noptica2-sdr/noptica.py

import serial
import queue
import threading

import SoapySDR
import numpy as np
from scipy.signal import blackmanharris


class BufferedSDR:
    def __init__(self, sdr, channels, bufsize, nbufs):
        self.sdr = sdr
        self.channels = channels
        self.bufsize = bufsize

        self.stream = None
        self.terminate = False
        self.thread = None
        self.queue = queue.Queue(nbufs)
        self.available_buffers = queue.Queue(nbufs)
        for _ in range(nbufs):
            self.available_buffers.put([np.array([0]*bufsize, np.complex64) for _ in self.channels])

    def start(self):
        self.stream = self.sdr.setupStream(SoapySDR.SOAPY_SDR_RX, SoapySDR.SOAPY_SDR_CF32, self.channels)
        try:
            self.thread = threading.Thread(target=self.thread_target)
            self.thread.start()
        except:
            self.sdr.closeStream(self.stream)
            raise

    def stop(self):
        self.terminate = True
        self.thread.join()
        self.sdr.closeStream(self.stream)

    def get(self):
        return self.queue.get()

    def dispose(self, buffers):
        self.available_buffers.put(buffers)

    def thread_target(self):
        self.sdr.activateStream(self.stream)
        try:
            while not self.terminate:
                buffers = self.available_buffers.get()
                sr = self.sdr.readStream(self.stream, buffers, self.bufsize)
                if sr.ret != self.bufsize:
                    print("SDR sampling error")
                    return
                self.queue.put(buffers)
        finally:
            self.sdr.deactivateStream(self.stream)


class DummyInductionHeater:
    def __init__(self, port, induction_min, induction_max):
        pass

    def start(self):
        pass

    def set(self, amount):
        print("induction", amount)

    def stop(self):
        pass


class InductionHeater:
    """Interface to the MHS5200A function generator driving the LC tank"""
    def __init__(self, port, induction_min, induction_max):
        self.port = port
        self.induction_min = induction_min
        self.induction_max = induction_max
        self.queue = queue.Queue(1)

    def start(self):
        self.serial = serial.Serial(self.port, 57600)
        self.thread = threading.Thread(target=self.thread_target)
        self.thread.start()

    def thread_target(self):
        while True:
            amount = self.queue.get()
            if amount is None:
                break

            amount = max(min(amount, 0.5), -0.5)
            freq = ((self.induction_min + self.induction_max)/2
                    + amount*(self.induction_max - self.induction_min))

            command = ":s1f{:010d}\n".format(int(freq*1e2))
            self.serial.write(command.encode())
            self.serial.readline()

    def set(self, amount):
        self.queue.put(amount, block=False)

    def stop(self):
        self.queue.put(None, block=True)
        self.thread.join()
        self.serial.close()


# https://gist.github.com/endolith/255291
def parabolic(f, x):
    xv = 1/2. * (f[x-1] - f[x+1]) / (f[x-1] - 2 * f[x] + f[x+1]) + x
    yv = f[x] - 1/4. * (f[x-1] - f[x+1]) * (xv - x)
    return (xv, yv)


class Stabilizer:
    def __init__(self, freq_sample, amp_threshold, freq_target, k, tuner):
        self.freq_sample = freq_sample
        self.amp_threshold = amp_threshold
        self.freq_target = freq_target
        self.k = k
        self.tuner = tuner

    def input(self, samples):
        spectrum = np.abs(np.fft.fft(samples*blackmanharris(len(samples)))[0:len(samples)//2])
        for i in range(len(spectrum)//100):
            spectrum[i] = 0
            spectrum[-i] = 0
        i = np.argmax(spectrum)
        true_i, amplitude = parabolic(spectrum, i)
        freq = 0.5 * self.freq_sample * true_i / len(spectrum)

        if amplitude > self.amp_threshold:
            tuning = (freq - self.freq_target)*self.k
        else:
            tuning = 0.0
        self.tuner.set(tuning)


def continuous_unwrap(last_phase, last_phase_unwrapped, p):
    # note: np.unwrap always preserves first element of array
    p = np.unwrap(p)
    glue = np.array([last_phase_unwrapped, last_phase_unwrapped + (p[0] - last_phase)])
    new_p0 = np.unwrap(glue)[1]
    return new_p0 + p - p[0]


class PositionTracker:
    def __init__(self, leakage_avg):
        self.last_phase = 0.0
        self.last_position = 0.0
        self.leakage = np.zeros(leakage_avg)
        self.leakage_ptr = 0

    def input(self, ref, meas):
        demod = np.conjugate(ref)*meas

        self.leakage[self.leakage_ptr] = np.real(np.sum(demod)/len(demod))
        self.leakage_ptr = (self.leakage_ptr + 1) % len(self.leakage)
        leakage = np.sum(self.leakage)/len(self.leakage)

        phase = np.angle(demod - leakage)
        position = continuous_unwrap(self.last_phase, self.last_position, phase)/(2.0*np.pi)
        self.last_phase = phase[-1]
        self.last_position = position[-1]
        return position, leakage