2020-11-23 06:35:23 +08:00
|
|
|
use dsp::iir::IIR;
|
|
|
|
use dsp::lockin::{
|
|
|
|
decimate, magnitude_phase, Lockin, ADC_SAMPLE_BUFFER_SIZE,
|
|
|
|
DECIMATED_BUFFER_SIZE, TIMESTAMP_BUFFER_SIZE,
|
|
|
|
};
|
|
|
|
|
|
|
|
use std::f64::consts::PI;
|
|
|
|
use std::vec::Vec;
|
|
|
|
|
|
|
|
const ADC_MAX: f64 = 1.;
|
|
|
|
const ADC_MAX_COUNT: f64 = (1 << 15) as f64;
|
|
|
|
|
|
|
|
/// Single-frequency sinusoid.
|
|
|
|
#[derive(Copy, Clone)]
|
|
|
|
struct PureSine {
|
|
|
|
// Frequency (in Hz).
|
|
|
|
frequency: f64,
|
|
|
|
// Amplitude in dBFS (decibels relative to full-scale). A 16-bit
|
|
|
|
// ADC has a minimum dBFS for each sample of -90.
|
|
|
|
amplitude_dbfs: f64,
|
|
|
|
// Phase offset (in radians).
|
|
|
|
phase_offset: f64,
|
|
|
|
}
|
|
|
|
|
|
|
|
/// Convert a dBFS voltage ratio to a linear ratio.
|
|
|
|
///
|
|
|
|
/// # Arguments
|
|
|
|
///
|
|
|
|
/// * `dbfs` - dB ratio relative to full scale.
|
|
|
|
///
|
|
|
|
/// # Returns
|
|
|
|
///
|
|
|
|
/// Linear value.
|
|
|
|
fn linear(dbfs: f64) -> f64 {
|
|
|
|
let base = 10_f64;
|
|
|
|
ADC_MAX * base.powf(dbfs / 20.)
|
|
|
|
}
|
|
|
|
|
|
|
|
/// Convert a linear voltage ratio to a dBFS ratio.
|
|
|
|
///
|
|
|
|
/// # Arguments
|
|
|
|
///
|
|
|
|
/// * `linear` - Linear voltage ratio.
|
|
|
|
///
|
|
|
|
/// # Returns
|
|
|
|
///
|
|
|
|
/// dBFS value.
|
|
|
|
fn dbfs(linear: f64) -> f64 {
|
|
|
|
20. * (linear / ADC_MAX).log10()
|
|
|
|
}
|
|
|
|
|
|
|
|
/// Convert a real ADC input value in the range `-ADC_MAX` to
|
|
|
|
/// `+ADC_MAX` to an equivalent 16-bit ADC sampled value. This models
|
|
|
|
/// the ideal ADC transfer function.
|
|
|
|
///
|
|
|
|
/// # Arguments
|
|
|
|
///
|
|
|
|
/// * `x` - Real ADC input value.
|
|
|
|
///
|
|
|
|
/// # Returns
|
|
|
|
///
|
|
|
|
/// Sampled ADC value.
|
|
|
|
fn real_to_adc_sample(x: f64) -> i16 {
|
|
|
|
let max: i32 = i16::MAX as i32;
|
|
|
|
let min: i32 = i16::MIN as i32;
|
|
|
|
|
|
|
|
let xi: i32 = (x / ADC_MAX * ADC_MAX_COUNT) as i32;
|
|
|
|
|
|
|
|
// It's difficult to characterize the correct output result when
|
|
|
|
// the inputs are clipped, so panic instead.
|
|
|
|
if xi > max {
|
|
|
|
panic!("Input clipped to maximum, result is unlikely to be correct.");
|
|
|
|
} else if xi < min {
|
|
|
|
panic!("Input clipped to minimum, result is unlikely to be correct.");
|
|
|
|
}
|
|
|
|
|
|
|
|
xi as i16
|
|
|
|
}
|
|
|
|
|
|
|
|
/// Generate `ADC_SAMPLE_BUFFER_SIZE` values of an ADC-sampled signal
|
|
|
|
/// starting at `timestamp_start`.
|
|
|
|
///
|
|
|
|
/// # Arguments
|
|
|
|
///
|
|
|
|
/// * `pure_signals` - Pure sinusoidal components of the ADC-sampled
|
|
|
|
/// signal.
|
|
|
|
/// * `timestamp_start` - Starting time of ADC-sampled signal in terms
|
|
|
|
/// of the internal clock count.
|
|
|
|
/// * `internal_frequency` - Internal clock frequency (in Hz).
|
|
|
|
/// * `adc_frequency` - ADC sampling frequency (in Hz).
|
|
|
|
///
|
|
|
|
/// # Returns
|
|
|
|
///
|
|
|
|
/// The sampled signal at the ADC input.
|
|
|
|
fn adc_sampled_signal(
|
|
|
|
pure_signals: &Vec<PureSine>,
|
|
|
|
timestamp_start: u64,
|
|
|
|
internal_frequency: f64,
|
|
|
|
adc_frequency: f64,
|
|
|
|
) -> [i16; ADC_SAMPLE_BUFFER_SIZE] {
|
|
|
|
// amplitude of each pure signal
|
|
|
|
let mut amplitude: Vec<f64> = Vec::<f64>::new();
|
|
|
|
// initial phase value for each pure signal
|
|
|
|
let mut initial_phase: Vec<f64> = Vec::<f64>::new();
|
|
|
|
// phase increment at each ADC sample for each pure signal
|
|
|
|
let mut phase_increment: Vec<f64> = Vec::<f64>::new();
|
|
|
|
let adc_period = internal_frequency / adc_frequency;
|
|
|
|
|
|
|
|
// For each pure sinusoid, compute the amplitude, phase
|
|
|
|
// corresponding to the first ADC sample, and phase increment for
|
|
|
|
// each subsequent ADC sample.
|
|
|
|
for pure_signal in pure_signals.iter() {
|
|
|
|
let signal_period = internal_frequency / pure_signal.frequency;
|
|
|
|
let phase_offset_count =
|
|
|
|
pure_signal.phase_offset / (2. * PI) * signal_period;
|
|
|
|
let initial_phase_count =
|
|
|
|
(phase_offset_count + timestamp_start as f64) % signal_period;
|
|
|
|
|
|
|
|
amplitude.push(linear(pure_signal.amplitude_dbfs));
|
|
|
|
initial_phase.push(2. * PI * initial_phase_count / signal_period);
|
|
|
|
phase_increment.push(2. * PI * adc_period / signal_period);
|
|
|
|
}
|
|
|
|
|
|
|
|
// Compute the input signal corresponding to each ADC sample by
|
|
|
|
// summing the contributions from each pure sinusoid.
|
|
|
|
let mut signal: [i16; ADC_SAMPLE_BUFFER_SIZE] = [0; ADC_SAMPLE_BUFFER_SIZE];
|
|
|
|
signal.iter_mut().enumerate().for_each(|(n, s)| {
|
|
|
|
*s = real_to_adc_sample(
|
|
|
|
amplitude
|
|
|
|
.iter()
|
|
|
|
.zip(initial_phase.iter())
|
|
|
|
.zip(phase_increment.iter())
|
|
|
|
.fold(0., |acc, ((a, phi), theta)| {
|
|
|
|
acc + a * (phi + theta * n as f64).sin()
|
|
|
|
}),
|
|
|
|
);
|
|
|
|
});
|
|
|
|
|
|
|
|
signal
|
|
|
|
}
|
|
|
|
|
|
|
|
/// Reference clock timestamp values in one ADC batch period starting
|
|
|
|
/// at `timestamp_start`. Also returns the number of valid timestamps.
|
|
|
|
///
|
|
|
|
/// # Arguments
|
|
|
|
///
|
|
|
|
/// * `reference_frequency` - External reference signal frequency (in
|
|
|
|
/// Hz).
|
|
|
|
/// * `timestamp_start` - Start time in terms of the internal clock
|
|
|
|
/// count. This is the start time of the current processing sequence
|
|
|
|
/// (i.e., for the current `ADC_SAMPLE_BUFFER_SIZE` ADC samples).
|
|
|
|
/// * `timestamp_stop` - Stop time in terms of the internal clock
|
|
|
|
/// count.
|
|
|
|
/// * `internal_frequency` - Internal clock frequency (in Hz).
|
|
|
|
///
|
|
|
|
/// # Returns
|
|
|
|
///
|
|
|
|
/// Tuple consisting of the number of valid timestamps in the ADC
|
|
|
|
/// batch period, followed by an array of the timestamp values.
|
|
|
|
fn adc_batch_timestamps(
|
|
|
|
reference_frequency: f64,
|
|
|
|
timestamp_start: u64,
|
|
|
|
timestamp_stop: u64,
|
|
|
|
internal_frequency: f64,
|
|
|
|
) -> (usize, [u16; TIMESTAMP_BUFFER_SIZE]) {
|
|
|
|
let reference_period = internal_frequency / reference_frequency;
|
|
|
|
let start_count = timestamp_start as f64 % reference_period;
|
|
|
|
let mut valid_timestamps: usize = 0;
|
|
|
|
let mut timestamps: [u16; TIMESTAMP_BUFFER_SIZE] =
|
|
|
|
[0; TIMESTAMP_BUFFER_SIZE];
|
|
|
|
|
|
|
|
let mut timestamp = (reference_period - start_count) % reference_period;
|
|
|
|
while timestamp < (timestamp_stop - timestamp_start) as f64 {
|
|
|
|
timestamps[valid_timestamps] = timestamp as u16;
|
|
|
|
timestamp += reference_period;
|
|
|
|
valid_timestamps += 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
(valid_timestamps, timestamps)
|
|
|
|
}
|
|
|
|
|
|
|
|
/// Lowpass biquad filter using cutoff and sampling frequencies.
|
|
|
|
/// Taken from:
|
|
|
|
/// https://webaudio.github.io/Audio-EQ-Cookbook/audio-eq-cookbook.html
|
|
|
|
///
|
|
|
|
/// # Arguments
|
|
|
|
///
|
|
|
|
/// * `corner_frequency` - Corner frequency, or 3dB cutoff frequency
|
|
|
|
/// (in Hz).
|
|
|
|
/// * `sampling_frequency` - Sampling frequency (in Hz).
|
|
|
|
///
|
|
|
|
/// # Returns
|
|
|
|
///
|
|
|
|
/// 2nd-order IIR filter coefficients in the form [b0,b1,b2,a1,a2]. a0
|
|
|
|
/// is set to -1.
|
|
|
|
fn lowpass_iir_coefficients(
|
|
|
|
corner_frequency: f64,
|
|
|
|
sampling_frequency: f64,
|
|
|
|
) -> [f32; 5] {
|
|
|
|
let normalized_angular_frequency: f64 =
|
|
|
|
2. * PI * corner_frequency / sampling_frequency;
|
|
|
|
let quality_factor: f64 = 1. / 2f64.sqrt();
|
|
|
|
let alpha: f64 = normalized_angular_frequency.sin() / (2. * quality_factor);
|
|
|
|
// All b coefficients have been multiplied by a factor of 2 in
|
|
|
|
// comparison with the link above in order to set the passband
|
|
|
|
// gain to 2.
|
|
|
|
let mut b0: f64 = 1. - normalized_angular_frequency.cos();
|
|
|
|
let mut b1: f64 = 2. * (1. - normalized_angular_frequency.cos());
|
|
|
|
let mut b2: f64 = b0;
|
|
|
|
let a0: f64 = 1. + alpha;
|
|
|
|
let mut a1: f64 = -2. * normalized_angular_frequency.cos();
|
|
|
|
let mut a2: f64 = 1. - alpha;
|
|
|
|
b0 /= a0;
|
|
|
|
b1 /= a0;
|
|
|
|
b2 /= a0;
|
|
|
|
a1 /= -a0;
|
|
|
|
a2 /= -a0;
|
|
|
|
|
|
|
|
[b0 as f32, b1 as f32, b2 as f32, a1 as f32, a2 as f32]
|
|
|
|
}
|
|
|
|
|
|
|
|
/// Check that a measured value is within some tolerance of the actual
|
|
|
|
/// value. This allows setting both fixed and relative tolerances.
|
|
|
|
///
|
|
|
|
/// # Arguments
|
|
|
|
///
|
|
|
|
/// * `actual` - Actual value with respect to which the magnitude of
|
|
|
|
/// the relative tolerance is computed.
|
|
|
|
/// * `computed` - Computed value. This is compared with the actual
|
|
|
|
/// value, `actual`.
|
|
|
|
/// * `fixed_tolerance` - Fixed tolerance.
|
|
|
|
/// * `relative_tolerance` - Relative tolerance.
|
|
|
|
/// `relative_tolerance`*`actual` gives the total contribution of the
|
|
|
|
/// relative tolerance.
|
|
|
|
///
|
|
|
|
/// # Returns
|
|
|
|
///
|
|
|
|
/// `true` if the `actual` and `computed` values are within the
|
|
|
|
/// specified tolerance of one another, and `false` otherwise.
|
|
|
|
fn tolerance_check(
|
|
|
|
actual: f32,
|
|
|
|
computed: f32,
|
|
|
|
fixed_tolerance: f32,
|
|
|
|
relative_tolerance: f32,
|
|
|
|
) -> bool {
|
|
|
|
(actual - computed).abs()
|
|
|
|
< max_error(actual, fixed_tolerance, relative_tolerance)
|
|
|
|
}
|
|
|
|
|
|
|
|
/// Maximum acceptable error from an actual value given fixed and
|
|
|
|
/// relative tolerances.
|
|
|
|
///
|
|
|
|
/// # Arguments
|
|
|
|
///
|
|
|
|
/// * `actual` - Actual value with respect to which the magnitude of the
|
|
|
|
/// relative tolerance is computed.
|
|
|
|
/// * `fixed_tolerance` - Fixed tolerance.
|
|
|
|
/// * `relative_tolerance` - Relative tolerance.
|
|
|
|
/// `relative_tolerance`*`actual` gives the total contribution of the
|
|
|
|
/// relative tolerance.
|
|
|
|
///
|
|
|
|
/// # Returns
|
|
|
|
///
|
|
|
|
/// Maximum acceptable error.
|
|
|
|
fn max_error(
|
|
|
|
actual: f32,
|
|
|
|
fixed_tolerance: f32,
|
|
|
|
relative_tolerance: f32,
|
|
|
|
) -> f32 {
|
|
|
|
relative_tolerance * actual.abs() + fixed_tolerance
|
|
|
|
}
|
|
|
|
|
|
|
|
/// Total noise amplitude of the input signal after sampling by the
|
|
|
|
/// ADC. This computes an upper bound of the total noise amplitude,
|
|
|
|
/// rather than its actual value.
|
|
|
|
///
|
|
|
|
/// # Arguments
|
|
|
|
///
|
|
|
|
/// * `noise_inputs` - Noise sources at the ADC input.
|
|
|
|
/// * `demodulation_frequency` - Frequency of the demodulation signal
|
|
|
|
/// (in Hz).
|
|
|
|
/// * `corner_frequency` - Low-pass filter 3dB corner (cutoff)
|
|
|
|
/// frequency.
|
|
|
|
///
|
|
|
|
/// # Returns
|
|
|
|
///
|
|
|
|
/// Upper bound of the total amplitude of all noise sources.
|
|
|
|
fn sampled_noise_amplitude(
|
|
|
|
noise_inputs: &Vec<PureSine>,
|
|
|
|
demodulation_frequency: f64,
|
|
|
|
corner_frequency: f64,
|
|
|
|
) -> f64 {
|
|
|
|
// There is not a simple way to compute the amplitude of a
|
|
|
|
// superpostition of sinusoids with different frequencies and
|
|
|
|
// phases. Although we can compute the amplitude in special cases
|
|
|
|
// (e.g., two signals whose periods have a common multiple), these
|
|
|
|
// do not help us in the general case. However, we can say that
|
|
|
|
// the total amplitude will not be greater than the sum of the
|
|
|
|
// amplitudes of the individual noise sources. We treat this as an
|
|
|
|
// upper bound, and use it as an approximation of the actual
|
|
|
|
// amplitude.
|
|
|
|
|
|
|
|
let mut noise: f64 = noise_inputs
|
|
|
|
.iter()
|
|
|
|
.map(|n| {
|
|
|
|
// Noise inputs create an oscillation at the output, where the
|
|
|
|
// oscillation magnitude is determined by the strength of the
|
|
|
|
// noise and its attenuation (attenuation is determined by its
|
|
|
|
// proximity to the demodulation frequency and filter
|
|
|
|
// rolloff).
|
|
|
|
let octaves = ((n.frequency - demodulation_frequency).abs()
|
|
|
|
/ corner_frequency)
|
|
|
|
.log2();
|
|
|
|
// 2nd-order filter. Approximately 12dB/octave rolloff.
|
|
|
|
let attenuation = -2. * 20. * 2_f64.log10() * octaves;
|
|
|
|
linear(n.amplitude_dbfs + attenuation)
|
|
|
|
})
|
|
|
|
.sum();
|
|
|
|
|
|
|
|
// Add in 1/2 LSB for the maximum amplitude deviation resulting
|
|
|
|
// from quantization.
|
|
|
|
noise += 1. / ADC_MAX_COUNT / 2.;
|
|
|
|
|
|
|
|
noise
|
|
|
|
}
|
|
|
|
|
|
|
|
/// Compute the maximum effect of input noise on the lock-in magnitude
|
|
|
|
/// computation.
|
|
|
|
///
|
|
|
|
/// The maximum effect of noise on the magnitude computation is given
|
|
|
|
/// by:
|
|
|
|
///
|
|
|
|
/// | sqrt((I+n*sin(x))**2 + (Q+n*cos(x))**2) - sqrt(I**2 + Q**2) |
|
|
|
|
///
|
2020-11-24 08:16:21 +08:00
|
|
|
/// * I is the in-phase component of the portion of the input signal
|
|
|
|
/// with the same frequency as the demodulation signal.
|
2020-11-23 06:35:23 +08:00
|
|
|
/// * Q is the quadrature component.
|
|
|
|
/// * n is the total noise amplitude (from all contributions, after
|
|
|
|
/// attenuation from filtering).
|
2020-11-24 08:16:21 +08:00
|
|
|
/// * x is the phase of the demodulation signal.
|
2020-11-23 06:35:23 +08:00
|
|
|
///
|
|
|
|
/// We need to find the demodulation phase (x) that maximizes this
|
2020-11-24 08:16:21 +08:00
|
|
|
/// expression. We can ignore the absolute value operation by also
|
|
|
|
/// considering the expression minimum. The locations of the minimum
|
|
|
|
/// and maximum can be computed analytically by finding the value of x
|
|
|
|
/// when the derivative of this expression with respect to x is
|
|
|
|
/// 0. When we solve this equation, we find:
|
|
|
|
///
|
|
|
|
/// x = atan(I/Q)
|
|
|
|
///
|
|
|
|
/// It's worth noting that this solution is technically only valid
|
|
|
|
/// when cos(x)!=0 (i.e., x!=pi/2,-pi/2). However, this is not a
|
|
|
|
/// problem because we only get these values when Q=0. Rust correctly
|
|
|
|
/// computes atan(inf)=pi/2, which is precisely what we want because
|
|
|
|
/// x=pi/2 maximizes sin(x) and therefore also the noise effect.
|
|
|
|
///
|
|
|
|
/// The other maximum or minimum is pi radians away from this
|
|
|
|
/// value.
|
2020-11-23 06:35:23 +08:00
|
|
|
///
|
|
|
|
/// # Arguments
|
|
|
|
///
|
|
|
|
/// * `total_noise_amplitude` - Combined amplitude of all noise
|
|
|
|
/// sources sampled by the ADC.
|
|
|
|
/// * `in_phase_actual` - Value of the in-phase component if no noise
|
|
|
|
/// were present at the ADC input.
|
|
|
|
/// * `quadrature_actual` - Value of the quadrature component if no
|
|
|
|
/// noise were present at the ADC input.
|
|
|
|
/// * `desired_input_amplitude` - Amplitude of the desired input
|
|
|
|
/// signal. That is, the input signal component with the same
|
|
|
|
/// frequency as the demodulation signal.
|
|
|
|
///
|
|
|
|
/// # Returns
|
|
|
|
///
|
|
|
|
/// Approximation of the maximum effect on the magnitude computation
|
|
|
|
/// due to noise sources at the ADC input.
|
|
|
|
fn magnitude_noise(
|
|
|
|
total_noise_amplitude: f64,
|
|
|
|
in_phase_actual: f64,
|
|
|
|
quadrature_actual: f64,
|
|
|
|
desired_input_amplitude: f64,
|
|
|
|
) -> f64 {
|
|
|
|
// See function documentation for explanation.
|
|
|
|
let noise = |in_phase_delta: f64, quadrature_delta: f64| -> f64 {
|
|
|
|
(((in_phase_actual + in_phase_delta).powf(2.)
|
|
|
|
+ (quadrature_actual + quadrature_delta).powf(2.))
|
|
|
|
.sqrt()
|
|
|
|
- desired_input_amplitude)
|
|
|
|
.abs()
|
|
|
|
};
|
|
|
|
|
2020-11-24 08:16:21 +08:00
|
|
|
let phase = (in_phase_actual / quadrature_actual).atan();
|
|
|
|
let max_noise_1 = noise(
|
|
|
|
total_noise_amplitude * phase.sin(),
|
|
|
|
total_noise_amplitude * phase.cos(),
|
|
|
|
);
|
|
|
|
let max_noise_2 = noise(
|
|
|
|
total_noise_amplitude * (phase + PI).sin(),
|
|
|
|
total_noise_amplitude * (phase + PI).cos(),
|
|
|
|
);
|
2020-11-23 06:35:23 +08:00
|
|
|
|
2020-11-24 08:16:21 +08:00
|
|
|
max_noise_1.max(max_noise_2)
|
2020-11-23 06:35:23 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
/// Compute the maximum phase deviation from the correct value due to
|
|
|
|
/// the input noise sources.
|
|
|
|
///
|
|
|
|
/// The maximum effect of noise on the phase computation is given by:
|
|
|
|
///
|
|
|
|
/// | atan2(Q+n*cos(x), I+n*sin(x)) - atan2(Q, I) |
|
|
|
|
///
|
|
|
|
/// See `magnitude_noise` for an explanation of the terms in this
|
|
|
|
/// mathematical expression.
|
|
|
|
///
|
2020-11-24 08:16:21 +08:00
|
|
|
/// This expression is harder to compute analytically than the
|
|
|
|
/// expression in `magnitude_noise`. We could compute it numerically,
|
|
|
|
/// but that's expensive. However, we can use heuristics to try to
|
|
|
|
/// guess the values of x that will maximize the noise
|
|
|
|
/// effect. Intuitively, the difference will be largest when the
|
|
|
|
/// Y-argument of the atan2 function (Q+n*cos(x)) is pushed in the
|
|
|
|
/// opposite direction of the noise effect on the X-argument (i.e.,
|
|
|
|
/// cos(x) and sin(x) have different signs). We can use:
|
|
|
|
///
|
|
|
|
/// * sin(x)=+-1 (+- denotes plus or minus), cos(x)=0,
|
|
|
|
/// * sin(x)=0, cos(x)=+-1, and
|
|
|
|
/// * the value of x that maximizes |sin(x)-cos(x)| (when
|
|
|
|
/// sin(x)=1/sqrt(2) and cos(x)=-1/sqrt(2), or when the signs are
|
|
|
|
/// flipped)
|
|
|
|
///
|
|
|
|
/// The first choice addresses cases in which |I|>>|Q|, the second
|
|
|
|
/// choice addresses cases in which |Q|>>|I|, and the third choice
|
|
|
|
/// addresses cases in which |I|~|Q|. We can test all of these cases
|
|
|
|
/// as an approximation for the real maximum.
|
2020-11-23 06:35:23 +08:00
|
|
|
///
|
|
|
|
/// # Arguments
|
|
|
|
///
|
|
|
|
/// * `total_noise_amplitude` - Total amplitude of all input noise
|
|
|
|
/// sources.
|
|
|
|
/// * `in_phase_actual` - Value of the in-phase component if no noise
|
|
|
|
/// were present at the input.
|
|
|
|
/// * `quadrature_actual` - Value of the quadrature component if no
|
|
|
|
/// noise were present at the input.
|
|
|
|
///
|
|
|
|
/// # Returns
|
|
|
|
///
|
|
|
|
/// Approximation of the maximum effect on the phase computation due
|
|
|
|
/// to noise sources at the ADC input.
|
|
|
|
fn phase_noise(
|
|
|
|
total_noise_amplitude: f64,
|
|
|
|
in_phase_actual: f64,
|
|
|
|
quadrature_actual: f64,
|
|
|
|
) -> f64 {
|
|
|
|
// See function documentation for explanation.
|
|
|
|
let noise = |in_phase_delta: f64, quadrature_delta: f64| -> f64 {
|
|
|
|
((quadrature_actual + quadrature_delta)
|
|
|
|
.atan2(in_phase_actual + in_phase_delta)
|
|
|
|
- quadrature_actual.atan2(in_phase_actual))
|
|
|
|
.abs()
|
|
|
|
};
|
|
|
|
|
|
|
|
let mut max_noise: f64 = 0.;
|
|
|
|
for (in_phase_delta, quadrature_delta) in [
|
|
|
|
(
|
|
|
|
total_noise_amplitude / 2_f64.sqrt(),
|
|
|
|
total_noise_amplitude / -2_f64.sqrt(),
|
|
|
|
),
|
|
|
|
(
|
|
|
|
total_noise_amplitude / -2_f64.sqrt(),
|
|
|
|
total_noise_amplitude / 2_f64.sqrt(),
|
|
|
|
),
|
|
|
|
(total_noise_amplitude, 0.),
|
|
|
|
(-total_noise_amplitude, 0.),
|
|
|
|
(0., total_noise_amplitude),
|
|
|
|
(0., -total_noise_amplitude),
|
|
|
|
]
|
|
|
|
.iter()
|
|
|
|
{
|
|
|
|
max_noise = max_noise.max(noise(*in_phase_delta, *quadrature_delta));
|
|
|
|
}
|
|
|
|
|
|
|
|
max_noise
|
|
|
|
}
|
|
|
|
|
|
|
|
/// Lowpass filter test for in-phase/quadrature and magnitude/phase
|
|
|
|
/// computations.
|
|
|
|
///
|
|
|
|
/// This attempts to "intelligently" model acceptable tolerance ranges
|
|
|
|
/// for the measured in-phase, quadrature, magnitude and phase results
|
|
|
|
/// of lock-in processing for a typical low-pass filter
|
|
|
|
/// application. So, instead of testing whether the lock-in processing
|
|
|
|
/// extracts the true magnitude and phase (or in-phase and quadrature
|
|
|
|
/// components) of the input signal, it attempts to calculate what the
|
|
|
|
/// lock-in processing should compute given any set of input noise
|
|
|
|
/// sources. For example, if a noise source of sufficient strength
|
|
|
|
/// differs in frequency by 1kHz from the reference frequency and the
|
|
|
|
/// filter cutoff frequency is also 1kHz, testing if the lock-in
|
|
|
|
/// amplifier extracts the amplitude and phase of the input signal
|
|
|
|
/// whose frequency is equal to the demodulation frequency is doomed
|
|
|
|
/// to failure. Instead, this function tests whether the lock-in
|
|
|
|
/// correctly adheres to its actual transfer function, whether or not
|
|
|
|
/// it was given reasonable inputs. The logic for computing acceptable
|
|
|
|
/// tolerance ranges is performed in `sampled_noise_amplitude`,
|
|
|
|
/// `magnitude_noise`, and `phase_noise`.
|
|
|
|
///
|
|
|
|
/// # Arguments
|
|
|
|
///
|
|
|
|
/// * `internal_frequency` - Internal clock frequency (Hz). The
|
|
|
|
/// internal clock increments timestamp counter values used to
|
|
|
|
/// record the edges of the external reference.
|
|
|
|
/// * `adc_frequency` - ADC sampling frequency (in Hz).
|
|
|
|
/// * `reference_frequency` - External reference frequency (in Hz).
|
|
|
|
/// * `demodulation_phase_offset` - Phase offset applied to the
|
|
|
|
/// in-phase and quadrature demodulation signals.
|
|
|
|
/// * `harmonic` - Scaling factor for the demodulation
|
|
|
|
/// frequency. E.g., 2 would demodulate with the first harmonic of the
|
|
|
|
/// reference frequency.
|
|
|
|
/// * `corner_frequency` - Lowpass filter 3dB cutoff frequency.
|
|
|
|
/// * `desired_input` - `PureSine` giving the frequency, amplitude and
|
|
|
|
/// phase of the desired result.
|
|
|
|
/// * `noise_inputs` - Vector of `PureSine` for any noise inputs on top
|
|
|
|
/// of `desired_input`.
|
|
|
|
/// * `time_constant_factor` - Number of time constants after which
|
|
|
|
/// the output is considered valid.
|
|
|
|
/// * `tolerance` - Acceptable relative tolerance for the magnitude
|
|
|
|
/// and angle outputs. The outputs must remain within this tolerance
|
|
|
|
/// between `time_constant_factor` and `time_constant_factor+1` time
|
|
|
|
/// constants.
|
|
|
|
fn lowpass_test(
|
|
|
|
internal_frequency: f64,
|
|
|
|
adc_frequency: f64,
|
|
|
|
reference_frequency: f64,
|
|
|
|
demodulation_phase_offset: f64,
|
|
|
|
harmonic: u32,
|
|
|
|
corner_frequency: f64,
|
|
|
|
desired_input: PureSine,
|
|
|
|
noise_inputs: &mut Vec<PureSine>,
|
|
|
|
time_constant_factor: f64,
|
|
|
|
tolerance: f32,
|
|
|
|
) {
|
|
|
|
let mut lockin = Lockin::new(
|
|
|
|
demodulation_phase_offset as f32,
|
|
|
|
(internal_frequency / adc_frequency) as u32,
|
|
|
|
harmonic,
|
|
|
|
IIR {
|
|
|
|
ba: lowpass_iir_coefficients(corner_frequency, adc_frequency),
|
|
|
|
y_offset: 0.,
|
|
|
|
y_min: -ADC_MAX_COUNT as f32,
|
|
|
|
y_max: (ADC_MAX_COUNT - 1.) as f32,
|
|
|
|
},
|
|
|
|
);
|
|
|
|
|
|
|
|
let mut timestamp_start: u64 = 0;
|
|
|
|
let time_constant: f64 = 1. / (2. * PI * corner_frequency);
|
|
|
|
let samples =
|
|
|
|
(time_constant_factor * time_constant * adc_frequency) as usize;
|
|
|
|
// Ensure the result remains within tolerance for 1 time constant
|
|
|
|
// after `time_constant_factor` time constants.
|
|
|
|
let extra_samples = (time_constant * adc_frequency) as usize;
|
|
|
|
let sample_count: u64 = (internal_frequency / adc_frequency) as u64
|
|
|
|
* ADC_SAMPLE_BUFFER_SIZE as u64;
|
|
|
|
|
|
|
|
let effective_phase_offset =
|
|
|
|
desired_input.phase_offset - demodulation_phase_offset;
|
|
|
|
let in_phase_actual =
|
|
|
|
linear(desired_input.amplitude_dbfs) * effective_phase_offset.cos();
|
|
|
|
let quadrature_actual =
|
|
|
|
linear(desired_input.amplitude_dbfs) * effective_phase_offset.sin();
|
|
|
|
|
|
|
|
let total_noise_amplitude = sampled_noise_amplitude(
|
|
|
|
noise_inputs,
|
|
|
|
reference_frequency * harmonic as f64,
|
|
|
|
corner_frequency,
|
|
|
|
);
|
|
|
|
let total_magnitude_noise = magnitude_noise(
|
|
|
|
total_noise_amplitude,
|
|
|
|
in_phase_actual,
|
|
|
|
quadrature_actual,
|
|
|
|
linear(desired_input.amplitude_dbfs),
|
|
|
|
);
|
|
|
|
let total_phase_noise =
|
|
|
|
phase_noise(total_noise_amplitude, in_phase_actual, quadrature_actual);
|
|
|
|
|
|
|
|
let pure_signals = noise_inputs;
|
|
|
|
pure_signals.push(desired_input);
|
|
|
|
|
|
|
|
for n in 0..(samples + extra_samples) {
|
|
|
|
let signal: [i16; ADC_SAMPLE_BUFFER_SIZE] = adc_sampled_signal(
|
|
|
|
&pure_signals,
|
|
|
|
timestamp_start,
|
|
|
|
internal_frequency,
|
|
|
|
adc_frequency,
|
|
|
|
);
|
|
|
|
let (valid_timestamps, timestamps) = adc_batch_timestamps(
|
|
|
|
reference_frequency,
|
|
|
|
timestamp_start,
|
|
|
|
timestamp_start + sample_count - 1,
|
|
|
|
internal_frequency,
|
|
|
|
);
|
|
|
|
|
|
|
|
let mut in_phase: [f32; ADC_SAMPLE_BUFFER_SIZE];
|
|
|
|
let mut quadrature: [f32; ADC_SAMPLE_BUFFER_SIZE];
|
|
|
|
let lockin_demodulate =
|
|
|
|
lockin.demodulate(signal, timestamps, valid_timestamps as u16);
|
|
|
|
match lockin_demodulate {
|
2020-11-25 15:44:39 +08:00
|
|
|
Ok((i, q)) => {
|
|
|
|
in_phase = i;
|
|
|
|
quadrature = q;
|
2020-11-23 06:35:23 +08:00
|
|
|
}
|
2020-11-25 15:44:39 +08:00
|
|
|
Err(_) => {
|
2020-11-23 06:35:23 +08:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
lockin.filter(&mut in_phase, &mut quadrature);
|
|
|
|
let (in_phase_decimated, quadrature_decimated) =
|
|
|
|
decimate(in_phase, quadrature);
|
|
|
|
|
|
|
|
let mut magnitude_decimated = in_phase_decimated.clone();
|
|
|
|
let mut phase_decimated = quadrature_decimated.clone();
|
|
|
|
|
|
|
|
magnitude_phase(&mut magnitude_decimated, &mut phase_decimated);
|
|
|
|
|
|
|
|
// Ensure stable within tolerance for 1 time constant after
|
|
|
|
// `time_constant_factor`.
|
|
|
|
if n >= samples {
|
|
|
|
for k in 0..DECIMATED_BUFFER_SIZE {
|
|
|
|
let amplitude_normalized: f32 =
|
|
|
|
magnitude_decimated[k] / ADC_MAX_COUNT as f32;
|
|
|
|
assert!(
|
|
|
|
tolerance_check(linear(desired_input.amplitude_dbfs) as f32, amplitude_normalized, total_magnitude_noise as f32, tolerance),
|
|
|
|
"magnitude actual: {:.4} ({:.2} dBFS), magnitude computed: {:.4} ({:.2} dBFS), tolerance: {:.4}",
|
|
|
|
linear(desired_input.amplitude_dbfs),
|
|
|
|
desired_input.amplitude_dbfs,
|
|
|
|
amplitude_normalized,
|
|
|
|
dbfs(amplitude_normalized as f64),
|
|
|
|
max_error(linear(desired_input.amplitude_dbfs) as f32, total_magnitude_noise as f32, tolerance)
|
|
|
|
);
|
|
|
|
assert!(
|
|
|
|
tolerance_check(
|
|
|
|
effective_phase_offset as f32,
|
|
|
|
phase_decimated[k],
|
|
|
|
total_phase_noise as f32,
|
|
|
|
tolerance
|
|
|
|
),
|
|
|
|
"phase actual: {:.4}, phase computed: {:.4}, tolerance: {:.4}",
|
|
|
|
effective_phase_offset as f32,
|
|
|
|
phase_decimated[k],
|
|
|
|
max_error(
|
|
|
|
effective_phase_offset as f32,
|
|
|
|
total_phase_noise as f32,
|
|
|
|
tolerance
|
|
|
|
)
|
|
|
|
);
|
|
|
|
|
|
|
|
let in_phase_normalized: f32 =
|
|
|
|
in_phase_decimated[k] / ADC_MAX_COUNT as f32;
|
|
|
|
let quadrature_normalized: f32 =
|
|
|
|
quadrature_decimated[k] / ADC_MAX_COUNT as f32;
|
|
|
|
assert!(
|
|
|
|
tolerance_check(
|
|
|
|
in_phase_actual as f32,
|
|
|
|
in_phase_normalized,
|
|
|
|
total_noise_amplitude as f32,
|
|
|
|
tolerance
|
|
|
|
),
|
|
|
|
"in-phase actual: {:.4}, in-phase computed: {:.3}, tolerance: {:.4}",
|
|
|
|
in_phase_actual,
|
|
|
|
in_phase_normalized,
|
|
|
|
max_error(
|
|
|
|
in_phase_actual as f32,
|
|
|
|
total_noise_amplitude as f32,
|
|
|
|
tolerance
|
|
|
|
)
|
|
|
|
);
|
|
|
|
assert!(
|
|
|
|
tolerance_check(
|
|
|
|
quadrature_actual as f32,
|
|
|
|
quadrature_normalized,
|
|
|
|
total_noise_amplitude as f32,
|
|
|
|
tolerance
|
|
|
|
),
|
|
|
|
"quadrature actual: {:.4}, quadrature computed: {:.4}, tolerance: {:.4}",
|
|
|
|
quadrature_actual,
|
|
|
|
quadrature_normalized,
|
|
|
|
max_error(
|
|
|
|
quadrature_actual as f32,
|
|
|
|
total_noise_amplitude as f32,
|
|
|
|
tolerance
|
|
|
|
)
|
|
|
|
);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
timestamp_start += sample_count;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
fn lowpass() {
|
|
|
|
let internal_frequency: f64 = 100e6;
|
|
|
|
let adc_frequency: f64 = 500e3;
|
|
|
|
let signal_frequency: f64 = 100e3;
|
|
|
|
let harmonic: u32 = 1;
|
|
|
|
let corner_frequency: f64 = 1e3;
|
|
|
|
let demodulation_frequency: f64 = harmonic as f64 * signal_frequency;
|
|
|
|
let demodulation_phase_offset: f64 = 0.;
|
|
|
|
let time_constant_factor: f64 = 5.;
|
|
|
|
let tolerance: f32 = 1e-2;
|
|
|
|
|
|
|
|
lowpass_test(
|
|
|
|
internal_frequency,
|
|
|
|
adc_frequency,
|
|
|
|
signal_frequency,
|
|
|
|
demodulation_phase_offset,
|
|
|
|
harmonic,
|
|
|
|
corner_frequency,
|
|
|
|
PureSine {
|
|
|
|
frequency: demodulation_frequency,
|
|
|
|
amplitude_dbfs: -30.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
&mut vec![
|
|
|
|
PureSine {
|
|
|
|
frequency: 1.1 * demodulation_frequency,
|
|
|
|
amplitude_dbfs: -20.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
PureSine {
|
|
|
|
frequency: 0.9 * demodulation_frequency,
|
|
|
|
amplitude_dbfs: -20.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
],
|
|
|
|
time_constant_factor,
|
|
|
|
tolerance,
|
|
|
|
);
|
|
|
|
}
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
fn lowpass_demodulation_phase_offset_pi_2() {
|
|
|
|
let internal_frequency: f64 = 100e6;
|
|
|
|
let adc_frequency: f64 = 500e3;
|
|
|
|
let signal_frequency: f64 = 100e3;
|
|
|
|
let harmonic: u32 = 1;
|
|
|
|
let corner_frequency: f64 = 1e3;
|
|
|
|
let demodulation_frequency: f64 = harmonic as f64 * signal_frequency;
|
|
|
|
let demodulation_phase_offset: f64 = PI / 2.;
|
|
|
|
let time_constant_factor: f64 = 5.;
|
|
|
|
let tolerance: f32 = 1e-2;
|
|
|
|
|
|
|
|
lowpass_test(
|
|
|
|
internal_frequency,
|
|
|
|
adc_frequency,
|
|
|
|
signal_frequency,
|
|
|
|
demodulation_phase_offset,
|
|
|
|
harmonic,
|
|
|
|
corner_frequency,
|
|
|
|
PureSine {
|
|
|
|
frequency: demodulation_frequency,
|
|
|
|
amplitude_dbfs: -30.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
&mut vec![
|
|
|
|
PureSine {
|
|
|
|
frequency: 1.1 * demodulation_frequency,
|
|
|
|
amplitude_dbfs: -20.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
PureSine {
|
|
|
|
frequency: 0.9 * demodulation_frequency,
|
|
|
|
amplitude_dbfs: -20.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
],
|
|
|
|
time_constant_factor,
|
|
|
|
tolerance,
|
|
|
|
);
|
|
|
|
}
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
fn lowpass_phase_offset_pi_2() {
|
|
|
|
let internal_frequency: f64 = 100e6;
|
|
|
|
let adc_frequency: f64 = 500e3;
|
|
|
|
let signal_frequency: f64 = 100e3;
|
|
|
|
let harmonic: u32 = 1;
|
|
|
|
let corner_frequency: f64 = 1e3;
|
|
|
|
let demodulation_frequency: f64 = harmonic as f64 * signal_frequency;
|
|
|
|
let demodulation_phase_offset: f64 = 0.;
|
|
|
|
let time_constant_factor: f64 = 5.;
|
|
|
|
let tolerance: f32 = 1e-2;
|
|
|
|
|
|
|
|
lowpass_test(
|
|
|
|
internal_frequency,
|
|
|
|
adc_frequency,
|
|
|
|
signal_frequency,
|
|
|
|
demodulation_phase_offset,
|
|
|
|
harmonic,
|
|
|
|
corner_frequency,
|
|
|
|
PureSine {
|
|
|
|
frequency: demodulation_frequency,
|
|
|
|
amplitude_dbfs: -30.,
|
|
|
|
phase_offset: PI / 2.,
|
|
|
|
},
|
|
|
|
&mut vec![
|
|
|
|
PureSine {
|
|
|
|
frequency: 1.1 * demodulation_frequency,
|
|
|
|
amplitude_dbfs: -20.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
PureSine {
|
|
|
|
frequency: 0.9 * demodulation_frequency,
|
|
|
|
amplitude_dbfs: -20.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
],
|
|
|
|
time_constant_factor,
|
|
|
|
tolerance,
|
|
|
|
);
|
|
|
|
}
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
fn lowpass_fundamental_111e3_phase_offset_pi_4() {
|
|
|
|
let internal_frequency: f64 = 100e6;
|
|
|
|
let adc_frequency: f64 = 500e3;
|
|
|
|
let signal_frequency: f64 = 111e3;
|
|
|
|
let harmonic: u32 = 1;
|
|
|
|
let corner_frequency: f64 = 1e3;
|
|
|
|
let demodulation_frequency: f64 = harmonic as f64 * signal_frequency;
|
|
|
|
let demodulation_phase_offset: f64 = 0.;
|
|
|
|
let time_constant_factor: f64 = 5.;
|
|
|
|
let tolerance: f32 = 1e-2;
|
|
|
|
|
|
|
|
lowpass_test(
|
|
|
|
internal_frequency,
|
|
|
|
adc_frequency,
|
|
|
|
signal_frequency,
|
|
|
|
demodulation_phase_offset,
|
|
|
|
harmonic,
|
|
|
|
corner_frequency,
|
|
|
|
PureSine {
|
|
|
|
frequency: demodulation_frequency,
|
|
|
|
amplitude_dbfs: -30.,
|
|
|
|
phase_offset: PI / 4.,
|
|
|
|
},
|
|
|
|
&mut vec![
|
|
|
|
PureSine {
|
|
|
|
frequency: 1.1 * demodulation_frequency,
|
|
|
|
amplitude_dbfs: -20.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
PureSine {
|
|
|
|
frequency: 0.9 * demodulation_frequency,
|
|
|
|
amplitude_dbfs: -20.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
],
|
|
|
|
time_constant_factor,
|
|
|
|
tolerance,
|
|
|
|
);
|
|
|
|
}
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
fn lowpass_first_harmonic() {
|
|
|
|
let internal_frequency: f64 = 100e6;
|
|
|
|
let adc_frequency: f64 = 500e3;
|
|
|
|
let signal_frequency: f64 = 50e3;
|
|
|
|
let harmonic: u32 = 2;
|
|
|
|
let corner_frequency: f64 = 1e3;
|
|
|
|
let demodulation_frequency: f64 = harmonic as f64 * signal_frequency;
|
|
|
|
let demodulation_phase_offset: f64 = 0.;
|
|
|
|
let time_constant_factor: f64 = 5.;
|
|
|
|
let tolerance: f32 = 1e-2;
|
|
|
|
|
|
|
|
lowpass_test(
|
|
|
|
internal_frequency,
|
|
|
|
adc_frequency,
|
|
|
|
signal_frequency,
|
|
|
|
demodulation_phase_offset,
|
|
|
|
harmonic,
|
|
|
|
corner_frequency,
|
|
|
|
PureSine {
|
|
|
|
frequency: demodulation_frequency,
|
|
|
|
amplitude_dbfs: -30.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
&mut vec![
|
|
|
|
PureSine {
|
|
|
|
frequency: 1.2 * demodulation_frequency,
|
|
|
|
amplitude_dbfs: -20.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
PureSine {
|
|
|
|
frequency: 0.8 * demodulation_frequency,
|
|
|
|
amplitude_dbfs: -20.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
],
|
|
|
|
time_constant_factor,
|
|
|
|
tolerance,
|
|
|
|
);
|
|
|
|
}
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
fn lowpass_second_harmonic() {
|
|
|
|
let internal_frequency: f64 = 100e6;
|
|
|
|
let adc_frequency: f64 = 500e3;
|
|
|
|
let signal_frequency: f64 = 50e3;
|
|
|
|
let harmonic: u32 = 3;
|
|
|
|
let corner_frequency: f64 = 1e3;
|
|
|
|
let demodulation_frequency: f64 = harmonic as f64 * signal_frequency;
|
|
|
|
let demodulation_phase_offset: f64 = 0.;
|
|
|
|
let time_constant_factor: f64 = 5.;
|
|
|
|
let tolerance: f32 = 1e-2;
|
|
|
|
|
|
|
|
lowpass_test(
|
|
|
|
internal_frequency,
|
|
|
|
adc_frequency,
|
|
|
|
signal_frequency,
|
|
|
|
demodulation_phase_offset,
|
|
|
|
harmonic,
|
|
|
|
corner_frequency,
|
|
|
|
PureSine {
|
|
|
|
frequency: demodulation_frequency,
|
|
|
|
amplitude_dbfs: -30.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
&mut vec![
|
|
|
|
PureSine {
|
|
|
|
frequency: 1.2 * demodulation_frequency,
|
|
|
|
amplitude_dbfs: -20.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
PureSine {
|
|
|
|
frequency: 0.8 * demodulation_frequency,
|
|
|
|
amplitude_dbfs: -20.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
],
|
|
|
|
time_constant_factor,
|
|
|
|
tolerance,
|
|
|
|
);
|
|
|
|
}
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
fn lowpass_third_harmonic() {
|
|
|
|
let internal_frequency: f64 = 100e6;
|
|
|
|
let adc_frequency: f64 = 500e3;
|
|
|
|
let signal_frequency: f64 = 50e3;
|
|
|
|
let harmonic: u32 = 4;
|
|
|
|
let corner_frequency: f64 = 1e3;
|
|
|
|
let demodulation_frequency: f64 = harmonic as f64 * signal_frequency;
|
|
|
|
let demodulation_phase_offset: f64 = 0.;
|
|
|
|
let time_constant_factor: f64 = 5.;
|
|
|
|
let tolerance: f32 = 1e-2;
|
|
|
|
|
|
|
|
lowpass_test(
|
|
|
|
internal_frequency,
|
|
|
|
adc_frequency,
|
|
|
|
signal_frequency,
|
|
|
|
demodulation_phase_offset,
|
|
|
|
harmonic,
|
|
|
|
corner_frequency,
|
|
|
|
PureSine {
|
|
|
|
frequency: demodulation_frequency,
|
|
|
|
amplitude_dbfs: -30.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
&mut vec![
|
|
|
|
PureSine {
|
|
|
|
frequency: 1.2 * demodulation_frequency,
|
|
|
|
amplitude_dbfs: -20.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
PureSine {
|
|
|
|
frequency: 0.8 * demodulation_frequency,
|
|
|
|
amplitude_dbfs: -20.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
],
|
|
|
|
time_constant_factor,
|
|
|
|
tolerance,
|
|
|
|
);
|
|
|
|
}
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
fn lowpass_first_harmonic_phase_shift() {
|
|
|
|
let internal_frequency: f64 = 100e6;
|
|
|
|
let adc_frequency: f64 = 500e3;
|
|
|
|
let signal_frequency: f64 = 50e3;
|
|
|
|
let harmonic: u32 = 2;
|
|
|
|
let corner_frequency: f64 = 1e3;
|
|
|
|
let demodulation_frequency: f64 = harmonic as f64 * signal_frequency;
|
|
|
|
let demodulation_phase_offset: f64 = 0.;
|
|
|
|
let time_constant_factor: f64 = 5.;
|
|
|
|
let tolerance: f32 = 1e-2;
|
|
|
|
|
|
|
|
lowpass_test(
|
|
|
|
internal_frequency,
|
|
|
|
adc_frequency,
|
|
|
|
signal_frequency,
|
|
|
|
demodulation_phase_offset,
|
|
|
|
harmonic,
|
|
|
|
corner_frequency,
|
|
|
|
PureSine {
|
|
|
|
frequency: demodulation_frequency,
|
|
|
|
amplitude_dbfs: -30.,
|
|
|
|
phase_offset: PI / 4.,
|
|
|
|
},
|
|
|
|
&mut vec![
|
|
|
|
PureSine {
|
|
|
|
frequency: 1.2 * demodulation_frequency,
|
|
|
|
amplitude_dbfs: -20.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
PureSine {
|
|
|
|
frequency: 0.8 * demodulation_frequency,
|
|
|
|
amplitude_dbfs: -20.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
],
|
|
|
|
time_constant_factor,
|
|
|
|
tolerance,
|
|
|
|
);
|
|
|
|
}
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
fn lowpass_adc_frequency_1e6() {
|
|
|
|
let internal_frequency: f64 = 100e6;
|
|
|
|
let adc_frequency: f64 = 1e6;
|
|
|
|
let signal_frequency: f64 = 100e3;
|
|
|
|
let harmonic: u32 = 1;
|
|
|
|
let corner_frequency: f64 = 1e3;
|
|
|
|
let demodulation_frequency: f64 = harmonic as f64 * signal_frequency;
|
|
|
|
let demodulation_phase_offset: f64 = 0.;
|
|
|
|
let time_constant_factor: f64 = 5.;
|
|
|
|
let tolerance: f32 = 1e-2;
|
|
|
|
|
|
|
|
lowpass_test(
|
|
|
|
internal_frequency,
|
|
|
|
adc_frequency,
|
|
|
|
signal_frequency,
|
|
|
|
demodulation_phase_offset,
|
|
|
|
harmonic,
|
|
|
|
corner_frequency,
|
|
|
|
PureSine {
|
|
|
|
frequency: demodulation_frequency,
|
|
|
|
amplitude_dbfs: -30.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
&mut vec![
|
|
|
|
PureSine {
|
|
|
|
frequency: 1.2 * demodulation_frequency,
|
|
|
|
amplitude_dbfs: -20.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
PureSine {
|
|
|
|
frequency: 0.8 * demodulation_frequency,
|
|
|
|
amplitude_dbfs: -20.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
],
|
|
|
|
time_constant_factor,
|
|
|
|
tolerance,
|
|
|
|
);
|
|
|
|
}
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
fn lowpass_internal_frequency_125e6() {
|
|
|
|
let internal_frequency: f64 = 125e6;
|
|
|
|
let adc_frequency: f64 = 500e3;
|
|
|
|
let signal_frequency: f64 = 100e3;
|
|
|
|
let harmonic: u32 = 1;
|
|
|
|
let corner_frequency: f64 = 1e3;
|
|
|
|
let demodulation_frequency: f64 = harmonic as f64 * signal_frequency;
|
|
|
|
let demodulation_phase_offset: f64 = 0.;
|
|
|
|
let time_constant_factor: f64 = 5.;
|
|
|
|
let tolerance: f32 = 1e-2;
|
|
|
|
|
|
|
|
lowpass_test(
|
|
|
|
internal_frequency,
|
|
|
|
adc_frequency,
|
|
|
|
signal_frequency,
|
|
|
|
demodulation_phase_offset,
|
|
|
|
harmonic,
|
|
|
|
corner_frequency,
|
|
|
|
PureSine {
|
|
|
|
frequency: demodulation_frequency,
|
|
|
|
amplitude_dbfs: -30.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
&mut vec![
|
|
|
|
PureSine {
|
|
|
|
frequency: 1.2 * demodulation_frequency,
|
|
|
|
amplitude_dbfs: -20.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
PureSine {
|
|
|
|
frequency: 0.8 * demodulation_frequency,
|
|
|
|
amplitude_dbfs: -20.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
],
|
|
|
|
time_constant_factor,
|
|
|
|
tolerance,
|
|
|
|
);
|
|
|
|
}
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
fn lowpass_low_signal_frequency() {
|
|
|
|
let internal_frequency: f64 = 100e6;
|
|
|
|
let adc_frequency: f64 = 500e3;
|
|
|
|
let signal_frequency: f64 = 10e3;
|
|
|
|
let harmonic: u32 = 1;
|
|
|
|
let corner_frequency: f64 = 1e3;
|
|
|
|
let demodulation_frequency: f64 = harmonic as f64 * signal_frequency;
|
|
|
|
let demodulation_phase_offset: f64 = 0.;
|
|
|
|
let time_constant_factor: f64 = 5.;
|
|
|
|
let tolerance: f32 = 1e-2;
|
|
|
|
|
|
|
|
lowpass_test(
|
|
|
|
internal_frequency,
|
|
|
|
adc_frequency,
|
|
|
|
signal_frequency,
|
|
|
|
demodulation_phase_offset,
|
|
|
|
harmonic,
|
|
|
|
corner_frequency,
|
|
|
|
PureSine {
|
|
|
|
frequency: demodulation_frequency,
|
|
|
|
amplitude_dbfs: -30.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
},
|
|
|
|
&mut vec![PureSine {
|
|
|
|
frequency: 1.1 * demodulation_frequency,
|
|
|
|
amplitude_dbfs: -20.,
|
|
|
|
phase_offset: 0.,
|
|
|
|
}],
|
|
|
|
time_constant_factor,
|
|
|
|
tolerance,
|
|
|
|
);
|
|
|
|
}
|