detailed question, but lots of details included! We do a demonstration in our group of how rocks are really springs (linearly elastic).  There is a 4 gauge bridge attached to a block of granite (or my coffee cup) and it gets powered from a 9V battery.  Attached photos show my coffee mug and a plot of it contracting as the coffee cools.  The gauges are 120 ohm.  Traditionally we have shown guests this demo on a strip chart recorder.  It's very sensitive and visual, but seems a bit odd telling them how advanced the lab is, then whipping a strip chart out  .

I've successfully put the bridge output into a 16 bit ADC on an Arduino with some internal amplification and got an electronic strip chart that can be projected on a classroom projector.  The signal was still small and that ADC board is overkill for what we need.  I'm going to build up a little amplifier to use the 10-bit ADC already on the Arduino.  This ADC goes by default from gnd to 5V, but can be modified with the analog reference pin.  The whole setup will be run on a 9V battery.

My idea (drawing attached) is to use the AD620 as the differential amp, and an OP177 providing a buffered divider offset for single rail operation.  The low-pass on the AD620 is from the datasheet with values calculated for about 26Hz Fc.  This setup does work, but has a few bumps.  First, the bridge is never perfectly balanced, so there is some offset.  With a very large gain, I end up saturating.  We are talking about changes of mV at the most... students are squeezing a block of granite, so the strain is very small!

My questions are:

1) Am I going about this totally backwards, or is this the right idea?
2) Should I just use unity gain, then offset again?
3) Should I power the amps from 5V on the Arduino, or from 9V and change the analog reference input on the Arduino?
4) Is the AD620 the right chip for the job?

Thank you in advance for your help!

[Rerouter]

as its a full bridge, you can sum into the op amp inputs a weak pull up / pull down to cancel the offset.

[geo_leeman]

That would work.  Any ideas on the ADC side or on chip selection? I'll probably make the voltage divider adjustable so that the user can move it as well.  Maybe use a rail-to-rail instead of the OP177.

[Rerouter]

your forgetting the magic of the instrumentation amplifiers, if you inject say 0.5V into the reference pin then a zero output corresponds to 0.5V allowing you to get away with adc's that cannot measure too close to ground,

note it is best if this voltage is buffered by an op amp as it should be low impedance,

[geo_leeman]

Right.  I was mainly thinking to allow the OP177 (the voltage divider buffer) to go rail to rail so that the reference to the instrumentation amp has full range to help offset any bridge imbalance.  I'm in the process of moving this week, but hopefully can pull out the granite block and test this design with the Arduino soon.  Thank you for the ideas!