> one low pass IIR filter is implemented on a fixed-pointed DSP,
> say, TI C5402.
>
> The IIR filter itself is stable.
> IF the input to the filter is one constant value (DC signal ),
> eg, X input= 1000;
>
> so, afte the stabilization, the output from the IIR filter is
> also 1000? or some value less than 1000?
If H(z) is the filter's transfer function, the ideal steady-state DC
output is H(1) * 1000 which could be any finite value, in principle.
I guess you're making the unstated assumption that H(1) = 1?
> Actually, I just wonder that the output will be less than 1000
> because of the finite word length effects. I am a new comer, so
> help me. Thanks
Finite arithmetic effects can disturb the values of |H(z)| in either
direction. Consider the possibility that coefficient quantization
shifts a nominally stable high-Q pole right onto the unit circle. The
same may happen to a zero, and in either case the shift may be
inwards too.
Martin
--
Quidquid latine scriptum sit, altum viditur.
Reply by Simon●May 23, 20062006-05-23
one low pass IIR filter is implemented on a fixed-pointed DSP, say, TI
C5402.
The IIR filter itself is stable.
IF the input to the filter is one constant value (DC signal ), eg, X
input= 1000;
so, afte the stabilization, the output from the IIR filter is also
1000? or some value less than 1000?
Actually, I just wonder that the output will be less than 1000 because
of the finite word length effects. I am a new comer, so help me. Thanks