INA118: sensing current in the 0-1uA range

Albero,

The issue is with your gain calculation.  It is: G=1+50k/Rg.  Thus, it is not 299 as you show in your Vcm vs Vout calulcator but rather G=1+50k/522=96.8 or G=981 (Rg=51ohm) - see below.  Also, in your Vcm vs Vout graph you use Vref=-5V but in you schematic Vref=0 - I used the latter.

For gain of 96.8 and Vref=0, simulated Vout is 43.7mV (close to 45mV calculated without including offset).

For gain of 981 and Vref=0, simulated Vout is 444mV (close to 456mV calculated without including offset) - see below.

Dear Marek,

thank you for your reply. In my tests, I used Rgain = 51 Ohm, so the Gain = 1 + 50k/51= 981.

In your formula, you add Vos (50uV) to (V+ - V-), then you multiply by the gain. By doing that, I nearly obtain the value neasured by the

Keithley SMU. How can I cancel/moderate the Vos effect on Vout ? Or can you suggest me a different INA with lower Vos (50uV seems to be

already a very low Vos) ?  Does a FET-input INA give a more accurate output ?

My best,

Alberto

Alberto,

The input voltage offset of INA118 has a Normal Gaussian distribution and thus may vary unit to unit with its limits as defined in the datasheet below.  For this reason, in order to take a precise measurement, you may NOT assume 50uV offset or -50uV - you must measure it.

Since you already calibrate the system, you may add a switch across R2 shunt resistor to calibrate the offset.  Shorting R2 will result in a zero differential input voltage and thus you may then use the output reading to back calculate the actual input offset voltage, Vos - see below.

In the example below with Vref=0, Vos=Vout/Gain = -12.524mV/981 = -12.7uV

Zero drift instrumentation amplifiers, like INA188, would have the lowest input voltage offset and offset drift BUT to get the most precise measurement calibration is the way to go as descibed above.