BF 4.3.1 + BiQuad Direct Form II math

quoting @Quick-Flash , Use Direct Form II for all biquad filter calculations. Previous code comments say this shouldn't be done, but reading online and testing has shown no ill effects when doing this. This should improve the CPU load of any filter that uses the biquad filter base. This includes the biquad type lpf and all notch filters.

I'd like people to get a good tune without the biquad calc changes using rpm filter plus dyn notch and then see if their tune still works.

• nothing to set, but the comparison is specifically for Notches and BiQuad LPF's
• A/B compare to BF 4.3.1 as stated above, and submit BBL of both

BF 4.3.1 + Smith + Kalman

• Betaflight 4.3.1 plus SmithPredictor plus KalmanFilter.
• Since KalmanFilter code is such a huge memory hog, this test-build removes MavLink, PPM, and Servos.

smith predictor defaults (CLI only):

smith_predict_str = 50
Allowed range: 0 - 100

smith_predict_delay = 40
Allowed range: 0 - 60

smith_predict_filt_hz = 5
Allowed range: 1 - 10000

kalman defaults (CLI only):

imuf_q = 10000
Allowed range: 1000 - 16000

imuf_w = 32
Allowed range: 0 - 64

BF 4.3.1 + SmithPredictor

• BetaFlight 4.3.1 plus Smith-Predictor
• Requires dev build Configurator (Fork of BetaFlight 10.8)
• flightControllerIdentifier EmuF and EmuF Icon embedded in HEX.

smith predictor defaults (CLI only):

smith_predict_str = 50
Allowed range: 0 - 100

smith_predict_delay = 40
Allowed range: 0 - 60

smith_predict_filt_hz = 5
Allowed range: 1 - 10000

What to Test/Feedback: (Feedback in EmuFlight's Discord or GitHub Discussions)

• A/B test a tuned quadcopter with Smith ON vs OFF and submit BBL as feedback.
• smith_predict_str = 0 is OFF. default is 50 (ON)
• set debug_mode = SMITH_PREDICTOR for Blackbox.

ALPHA EmuConfigurator

• fork of of betaflight configurator 10.8-maintenance
• points to https://github.com/emuflight/development for HEX, not official releases
• only linux version was tested