Hurricos
commented
8 months ago I see the issue with my proposed change:
diff --git a/muddersMIMA_firmware/brakeLights.cpp b/muddersMIMA_firmware/brakeLights.cpp
index 8e4f1d1..8d94a43 100755
--- a/muddersMIMA_firmware/brakeLights.cpp
+++ b/muddersMIMA_firmware/brakeLights.cpp
@@ -40,7 +40,7 @@ uint8_t brakeLights_handler(void)
//JTS2doNow: When brake pressed, gpio_getBrakePosition_bool() alternates between "Lights ON" & "Lights OFF"
if (joystickPercent < JOYSTICK_MIN_ALLOWED_PERCENT) { gpio_brakeLights_turnOff(); } //joystick input too low
else if(joystickPercent < TURN_BRAKE_LIGHTS_ON_BELOW_JOYSTICK_PERCENT) { gpio_brakeLights_turnOn(); } //strong regen
- else if(gpio_getBrakePosition_bool() == BRAKE_LIGHTS_ARE_ON) { gpio_brakeLights_turnOff(); }
+ else if(gpio_getBrakePosition_bool() == BRAKE_LIGHTS_ARE_ON) { gpio_brakeLights_floatPin(); } //not controlling brake light, so don't assert it
else { gpio_brakeLights_floatPin(); }
}
else if(brakeLightMode == BRAKE_LIGHT_FORCE_ON) { gpio_brakeLights_turnOn(); }The issue is that in mode 1, if the brake pedal is pressed, BRAKE_D goes high, making BRAKE_uC go high and activating the BRAKE_EN input of U1, the high-side FET driver. We need to drive the high-side FET low for long enough to disable the FET before we can resolve whether the brake switch is supplying voltage on BRAKE_RAW.
Edit: I believe you've commented on the forums RE: changes to improve brake-off detection; separately, it seems this TODO requests it intentionally, but the implementing commit 15e8938a321050a84cbf16b3732b58566b6ac1a8 still includes the TODO:
I'll review this TODO more closely before I have more questions.
Continuous BRAKE detection isn't working in mode 1: when one connects the brake switch, it is initially detected as pressed, but 250 < interval < 750ms later is read as 'released' even if the brake is still depressed (before alternating back to pressed). This impairs the effectiveness of any solutions which monitor brake position continuously.
I think I know why this is happening:
In mode 1 (=>
BRAKE_LIGHT_AUTOMATIC), the initial brake press is read without issue (great for safety!), because if LiControl isn't driving the brake light,However, once [the FET driver is fully released?], the brake lights [are detected as on by the brake switch], and in mode 1 (=>
BRAKE_LIGHT_AUTOMATIC) the firmware assumes it is the one driving it on and turns it off, forcing the GPIO input off (though brake lights and other things feeling the switch stay on).I'll test the change I have in mind for this; it involves tracking whether LiControl is asserting the brake on, and leaving the brake GPIO floating otherwise.
@doppelhub please let me know if there are any concerns with the electronics side of this. I'll live if I fry my own boards, but not if this is misguided and risks other users' safety.I think I see the issue; see comment below.Additional note: I believe that TODO is unrelated to BRAKE_LIGHT_PULSE, which is for LiControl-driven braking; note that as a business, publishing firmware which makes brake lights flash may run afoul of NHTSA recommendations and federal law.