DC Differential amplification with gain

Dear, Analogy Device

I have a two resistors (R1 and R2, they are about 5000 ohm) in series with a square wave current passing through (250nA - 500nA). I am looking for differential amplifier to enhance the voltage signal across each of the resistor. After differentiating the voltage across each resistor with small gain, I want to subtract the two output signal from the two differential amplifiers.

I am looking at the AD524 or AD624. I don't know it is good for my approach since I need very high precised voltage signal. the output signal should have resolution down to 1nV with no gain. And the two differential amplifier should have the exact gain. Do you have any recommendation a amplifier has low noise, high gain accuracy, and high input impedance.



  • 0
    •  Analog Employees 
    on Sep 21, 2015 5:43 PM over 5 years ago

    Hi Jason,

    AD524 and AD624 might not be the suitable amplifiers for your application. The AD524 arequires an input bias current of 25 nA - 50 nA while the AD624 requires an input bias current of 15nA - 50 nA depending on the grade of the part, the amplifiers will draw too much current from the source. You need to use an amplifier with a small input bias current relative to the input source of 250 nA- 500 nA, so that most of the current will pass through the sense resistors. In addition to that, you need to consider the effect of the input offset and output offset voltages of the device since it will introduce dc offset error at the output of the amplifier. The output offset of the AD524 and AD624 is in few mV, and this will be a problem when you configured the amplifiers at low gain. At higher gains, errors contributed by the input will be multiplied and this error will be reflected at the output.

    Can I ask what is your gain requirements for G1 and G2? With this, we can provide an appropriate parts for your application.



  • HI Gunao

    G1 and G2 should be equal and in the range 100-500 v/v.

  • Do you have access to the clock that is generating the square wave current?

    If so, then you could use synchronous detection methods to eliminate a slew of imperfections in the amplifiers.


  • I do have the access to the clock of the square wave, do u mean using lock-in as synchronous detection to measure the resistance change?

  • Looks like your requirements are 1nV resolution with the voltage drop over the resistors being about 2.5mV max. That is over 21 bits of resolution. You won't be able to get there simply relying on gain matching.

    Some method of (self) calibration will be needed. Having a synchronous clock allows you to use switches to select different measurement nodes and use the same amplifier to avoid gain changes. You need to have time for the transients to settle and likely a lot of averaging to get your noise floor below 1nV.

    How are you generating the sample currents? Not exactly trivial to get to the required precision...


    Alternatively, is there a way to reconfigure the setup to form a complete Wheatstone bridge?...