The circuit in Figure 1 illustrates a classic use of a long time-constant. In this case, it’s an attack/release circuit that might be found in an automatic gain control. Resistor R1 and capacitor C1 form the time constant for the attack while R2, pot R3, and C1 form the release time-constant. Notice that R3 is a 1M potentiometer. A pot this large can pose a purchasing problem since it may not be available in all tapers.
The pot is large because the value of C1 is limited to approximately 5 µF to hold down the charging current during the attack phase. If a release time-constant much longer than five seconds is required, the high resistance value needed for the pot becomes a more significant concern.
The circuit in Figure 2 implements one possible solution to the problem. Op amp U2B and Q1 form a controlled current sink to ground through sense resistor R9. Potentiometer R6 (any convenient value and taper) sets the release current while R5 and R7 establish the maximum and minimum currents, respectively. If the top end of R6 was connected to a fixed voltage, the circuit would be a simple adjustable current sink. This would form a linear, constant-current discharge of C3.
In many applications, however, a linear discharge isn’t appropriate. By connecting R6 (through R5) to the buffered voltage on C3, the discharge current decreases as C3 discharges—just like the exponential discharge in Figure 1. The resistance RΦ indicated in Figure 2 represents the approximate equivalent discharge resistance. The minimum effective value of RΦ can’t be any smaller than R9. A value of around 1k to 2k for R9 works well. At very low currents it may be better to make R9 larger—about 10k or so.
The attack time is given by R4*C3. The release time is given by RΦ*C3. The maximum effective resistance of RΦ is R9 * (R5 + R6 + R7)/R7; the minimum effective resistance is R9 * (R5 + R6 + R7)/(R7 + R6).
R8 and C2 provide stability for U2B. U2A needs to be a FET-input op amp, such as the LF353 or TL072, to avoid loading the time-constant capacitor C3. Almost any op amp will suffice for U2B. The transistor Q1 needs to have reasonably low leakage current to control the very low current involved.
In use, the circuit behaves well when the effective resistance is on the order of several megohms. It may become twitchy if RΦ is increased much over 10 MW. If a high effective resistance is required, a small NMOS or DMOS transistor (such as the 2N7000) can be substituted for Q1.