Skip to main content

Impulse response of DT ideal low pass filter

Short answer

The impulse response of an ideal discrete-time lowpass filter with cutoff ω_c is h[n] = sin(ω_c n)/(π n) = (ω_c/π) sinc(ω_c n/π) for n ≠ 0, and h[0] = ω_c/π. It is non‑causal and not absolutely summable (unstable).

Derive formula to calculate the impulse response of ideal low pass filter in discrete time.

An ideal discrete-time low pass filter passes all frequencies below a cut-off ωc\omega_c with unit gain and blocks the rest. Its frequency response over one period (−π≤ω≤π-\pi\le\omega\le\pi) is

H(ejω)={1,∣ω∣≤ωc0,ωc<∣ω∣≤πH(e^{j\omega})=\begin{cases}1, & |\omega|\le\omega_c\\ 0, & \omega_c<|\omega|\le\pi\end{cases}

and it repeats every 2π2\pi.

        H(e^jw)
          1 +-----+
            |     |
 ----+------+-----+------+---> w
    -pi   -wc  0  wc     pi

Derivation

The impulse response is the inverse DTFT:

h[n]=12π∫−ππH(ejω)ejωn dω=12π∫−ωcωcejωn dω=12π[ejωnjn]−ωcωc=12π⋅ejωcn−e−jωcnjn=12π⋅2jsin⁡(ωcn)jn=sin⁡(ωcn)πn,n≠0\begin{aligned} h[n] &= \frac{1}{2\pi}\int_{-\pi}^{\pi}H(e^{j\omega})e^{j\omega n}\,d\omega = \frac{1}{2\pi}\int_{-\omega_c}^{\omega_c}e^{j\omega n}\,d\omega \\ &= \frac{1}{2\pi}\left[\frac{e^{j\omega n}}{jn}\right]_{-\omega_c}^{\omega_c} = \frac{1}{2\pi}\cdot\frac{e^{j\omega_c n}-e^{-j\omega_c n}}{jn} \\ &= \frac{1}{2\pi}\cdot\frac{2j\sin(\omega_c n)}{jn} = \frac{\sin(\omega_c n)}{\pi n}, \qquad n\ne 0 \end{aligned}

At n=0n=0:

h[0]=12π∫−ωcωcdω=ωcπh[0]=\frac{1}{2\pi}\int_{-\omega_c}^{\omega_c}d\omega=\frac{\omega_c}{\pi}

which is also the limit of sin⁡ωcnπn\frac{\sin\omega_c n}{\pi n} as n→0n\to0. So

h[n]=sin⁡(ωcn)πn=ωcπ sinc ⁣(ωcnπ)h[n]=\frac{\sin(\omega_c n)}{\pi n}=\frac{\omega_c}{\pi}\,\mathrm{sinc}\!\left(\frac{\omega_c n}{\pi}\right)

where sinc x=sin⁡πxπx\mathrm{sinc}\,x=\frac{\sin \pi x}{\pi x} (the normalised sinc).

Remarks

  • h[n]h[n] is a sampled sinc: maximum ωc/π\omega_c/\pi at n=0n=0, zero crossings where ωcn\omega_c n is a multiple of π\pi.
  • h[n]≠0h[n]\ne0 for n<0n<0, so the ideal LPF is non-causal.
  • ∑∣h[n]∣\sum|h[n]| diverges (decays only as 1/n1/n), so it is not BIBO stable. Hence it cannot be built exactly; practical filters truncate/window h[n]h[n] and delay it.
  • Example: for ωc=π/2\omega_c=\pi/2, h[n]=sin⁡(πn/2)πnh[n]=\frac{\sin(\pi n/2)}{\pi n}: h[0]=0.5h[0]=0.5, h[±1]=1/π≈0.318h[\pm1]=1/\pi\approx0.318, h[±2]=0h[\pm2]=0, h[±3]=−1/(3π)≈−0.106h[\pm3]=-1/(3\pi)\approx-0.106.

✓ Answer checked by Claude Opus 5.5 (high effort)

All Signals and Systems chapter 6 questions →