# Instrumentation (or diff) Amp saturating when measuring low voltage

I need to measure (amplify) a very low voltage, by low I mean in the order of 20 uV. Right now, using a lab multimeter, I am clearly and repeatably measuring the voltages so I thought that amplifying it to about 1V in order to measure it with a 10bit ADC would be easy. Of course I am having trouble with it.

After many tests with different resistance values and circuit config, I stepped back to a simple 2 stages differential amplifier, with a TL082 (2 op amps package, Input impedance of 10^12ohm). The voltage supply is of -15 to 15V so I have a quite good margin here.

simulate this circuit – Schematic created using CircuitLab

As you can see, I have a strain gauge circuit to the left. Since it was ripped from from a 10\$ kitchen scale, the 2 gauges and the resistors are glued and unavailable. Still, being able to measure it with the multimeter made me confident that I could simply amplify the signal.

The problem is that my equivalent gain is (200k/1k)*(100k/1k) = 20000 (which I want). If I try to measure the voltage Vin, Vout goes to about half the positive rail voltage (here, around 7.5V), then bending the straing gauge assembly (varying Vin) shows a seemingly linear augmentation of Vout. According to my calculation an measurements, when Vin = 0.03mV, Vout should be of 0.03mV * 20000 = 0,6V. So, reading 7.5V at Vout is seemingly incorrect in the first place.

What is strange is that if I simply measure the voltage across a floating resistor of 1k, instead of reading ~0V (plus noise) I read a Vout of 7,5V. But if measure a floating resistor of 200k, damn I measure around 0,7V, which is exactly the kind of value I would expect of a noise being amplified 20000 times.

So, is anyone up with something on this? Oh, I should mention that:

• First, we used a full instrumentation amplifier to provide the 20000 gain and the behaviour was similarly wrong, so we stepped back to this "simpler" circuit.
• We have access to lab strain gauges and will probably try to use them instead of the ripped off bridge from the kitchen scale.
• I am a TA, not a student with a homework ;) I want to use this a pedagogical tool for my student.

Thanks everyone!

• What op-amp power rails do you have and if you say +15 volts and 0V think again! Commented Nov 28, 2014 at 22:14
• +15 and -15 V ;) Commented Nov 28, 2014 at 23:27
• 10mv input offset * 20000 gain is about 200V...
– user16324
Commented Nov 29, 2014 at 0:40