I'm trying to simulate the parasitic insensitive switched capacitor circuit shown here: enter image description here

I am using cadence virtuoso. My input is a sine wave of 1 V amplitude, 0 DC and 1 kHz frequency. I am using non-overlapping clocks with period 1us and pulse width 0.4 us. Both caps are 1pF. The schematic is shown below:

enter image description here

The opamp is generated using a vcvs with open-loop gain 1e20.

However, the output of my integrator seems to have a huge unexpected gain, instead of the predicted gain of 1. (Check out the swing in the next image). It also has some strange offset.

enter image description here

Does anyone know where I am going wrong with my simulation? Thanks.

  • \$\begingroup\$ Which component is the VCVS? \$\endgroup\$
    – Voltage Spike
    Jul 13, 2019 at 22:07
  • \$\begingroup\$ Voltage controlled voltage source from the analoglib. It can act as an opamp if your gain is very high. \$\endgroup\$ Jul 15, 2019 at 2:10
  • 1
    \$\begingroup\$ I know what a VCVS is, which designator is the VCVS? Are you talking about the opamp? I've never heard of an VCVS having open loop gain \$\endgroup\$
    – Voltage Spike
    Jul 15, 2019 at 4:30
  • \$\begingroup\$ Try limiting the bandwidth of your opamp \$\endgroup\$
    – Mike
    Jul 15, 2021 at 0:00

2 Answers 2


The gain of the integrator is related to the clock frequency. Notice, \$K \approx C_1/(C_2T_{clk})\$ . (see David Johns, Ken Martin, 1st ed. 403).

enter image description here

I put together a simulation in LTspice. The concept should be the same in Cadence. Notice with a fixed input frequency, as the sampling clock frequency gets higher and higher relative to the fixed input signal frequency, the relative clock period is getting smaller and the gain is getting larger, as the equation predicts.

Also, using a vcvs allows for the gain to grow without boundaries (it would saturate to rails with more realistic amplifier). Two possible ways to adjust this are lower the oversampling ratio, and or adjust capacitance ratio. Also, the vcvs output can't swing below 0 like the input without additional offset applied (like subtraction).

edit. Didn't notice this was two years old, but it's a useful concept for anyone learning about switched capacitor integrators. And not obvious.

  • \$\begingroup\$ Also 10^20 is too high for VCVs gain -- try 10^6, or even 10^3 to start debugging. 10^20 will bring in additional convergence issues in SPICE not related to debugging the circuit because it exceeds the resolution of even double precision floating point numbers. \$\endgroup\$
    – jp314
    Nov 15, 2021 at 4:39

Some things to try:

To solve the offset problem try putting a large resistor to ground on the input of the VCVS, like 1e12 or larger. That way the simulation can find the DC operating point on startup.

To solve the gain issue, use an actual integrator with resistors enter image description here


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.