This issue collects various smaller bugs / improvements w.r.t. fixpoint processing
[x] : When selecting and designing a new filter type without fixpoint implementation, the old fixpoint response is plotted (and repeated for IIR filters if the previous was too short). Desired behaviour: switch back to float and issue a warning, color the run button in red
[x] : When the FFT window is changed (even without modifying the number of data points), stimulus and response are recalculated. This can be quite time consuming for fixpoint simulations, it would be sufficient to recalculate the FFT
[x] : When transient simulation is in fixpoint mode, selecting a new filter, window type or whatever triggers a new fixpoint simulation. Even in Autorun mode this should only happen when a new filter has been designed. This probably happens because a new filter ('filt_changed') updates the input_fixpoint_specs widgets which in turn trigger a auto run. Probably auto run should be disabled for fixpoint simulations. This would solve some other minor hiccups as well.
[x] : A fixpoint simulation that was triggered by changing of specs (autorun), starts with 'fx_sim' : 'finish' and finishes with 2x 'fx_sim' : 'finish'. Starting with the "RUN" button in the h[n] tab, doesn't emit the initial 'finish' but also produces 2x 'finish' at the end. Only running from "Sim. HDL" produces the expected results (but skips the 'init' phase)
[x] : When a filter without fixpoint model is selected, selecting the fixpoint tab crashes with
File "/home/cmuenker/Daten/design/python/git/pyfda/pyfda/input_widgets/input_fixpoint_specs.py", line 461, in resize_img img_w, img_h = self.img_fixp.width(), self.img_fixp.height() AttributeError: 'NoneType' object has no attribute 'width'
This issue collects various smaller bugs / improvements w.r.t. fixpoint processing
[x] : When selecting and designing a new filter type without fixpoint implementation, the old fixpoint response is plotted (and repeated for IIR filters if the previous was too short). Desired behaviour: switch back to float and issue a warning, color the run button in red
[x] : When the FFT window is changed (even without modifying the number of data points), stimulus and response are recalculated. This can be quite time consuming for fixpoint simulations, it would be sufficient to recalculate the FFT
[x] : When transient simulation is in fixpoint mode, selecting a new filter, window type or whatever triggers a new fixpoint simulation. Even in Autorun mode this should only happen when a new filter has been designed. This probably happens because a new filter ('filt_changed') updates the input_fixpoint_specs widgets which in turn trigger a auto run. Probably auto run should be disabled for fixpoint simulations. This would solve some other minor hiccups as well.
[x] : A fixpoint simulation that was triggered by changing of specs (autorun), starts with 'fx_sim' : 'finish' and finishes with 2x 'fx_sim' : 'finish'. Starting with the "RUN" button in the h[n] tab, doesn't emit the initial 'finish' but also produces 2x 'finish' at the end. Only running from "Sim. HDL" produces the expected results (but skips the 'init' phase)
[x] : When a filter without fixpoint model is selected, selecting the fixpoint tab crashes with
File "/home/cmuenker/Daten/design/python/git/pyfda/pyfda/input_widgets/input_fixpoint_specs.py", line 461, in resize_img img_w, img_h = self.img_fixp.width(), self.img_fixp.height() AttributeError: 'NoneType' object has no attribute 'width'