# Instrumentation Amplifier - how to use it correctly

I am trying to perform some measurements based on a special technique used to detect corrosion in rebars of concrete bloc. The idea is to measure the potential difference between an electrode and the rebar: I tried it with a conventional voltmeter. It worked just fine. The potential difference lies between 0V and 1V.

Now, I want to do it with a micro controller. My idea was then to connect the two inputs (electrode and rebar potential) to the input of an instrumentation amplifier and then to read the output with an ADC input of the micro controller.

To do so, I used the INA118.

I left the Rg input unconnected to have a unity gain.

V+ = 5V V- = GND (therefore "single supply operation") Vo = ADC input (with 100nF capacitor to GND) Ref = GND

I also connected two resistors of 1MOhm between Vin-,+ and GND.

Now the problem is that I have nothing on the output of the INA118. More precisely I have 30mV always; even when I short-circuit Vin+ and Vin-.

What is wrong with my circuit?

Should I add an offset voltage to the inputs of the INA118? I have read that: "With single supply operation, V+ and V– must be both 0.98V above ground for linear operation. "

Or is it that the input bias current is too low?

Thank you

Image of the circuit: • Can you post a schematic of the microcontroller circuit you use? – AndrejaKo Apr 14 '12 at 19:52
• As a new user, I am not alowed to post a picture apparently. – Pelotudo Apr 14 '12 at 21:23
• What is the actual voltage at the inputs ? – Rocketmagnet Apr 14 '12 at 21:24
• I will have to check. I don't have the circuit right now. The circuit is more or less like the upper one of figure 3 from the datasheet: ti.com/lit/ds/symlink/ina118.pdf The difference is that the Rg is not present, the resistors are 1MOhm and the output is connected to the ADC of a PIC dsPIC33FJ. – Pelotudo Apr 14 '12 at 21:29
• FYI, here's a classic text on InAmps: A Designer's Guide to Instrumentation Amplifiers. whole document, or broken up into chapters – Nick Alexeev Apr 15 '12 at 9:18

Input voltage range issue. With 0 & +5 supplies, the input range is only 1 to 4 volts. Your inputs are below the range where this amp will work. Check "linear input range" on datasheet.

One fix is to get negative voltage on V-. Another is to bias one end of the voltage measurement. Eg, bias the low side to 1.5 V and leave the high side unbiased/high impedance.

• I see what you mean. I could try the two options. Regarding the solution of providing a negative voltage, I don't have a negative voltage with respect to the ground of the microcontroller. However, I have access to a 12V output power supply. Could I set Vss to gnd, Vdd to 12V and Vref to 6V in order to have the In Amp just as if it was supplied to +-6Volts? – Pelotudo Apr 14 '12 at 22:07
• You could. But you'd still have to bias to voltage signal. V- would still be 0 V, the minimum input 1 V, and the voltage input could not be grounded. – user9224 Apr 14 '12 at 22:28

As has already been mentioned, your circuit doesn't work because you violated the amplifier's common mode range. This is yet another example where reading the datasheet carefully would have avoided a obvious mistake.

A simple fix is to get a amp with common mode range that extends to ground. However, you can easily bias the amp you have near the middle of its range like so: R1 and R2 form a voltage divider to make 1/2 the supply voltage. C1 attenuates nasty frequencies that might be on the supply that are high enough so that the amp's active circuitry can't eliminate the noise as common mode signal. R4 and R3 apply this 1/2 supply bias voltage to each of the inputs.

• Does this solution assume we know the impedance of the voltage source, and that the impedance is constant ? – Rocketmagnet Apr 15 '12 at 17:04
• @Rocket: It assumes that the voltage source can handle 2 MOhm differential. This is the same loading that the OP applied in his original circuit, so I just copied it since we have not been given any other specs. This is basically the OP's circuit with the two 1 MOhm resistors going to 1/2 the supply voltage instead of ground. – Olin Lathrop Apr 15 '12 at 17:19
• Yeah, I was a bit dubious about his pull down resistors for the same reason. I mean, isn't the point of an InAmp to cope with very high impedance sources? How much drive does a piece of concrete have anyway? – Rocketmagnet Apr 15 '12 at 17:23
• @Rocket: InAmps do generally have high imput impedance, but that's not the only reason for using them. The main point is true differential measurement with a fixed and known gain. Sometimes high imput impedance is require, but sometimes not. I don't know what the impedance of the concrete and rebar battery is, but 2 MOhm sounds high enough to measure its voltage. – Olin Lathrop Apr 15 '12 at 17:55

Resistance of concrete (if dry) might be quite high, up to GOhm and unstable (depend on humidity -can change 3 orders of magnitude! area of your electrode, force applied to it,..). So 1MOhm input resistance can simply shorten the voltage You want to measure. I would use FET input Instrumentation Amplifier. Or use High Input impedance voltmeter with computer interface. added: I have checked: the meter for concrete I have found in internet had 100MOhm input resistance, 2MOhm upper range and sponge type electrodes for good contact. May be problem is in contact resistance between electrode and concrete?.