I dont know about baseflight, but my personal tarduino version has this problem. Stick calib becomes a very slow, what was 1 trim pulse before is 8 trim pulses now. I do use stick calib a lot, very nice feature.
I did see computeRC still uses this 4 times averaging filter and the 4us deadband thing. I think that is not necessary with stm32 1us accuracy and propose this patch i have been using for some time now. It DOES make display jumpier, but it also gives your controls crispness and precision never heard of before.
Code: Select all
Index: src/mw.c
===================================================================
--- src/mw.c (revision 421)
+++ src/mw.c (working copy)
@@ -233,23 +233,10 @@
void computeRC(void)
{
- static int16_t rcData4Values[8][4], rcDataMean[8];
- static uint8_t rc4ValuesIndex = 0;
- uint8_t chan, a;
-
- rc4ValuesIndex++;
- for (chan = 0; chan < 8; chan++) {
- rcData4Values[chan][rc4ValuesIndex % 4] = rcReadRawFunc(chan);
- rcDataMean[chan] = 0;
- for (a = 0; a < 4; a++)
- rcDataMean[chan] += rcData4Values[chan][a];
-
- rcDataMean[chan] = (rcDataMean[chan] + 2) / 4;
- if (rcDataMean[chan] < rcData[chan] - 3)
- rcData[chan] = rcDataMean[chan] + 2;
- if (rcDataMean[chan] > rcData[chan] + 3)
- rcData[chan] = rcDataMean[chan] - 2;
- }
+ uint8_t chan;
+ for (chan = 0; chan < 8; chan++) {
+ rcData[chan] = rcReadRawFunc(chan);
+ }
}
static void mwArm(void)
Me thinks if user doesent like the quantization and jumpiness of the real world then filter it in the GUI.
3 days from Japan to Slovenia! Can't wait to test this little thing 
