Use LM2902N as the op amps to build an instrumental amplifier

[convine]

Hi,everyone!

I'm trying to build an Instrumentation amplifier that would do a 10uV -> 10mV amplification. The motivation is to measure uA currents on a small enough shunt resistor (1-10Ohm).

For a proof of concept, I've built a circuit like the one below with a distinction that I'm powering it with two 9V batteries and am using LM2902N(datasheet for reference:http://www.kynix.com/uploadfiles/pdf9675/LM2902N.pdf) as the op amps and Rg is somewhat different. It has much less that the required 1000x gain.



However, when measuring the results for different input voltages, the gain seems to vary:



I was expecting it to be constant.

What is a reasonable place to look for the causes of the variations?
Cheap resistors?
Inadequate op amp?
Breadboard prototyping?
2x9V batteries as power supply?
other?

[darkfenriz]

The opamp has a huge input offset voltage (1.5mV) compared to what you're trying to amplify (0.01mV).
This is somewhat difficult area of ultra precision if you want to use this for measurement and you may struggle finding a suitable chip (chopper-stabilized anyone?)

Did you make the chart using AC or DC signals?

[nickb]

10uV is tiny and requires a great deal of skill of to design anything with any accuracy. The unexpected will catch you out e.g. metallic contacts produce thermocouple effects. So rule #1 is make the input signal as big as you can. Next use a part designed for the job e.g AD8230 Datasheet and Product Info | Analog Devices or go for chopper stabilised. You will need to think about how the tolerance of resistors affects the offset and gain and also consider the effects of temperature coefficients.

Go and find as many articles as you can on the subject.

[Dave H]

I'm trying to build an Instrumentation amplifier that would do a 10uV -> 10mV amplification. The motivation is to measure uA currents on a small enough shunt resistor (1-10Ohm).

To measure uA currents with high accuracy you could use an op-amp current to voltage converter. Use a low input bias current op-amp and a 1k feedback resistor to convert 10uA to 10mV or 100k to convert 10uA to 1V (output is -Iin*Rf)

[J M Fahey]

I'm trying to build an Instrumentation amplifier that would do a 10uV -> 10mV amplification. The motivation is to measure uA currents on a small enough shunt resistor (1-10Ohm).

For a proof of concept, I've built a circuit like the one below with a distinction that I'm powering it with two 9V batteries and am using LM2902Nas the op amps and Rg is somewhat different. It has much less that the required 1000x gain.

With due respect, I think you are trying to bite more than youn cann chew, you chose a very ambitious goal for your first Instrumentation project.
* trying to measure 10uV (is it full scale measurement?) with any precision means your acceptable error goes from 1uV (quite coarse 10% full scale) to 0.1uV for anything approaching a cheap $10 multimeter ... a TALL mountain to climb indeed.

* you are showing us a classic textbook example, dated by definition.
And using a still made but practically obsolete Op Amp, with very limited performance (and that being kind).
WILL work reasonably well, specially as a Classroom example, but donīt expect much, typical gain quoted is around 10X so expecting 1000X is stretching it way beyond reasonable.
No mention is made about input values , but I bet they are WAY higher than 10uV.

However, when measuring the results for different input voltages, the gain seems to vary:

I was expecting it to be constant.

May be wrong but rather than straight gain, a unit less ratio, your graph seems to be labelled in Voltage values.
The horizontal scale seems to go from 5mV (5.00E-03 meaning 5.00 "voltage" * "0.001") to 45mV , in 10mV steps, and the vertical one is again labeled in "voltage" but also as "gain" , which has no units.
In fact vertical scale shows what might be output voltage , going from 0.00E+00 (0 V) to 15V (1.5V * 10), and the blue line alone should indicate "gain" (top right side of image says so) , yet vertical scale is also labelled "gain"(left side of image also says so) ..... WTF?

Not sure how you achieved these results either, are they measured/calculated/simulated?

What is a reasonable place to look for the causes of the variations?
Cheap resistors?

You mean low precision? ... probably.

Inadequate op amp?

Definitely.

Breadboard prototyping?

Not if you are working carefully, in a proper environment, free of interference, etc.

2x9V batteries as power supply?

Should not influence results, but classic Lab work typically uses high precision regulated supplies, if anything because they are available.
2 x 9V batteries is fine but hints at "garage/kitchen table experimenting" .
Not blaming you, we all do such things everyday, just not testing such critical circuits (and thatīs an understatement).

I suggest you use a modern, Instrumentation specific Op Amp, and a modern circuit, with precision parts, such as shown in:
LT1167 - Single Resistor Gain Programmable, Precision Instrumentation Amplifier - Linear Technology

Notice the example shown uses:
* a precision modern high quality Op Amp
* a precision voltage reference
* offset adjust

and even so, it claims input offset voltage 10uV to 60uV , higher than what you are trying to measure, while as I said above, offset voltage and errors in general should be from 10 to 100 times smaller than what you are trying bto measure.

[Dave H]

The motivation is to measure uA currents on a small enough shunt resistor (1-10Ohm).

Why make life difficult by using such a low value shunt to measure uA? Won't the circuit under test tolerate a higher voltage drop across the shunt? Can you post a schematic?

If the aim is to measure very small currents with minimum voltage drop then the circuit I posted will do it. It can measure fA (with a suitable op-amp and layout) with only the op-amp's input offset voltage drop.