2.3 LTI Continuous-Time Systems 141which corresponds to the accumulation of values ofx(t)in a segment [t−T,t] divided by its
lengthT, or the average ofx(t)in [t−T,t]. Use the convolution integral to find the response of the
averager to a ramp.
Solution
To find the ramp response using the convolution integral we first needh(t). The impulse response
of an averager can be found by lettingx(t)=δ(t)andy(t)=h(t)or
h(t)=1
T
∫tt−Tδ(τ)dτIft<0 or ift−T>0 this integral is zero as in these two situationst=0, where the delta function
occurs, is not included in the integral limits. However, whent−T<0 andt>0, or 0<t<T, the
integral is 1 as the origint=0, whereδ(t)occurs, is included in this interval. Thus, the impulse
response of the analog averager is
h(t)={ 1
T^0 <t<T
0 otherwiseWe then have that the outputy(t), for a given inputx(t), is given by the convolution integral
y(t)=∫∞
−∞h(τ)x(t−τ)dτ=∫T
01
T
x(t−τ)dτwhich can be shown to equal the definition of the averager by a change of variable. Indeed, let
σ=t−τ, so whenτ=0 thenσ=t, and whenτ=Tthenσ=t−T. Moreover, we have that
dσ=−dτ. The above integral becomes
y(t)=−1
T
t∫−Ttx(σ)dσ=1
T
∫tt−Tx(σ)dσThus, we have that
y(t)=1
T
∫t0x(t−τ)dτ=1
T
∫tt−Tx(σ)dσ (2.20)If the input is a ramp,x(t)=tu(t), the ramp responseρ(t)is
ρ(t)=1
T
∫tt−Tx(σ)dσ=1
T
∫tt−Tσu(σ)dσ