Skip to main content

Why gain enhancement is needed in Sallen-Key biquad

Short answer

In Sallen-Key biquad, Q and gain K are interdependent, so passband gain is fixed once Q is chosen. Gain enhancement increases overall gain without altering ωo and Q: set op-amp gain μ larger than design K, and return feedback capacitor C1 to a resistive divider tap such that the tap voltage is K·VB. Then overall gain becomes μ=K/β with β=Ry/(Rx+Ry).

Why gain enhancement is needed in Sallen and Key biquad? Explain the gain enhancement in Sallen and Key low pass biquad.

Why gain enhancement is needed

In the Sallen-Key low-pass biquad the gain KK is not a free design parameter: it also sets QQ.

ωoQ=1R1C1+1R2C1+1−KR2C2\frac{\omega_o}{Q} = \frac{1}{R_1C_1} + \frac{1}{R_2C_1} + \frac{1-K}{R_2C_2}
  • Equal-element design: K=3−1/QK = 3 - 1/Q, so KK always lies between 1 and 3 (e.g. 1.586 for Butterworth).
  • Unity-gain design (Design 2): K=1K = 1.

Once QQ is chosen, the passband gain is fixed. If the specification needs a larger gain (say 10), the circuit cannot give it directly, so the gain must be enhanced without changing ωo\omega_o and QQ. (If the gain is too large, it is reduced with an input voltage divider instead.)

Gain enhancement in the Sallen-Key low-pass biquad

The idea is to let the op-amp amplify by a larger factor μ\mu while the feedback capacitor C1C_1 still sees only the voltage KVBK V_B that the design needs.

  1. Set the op-amp's non-inverting gain to μ=1+RB/RA\mu = 1 + R_B/R_A, larger than the design value KK.
  2. Connect a resistive divider RxR_x, RyR_y across the output. Return C1C_1 to the tap of this divider (not to VoV_o). The tap voltage is
Vt=βVo,β=RyRx+RyV_t = \beta V_o, \qquad \beta = \frac{R_y}{R_x + R_y}
  1. Choose β=K/μ\beta = K/\mu, so that Vt=βμVB=KVBV_t = \beta\mu V_B = K V_B.
               C1
        +------||-----+ tap (beta*Vo)
        |             |
Vin -R1-+- R2 -+------|+\       Rx
               |      |  >--+--/\/--+
              === C2 +|-/   |       |
               |     |      Vo     Ry
              GND  RA/RB           |
                   network        GND

Analysis: the node equations are the same as in the normal circuit, with VoV_o replaced by βVo=KVB\beta V_o = K V_B at C1C_1. So the denominator (and therefore ωo\omega_o and QQ) is unchanged:

VoVin=μ ωo2s2+(ωo/Q)s+ωo2,μ=Kβ\frac{V_o}{V_{in}} = \frac{\mu\,\omega_o^2}{s^2 + (\omega_o/Q)s + \omega_o^2}, \qquad \mu = \frac{K}{\beta}

The passband gain rises from KK to K/βK/\beta.

Design steps:

  1. Design the normal circuit for the required ωo\omega_o and QQ, which gives KK.
  2. Required gain HH: set μ=H\mu = H, i.e. RB/RA=H−1R_B/R_A = H - 1.
  3. Divider ratio β=K/H\beta = K/H.
  4. Keep Rx∥RyR_x \parallel R_y much smaller than the reactance 1/(ωoC1)1/(\omega_o C_1) so the divider is not loaded by C1C_1.

Example: Butterworth equal-element design (K=1.586K = 1.586) needing H=10H = 10: RB/RA=9R_B/R_A = 9 and β=0.1586\beta = 0.1586.

A simpler alternative is to cascade the biquad with a non-inverting amplifier of gain H/KH/K, at the cost of one more op-amp.

✓ Answer checked by Claude Sonnet 5.5

All Filter Design chapter 6 questions →