Use the Miller effect, which originally described undesired impedance, in a novel way to create a variable gain circuit.
The Miller effect, first reported a century ago, causes impedances connected between input and output of a voltage amplifier to be reflected in the amplifier input impedance, proportionately scaled by amplifier gain. Although first seen as only an undesired parasitic-capacitance multiplier limiting bandwidth and stability, the Miller effect has also been incorporated into useful topologies, like analog oscilloscope timebase integrators.
This Design Idea suggests another cool way to use it, based on the fact that if amplifier gain (A) is made variable, then the reflected Miller impedance (Zm) or susceptance (Ym) will be too (Figure 1).
Note in particular the interesting effect of allowing the range of gain factor A to include A = +1 and thus (1 – A) = (1 – 1) = 0, which causes Zm (i.e., Lm or Rm) to go (theoretically) to infinity and Ym (Cm) to zero.
A practical realization of the Miller effect component synthesis circuit is suggested in Figure 2.
Selection of the gain-setting potentiometer (simple analog trimmer, precision pot with turns-counting dial, or the illustrated digital pot) reflects application requirements. Selection of A1 and A2 accommodates needed bandwidth, voltage compliance, and current drive capability, while the R2/R1 ratio determines gain adjustment range: (+1 to -R2/R1). For example, given the components shown (e.g., 10-bit resolution AD5292-20 digital pot), R1 = R2, Yref = 0.5uF reference capacitor, and jumper J1 connected to ground, Ym can be synthesized from ~zero (single digit pF) to 1.0uF in 1k steps of ~1nF (actually 977pF):
The resulting circuit is useful in prototyping, testing, post-production trimming, tuning, and calibration.
However, the reader will have noticed an obvious limitation. The synthesized reactance has only one active terminal, with the other implicitly grounded. This is problematic in many potential applications, but when access to both terminals is required, a solution exists (Figure 3).