mirror of
https://github.com/nqrduck/nqrduck-spectrometer-limenqr.git
synced 2024-12-22 23:47:47 +00:00
Fixed problem with pulse clipping.
This commit is contained in:
parent
a8c728a123
commit
51af43d74c
1 changed files with 8 additions and 2 deletions
|
@ -203,8 +203,14 @@ class LimeNQRController(BaseSpectrometerController):
|
|||
# Apply the shift by multiplying the time domain signal
|
||||
pulse_amplitude = (pulse_amplitude * shift_signal)
|
||||
|
||||
# Normalize the pulse amplitude
|
||||
pulse_amplitude /= np.max(pulse_amplitude)
|
||||
# Apply the relative amplitude
|
||||
pulse_amplitude *= parameter.get_option_by_name(TXPulse.RELATIVE_AMPLITUDE).value
|
||||
|
||||
# Clip the pulse amplitude to a minimum and maximum value of -0.99 and 0.99
|
||||
# this is kind of ugly but it prevents some kind of issue with the pulse clipping
|
||||
# I'm not sure why this happens but it seems to be related to the pulse shape
|
||||
# rectangular pulses seem to be the most effected by this
|
||||
pulse_amplitude = np.clip(pulse_amplitude, -0.99, 0.99)
|
||||
|
||||
if len(lime.pfr) == 0:
|
||||
# Add the TX pulse to the pulse frequency list (lime.pfr)
|
||||
|
|
Loading…
Reference in a new issue