i have designed a gm-C bandpass filter used in ir receiver module. in my design, C is integrated on-chip (f0=gm/C), so C will change with process variation (may be 20% ~ 30% variation). i want to calibrate the variation of C with process through varing gm (bias current of gm), but how to calibrate this in detail?
besides, in simulation, i can get the frequency response(amplitude vs frequency) of my bandpass filter (ac analysis), but for real chip, how to measure the frequency response of bandpass filter? how to add the stimulus and how to get the frequency response curve?
thanks all.

There are a lot of ways of tuning the time constants. One is to use a PLL as Dan Clement said. You can also use an integrator and adjust things based on the ramp. For measurements, you have to use buffers and calibrate out its response. See the paper below(Fig. 1c, Fig. 8b) and references [3,9].

Nagendra Krishnapura, Abhishek Agrawal, and Sameer Singh, "A High IIP3 Third Order Elliptic Filter with Current Efficient Feedforward Compensated Opamps," IEEE Transactions on Circuits and Systems II-Express Briefs, vol. 58, no. 4, pp. 205-209, April 2011.

besides, nrk1, can you give me a copy of the paper you said? i cannot get it from web, thanks.

I inherited the filter and PLL architecture from someone else so I did not have the luxury of doing it a better way...

Technically it's really more of a DLL...

I did a master slave tuning topology but I don't like the way it works. You can't tune out the transport delay of the filter and end up a systematic error between master and slave.

And since I used the Bram Nauta Gm cell, I had to tune using a regulator. The regulator makes the loop dynamics tricky.

But my simulations indicate a centered tolerance of two percent and that meets the needs for the system.

The filter is not sw cap as it has three programmable center frequencies up to 8MHz.