diff options
author | Matthias P. Braendli <matthias.braendli@mpb.li> | 2018-12-05 11:19:07 +0100 |
---|---|---|
committer | Matthias P. Braendli <matthias.braendli@mpb.li> | 2018-12-05 11:19:07 +0100 |
commit | 31b65e41043900c0cadd80961f4b22cdfc171e7d (patch) | |
tree | cdeceac026a2d1e0fe8c00af5d0f867767d17ef4 /python/dpd/RX_Agc.py | |
parent | 5cf52c74e9eb6bf8a82af4509ff3eb5106f928f9 (diff) | |
download | dabmod-31b65e41043900c0cadd80961f4b22cdfc171e7d.tar.gz dabmod-31b65e41043900c0cadd80961f4b22cdfc171e7d.tar.bz2 dabmod-31b65e41043900c0cadd80961f4b22cdfc171e7d.zip |
Get GUI to communicate with DPDCE
Diffstat (limited to 'python/dpd/RX_Agc.py')
-rw-r--r-- | python/dpd/RX_Agc.py | 45 |
1 files changed, 27 insertions, 18 deletions
diff --git a/python/dpd/RX_Agc.py b/python/dpd/RX_Agc.py index 0cc18b8..2a2f548 100644 --- a/python/dpd/RX_Agc.py +++ b/python/dpd/RX_Agc.py @@ -13,6 +13,7 @@ import numpy as np import matplotlib matplotlib.use('agg') import matplotlib.pyplot as plt +from typing import Tuple import dpd.Adapt as Adapt import dpd.Measure as Measure @@ -39,31 +40,39 @@ class Agc: self.measure = measure self.adapt = adapt self.min_rxgain = c.RAGC_min_rxgain + self.max_rxgain = c.RAGC_max_rxgain self.rxgain = self.min_rxgain self.peak_to_median = 1./c.RAGC_rx_median_target - def run(self): + def run(self) -> Tuple[bool, str]: self.adapt.set_rxgain(self.rxgain) - for i in range(2): - # Measure - txframe_aligned, tx_ts, rxframe_aligned, rx_ts, rx_median= \ - self.measure.get_samples() - - # Estimate Maximum - rx_peak = self.peak_to_median * rx_median - correction_factor = 20*np.log10(1/rx_peak) - self.rxgain = self.rxgain + correction_factor - - assert self.min_rxgain <= self.rxgain, ("Desired RX Gain is {} which is smaller than the minimum of {}".format( - self.rxgain, self.min_rxgain)) - - logging.info("RX Median {:1.4f}, estimated peak {:1.4f}, correction factor {:1.4f}, new RX gain {:1.4f}".format( - rx_median, rx_peak, correction_factor, self.rxgain - )) - + # Measure + txframe_aligned, tx_ts, rxframe_aligned, rx_ts, rx_median=self.measure.get_samples() + + # Estimate Maximum + rx_peak = self.peak_to_median * rx_median + correction_factor = 20*np.log10(1/rx_peak) + self.rxgain = self.rxgain + correction_factor + + measurements = "RX Median {:1.4f}, estimated peak {:1.4f}, correction factor {:1.4f}, new RX gain {:1.4f}".format( + rx_median, rx_peak, correction_factor, self.rxgain) + logging.info(measurements) + + if self.rxgain < self.min_rxgain: + w = "Warning: calculated RX Gain={} is lower than minimum={}. RX feedback power is too high!".format( + self.rxgain, self.min_rxgain) + logging.warning(w) + return (False, "\n".join([measurements, w])) + elif self.rxgain > self.max_rxgain: + w = "Warning: calculated RX Gain={} is higher than maximum={}. RX feedback power should be increased.".format( + self.rxgain, self.max_rxgain) + logging.warning(w) + return (False, "\n".join([measurements, w])) + else: self.adapt.set_rxgain(self.rxgain) time.sleep(0.5) + return (True, measurements) def plot_estimates(self): """Plots the estimate of for Max, Median, Mean for different |