Problems with DC analysis of integrator

Discussion in 'The Projects Forum' started by srtquery, Mar 10, 2015.

  1. srtquery

    Thread Starter New Member

    Mar 9, 2015
    I am working on circuit to convert micro ampere current into voltage. For this purpose, I am using an integrator circuit using LMC6001 as the op-amp. I want to convert DC current in the micro ampere range into voltage. My circuit output for AC analysis matches with the simulation results, where the input is a sine wave or a square wave. Unfortunately, I get no circuit output for DC analysis, when the SPICE simulation software gives the correct output as calculated mathematically. I am using a feedback capacitor of 2nf and a resistor of 10M ohms as feedback resistance in the Integrator circuit. I am using these values of capacitance and resistor so that my integrator circuit can work with in a frequency range of 8-20 Hz.

    In DC analysis, for discharging the capacitor I am using a JFET (2N4117) that has to be given a clock pulse with a period of 50ms and the JFET will be switched off for 1msec. I want my capacitor to discharge in this 1ms and then start integrating once again for the next 49ms. For the rest of the 49ms, the JFET will not be working as it will be in open circuit. I am placing this JFET in such a way that the source and drain are across the feedback resistor and capacitor in the circuit. I am providing the gate with a clock pulse of amplitude of -5V for 49ms. This for the off condition and 0V for the rest of 1msec : this is the on condition for the JFET.

    The main problem in this circuit is that the switching action of the JFET works fine in AC analysis, but not in the DC analysis.
    I would be grateful for any help / suggestions.

    P.S : Do you think that it’s because of testing this micro ampere- range circuit on a breadboard ?
    Does any precaution needs to be taken for the circuit testing ? Any design changes that need to be done
  2. DickCappels


    Aug 21, 2008
    A schematic diagram, if you have it might help in the diagnosis in spite of your clear description. For example, is the source of the JFET connect to the opamp's inverting input or to the opamp's output?

    If you post your SPICE simulation file, perhaps somebody can find the problem with the DC simulation.

    Those plastic breadboards are good for testing LEDs and such, but now that you are in the microamp range the parasitics such as leakage current and capacitance to noise sources, including sources not on the board, could be a source of significant error.

    The 10 Meg resistor could be a source of error -100 na/volt out of the integrator, and since you are clamping (dumping) the integrator with the JFET the 10 Meg might not be needed, except perhaps to prevent latch-up if your opamp is susceptible to it.

    When using a JFET it is a good idea to put a resistor between the gate and source and drive the gate through a diode so that the drive signal can only reverse biase the gate, as forward bias would cause current to flow and that's probably a bad idea. The arrangement would be similar to that shown in the circuit below, except your JFET will be connected to the rest of the circuit in a different way. I used MOSFETs (2N7000) and transmission gates (74HC4066, 74HC4053) so I didn't have to worry about gate current.


    With a 49 ms integration time, you will probably suffer significant accuracy problems because the time constant of the 10 meg resistor and the .002 uF integrating capacitor is only 200 milliseconds; another reason to eliminate the 10 Meg resistor.

    When testing the real-world circuit, shielding is a good idea because things like charged objects moving near the circuit (like your hand or electrons in AC mains wiring) can affect the integrator's output. Locking your integration period to an integer multiple of the AC power can reduce the effects of the AC power line if you cannot shield adequately.

    The IDSS of the 2N4117 is very small. I would check during the test to confirm that it is actually discharging the .002 uf capacitor adequately (on paper it is fine, but close enough to warrant confirmation).
  3. #12


    Nov 30, 2010
    Microamps are certainly measurable. I've worked as low as 10 picoamps. :eek:
    Still, you have to arrange your physical parts for lowest leakage or the errors can be significant.
    (The idea of a simulator that doesn't work is possible.)
  4. srtquery

    Thread Starter New Member

    Mar 9, 2015

    Actually the circuit I have explained above is for 0.1uA(or 100nA) to a range of 15picoamps. The main problem is the switching of jfet(2N4117). We are not able to achieve it practically in our lab. Moreover, the off current of the jfet is in nanoamps. So, the circuit,s range is also decreasing.