lowpass: robustify
This commit is contained in:
parent
8d68504026
commit
beeb43bf8b
@ -15,19 +15,20 @@ impl<N: ArrayLength<i32>> Lowpass<N> {
|
||||
///
|
||||
/// # Args
|
||||
/// * `x`: Input data
|
||||
/// * `k`: Log2 time constant
|
||||
/// * `k`: Log2 time constant, 1..32
|
||||
///
|
||||
/// # Return
|
||||
/// Filtered output y
|
||||
pub fn update(&mut self, x: i32, k: u8) -> i32 {
|
||||
debug_assert!((1..32).contains(&k));
|
||||
// This is an unrolled and optimized first-order IIR loop
|
||||
// that works for all possible time constants.
|
||||
// Note the zero(s) at Nyquist and the benign overflow (DF-I).
|
||||
// Note the zero(s) at Nyquist.
|
||||
let mut x0 = x;
|
||||
let mut x1 = self.xy[0];
|
||||
self.xy[0] = x0;
|
||||
for y1 in self.xy[1..].iter_mut() {
|
||||
x0 = *y1 + (((x0 >> 1) + (x1 >> 1) - *y1 + (1 << (k - 1))) >> k);
|
||||
x0 = *y1 + (((x0 >> 2) + (x1 >> 2) - (*y1 >> 1) + (1 << (k - 1))) >> k);
|
||||
x1 = *y1;
|
||||
*y1 = x0;
|
||||
}
|
||||
|
Loading…
Reference in New Issue
Block a user