Merge #205

205: pll: refine gains r=jordens a=jordens

this decouples frequency and phase gain

Co-authored-by: Robert Jördens <rj@quartiq.de>

dsp/src/pll.rs

@@ -10,12 +10,14 @@ use serde::{Deserialize, Serialize};
 /// stable for any gain (1 <= shift <= 30). It has a single parameter that determines the loop
 /// bandwidth in octave steps. The gain can be changed freely between updates.
 ///
-/// The frequency and phase settling time constants for an (any) frequency jump are `1 << shift`
+/// The frequency and phase settling time constants for a frequency/phase jump are `1 << shift`
 /// update cycles. The loop bandwidth is about `1/(2*pi*(1 << shift))` in units of the sample rate.
+/// While the phase is being settled within one turn, there is a typically very small frequency
+/// overshoot.
 ///
 /// All math is naturally wrapping 32 bit integer. Phase and frequency are understood modulo that
 /// overflow in the first Nyquist zone. Expressing the IIR equations in other ways (e.g. single
-/// (T)-DF-{I,II} biquad/IIR) would break on overflow.
+/// (T)-DF-{I,II} biquad/IIR) would break on overflow (i.e. every cycle).
 ///
 /// There are no floating point rounding errors here. But there is integer quantization/truncation
 /// error of the `shift` lowest bits leading to a phase offset for very low gains. Truncation
@@ -23,8 +25,8 @@ use serde::{Deserialize, Serialize};
 /// efficiently by dithering.
 ///
 /// This PLL does not unwrap phase slips during lock acquisition. This can and should be
-/// implemented elsewhere by (down) scaling and then unwrapping the input phase and (up) scaling
-/// and wrapping output phase and frequency. This affects dynamic range accordingly.
+/// implemented elsewhere by unwrapping and scaling the input phase and un-scaling
+/// and wrapping output phase and frequency. This affects dynamic range, gain, and noise accordingly.
 ///
 /// The extension to I^3,I^2,I behavior to track chirps phase-accurately or to i64 data to
 /// increase resolution for extremely narrowband applications is obvious.
@@ -39,25 +41,39 @@ pub struct PLL {
 }
 
 impl PLL {
-    /// Update the PLL with a new phase sample.
+    /// Update the PLL with a new phase sample. This needs to be called (sampled) periodically.
+    /// The signal's phase/frequency is reconstructed relative to the sampling period.
     ///
     /// Args:
     /// * `input`: New input phase sample.
-    /// * `shift`: Error scaling. The frequency gain per update is `1/(1 << shift)`. The phase gain
-    ///   is always twice the frequency gain.
+    /// * `shift_frequency`: Frequency error scaling. The frequency gain per update is
+    ///   `1/(1 << shift_frequency)`.
+    /// * `shift_phase`: Phase error scaling. The phase gain is `1/(1 << shift_phase)`
+    ///   per update. A good value is typically `shift_frequency - 1`.
     ///
     /// Returns:
     /// A tuple of instantaneous phase and frequency (the current phase increment).
-    pub fn update(&mut self, x: i32, shift: u8) -> (i32, i32) {
-        debug_assert!((1..=30).contains(&shift));
-        let bias = 1i32 << shift;
+    pub fn update(
+        &mut self,
+        x: i32,
+        shift_frequency: u8,
+        shift_phase: u8,
+    ) -> (i32, i32) {
+        debug_assert!((1..=30).contains(&shift_frequency));
+        debug_assert!((1..=30).contains(&shift_phase));
         let e = x.wrapping_sub(self.f);
         self.f = self.f.wrapping_add(
-            (bias >> 1).wrapping_add(e).wrapping_sub(self.x) >> shift,
+            (1i32 << (shift_frequency - 1))
+                .wrapping_add(e)
+                .wrapping_sub(self.x)
+                >> shift_frequency,
         );
         self.x = x;
         let f = self.f.wrapping_add(
-            bias.wrapping_add(e).wrapping_sub(self.y) >> (shift - 1),
+            (1i32 << (shift_phase - 1))
+                .wrapping_add(e)
+                .wrapping_sub(self.y)
+                >> shift_phase,
         );
         self.y = self.y.wrapping_add(f);
         (self.y, f)
@@ -70,8 +86,8 @@ mod tests {
     #[test]
     fn mini() {
         let mut p = PLL::default();
-        let (y, f) = p.update(0x10000, 10);
-        assert_eq!(y, 0xc2);
+        let (y, f) = p.update(0x10000, 8, 4);
+        assert_eq!(y, 0x1100);
         assert_eq!(f, y);
     }
 
@@ -79,17 +95,17 @@ mod tests {
     fn converge() {
         let mut p = PLL::default();
         let f0 = 0x71f63049_i32;
-        let shift = 10;
-        let n = 31 << shift + 2;
+        let shift = (10, 9);
+        let n = 31 << shift.0 + 2;
         let mut x = 0i32;
         for i in 0..n {
             x = x.wrapping_add(f0);
-            let (y, f) = p.update(x, shift);
+            let (y, f) = p.update(x, shift.0, shift.1);
             if i > n / 4 {
                 assert_eq!(f.wrapping_sub(f0).abs() <= 1, true);
             }
             if i > n / 2 {
-                // The remaining error is removed by dithering.
+                // The remaining error would be removed by dithering.
                 assert_eq!(y.wrapping_sub(x).abs() < 1 << 18, true);
             }
         }