Flight 1 April 13, 2014 6:00 pm Maybe 15 minutes of flight, consisting mostly of takeoff/landing practice and low height hover (2-10 ft). Max height 20-30 ft. Values: Stabilize P 3.5 P 0.1550 I 0.7700 D 0.0160 Ch 6 - set I Results: Generally flew well and landed fine, except for a few instabilities where it started wobbling after a slight throttle decrease intended to start landing (or maybe wobbles started first and then throttle decrease - see later). Towards end of flight, taking off required noticeably more throttle due to low battery. Elevator channel required reversing (pitch) which may have thrown off controls for second part of flight (roll motion was controllable but pitch was difficult to control). Yaw motion was controllable. A few times in wind it was wobbly, but allowing it to stabilize by itself seemed to work (at that point in flight also batteries were lower power level, while at beginning of flight it was much more wobbly descending from high altitude). Had some low-height 'crash' landings, which (?) caused the APM to turn sideways and later affected flight performance by having craft constantly tilted. Two propellers moved from their original position (the counterclockwise ones, moved clockwise as seen from top) but the propeller holding wire did its job. Suggestions: Likely only slight changes will be necessary. Online forum http://forums.openpilot.org/topic/6777-tuning-observation-ki/ mentioned decreasing rate I. Also increasing rate D is suggested to have more control since fast oscillations were not observed. APM needs to be centered and better attached, after which accelerometer needs to be re-zeroed. Create a curve for throttle on transmitter for better vertical controllability. increase rate D a little (could be more stable) zero rate I a bit (oscillates at height) lower stabilize P a little Changes: New settings (written to device after flight): Stabilize P 3.45 P 0.1550 I 0.0050 D 0.0170 Ch6 - control stabilize P (range 3.0 - 4.0, start at 3.45) Radio ranges on APM calibrated with reversed elevator channel Throttle curve set up on the transmitter: 0 33 50 65 100 Propellers re-attached and re-marked Motor bolts checked and marked APM Accelerometer re-attached and re-calibrated Flight 2 April 20, 2014 7:30 pm Reaching (?) 30 minutes of flight time, numerous takeoff/landings and a few longer continuous hover flights (up to few minutes). Max height 20-30 ft. Tested sharper maneuvers and response to fast stick movements - stable even if not as fast as could be. (Mid flight changes) Stab P set to 3.52 D to 0.0180 Final settings at end of flight Stab P 3.52 P 0.1550 I 0.2790 D 0.0180 Results Flew well, no observed instabilities Stick response works best with stick moved towards end of range, ie middle of stick is not sensitive unlike the case with Cathi or some other models - action requires noticeable movement of the stick but once that is done the quadcopter responds appropriately Stick response was tested with roll nearly full stick input, and then throw back to zero - quadcopter responded appropriately and instabilities were not evident Quadcopter would sometimes rotate (yaw) during hover Quadcopter would often descend after sudden movements (such as stick returned to zero - horizontal motion would stop appropriately but during that time it would begin descending and continue doing so unless throttle was raised). Perhaps this was due to low battery, but more testing is required. Throttle control seems to be improved with curve set up on remote, although hover begins above 50% mark and it is possible that the curve as it is set up may lead to over-correct easily if medium speed ascents are desired. Perhaps raise both left and right quarter-points so the curve is steeper in bottom section, then reasonably flat at above 50%, and steep again. One CCW propeller has moved slightly from its position, with a zip tie protector keeping it in place - likely during a crash. It does not seem necessary to re-attach at this point (hot glue still providing a good amount of friction). One bolt on other CCW motor might have moved and should be double-checked (done). Other bolts and props remained in place. APM remained flat (roll) but has changed in pitch (ie reads 'nose up' on ground) - needs to be re-checked in room where initial calibration was done. Suggestions Find a better way to attach APM to the assembly, as currently it is easy to somehow change pitch/roll settings Buy (make: it is a voltage divider for 0-3.3V output of 0-51.8V range) a voltage indicator than can be hooked up to APM, and LEDs http://plane.ardupilot.com/wiki/voltage_current-3/ It will be necessary to get more connectors for the bolts though http://copter.ardupilot.com/wiki/common-using-a-current-sensor/ -> use a servo connector for the voltage sensor instead of the weird power module thing https://www.sparkfun.com/products/10644 http://dlnmh9ip6v2uc.cloudfront.net/datasheets/Sensors/Current/DC%20Voltage%20and%20Current%20Sense%20PCB%20with%20Analog%20Output.pdf Have timer set for next flight Adjust throttle curve Figure out why it is descending (suggestion to adjust Throttle_P and Throttle_I but not sure which direction - try slight increase in both) See if the slight yaw rate problem can be fixed somehow - potentially motors not straight? Not much to do about that. Try increasing Stab_Yaw_P Double check all frame/structural bolts Lower stab_P very slightly to get rid of slow wobbles Some of the above suggestions came from https://code.google.com/p/arducopter/wiki/AC2_Tweaks GPS connections: http://diydrones.ning.com/forum/topics/apm-2-5-with-neo-6m-ublox-gps-not-3dr-lea-6?id=705844%3ATopic%3A1004467&page=8#comments Actions Bolts and props double-checked APM re-attached using two aluminum supports and better foam for vibration isolation, accelerometer re-calibrated (accuracy was within 0.1 degrees) Found one loose bolts on a central battery holder, two loose bolts on camera holder - tightened all Also noticed hot glue on bottom of camera servo has come undone Re-tightened both power terminal screws/nuts Stab_P lowered from 3.52 to 3.515 Stab_Yaw increased from 4.5 to 4.55 Throttle Accel P increased from 0.75 to 0.76 Throttle Accel I increased from 1.5 to 1.52 Throttle curve adjusted to 0,40,57,65.5,100 -> gives increased control in hover region and slightly increased control for descent region // (wrong setting - transmitter does not allow this) Pitch curve adjusted to 0,20,50,80,100 -> gives faster response in the middle of stick range Timer set up on transmitter to use during flight - set to 30 minutes Flight 3 April 21, 2014 6:30 pm Timed 20 minutes total flight time, with a few brief landings thrown in. Medium heights were attained throghout (10-30 ft) and controllability was tested, roll pitch and yaw. One crash occurred when stick moved from one extreme to other side, causing the copter to over-react initially but it did not have enough motor power/response to react properly, thus it became unstable losing altitude and crashed. Crash was from a height of about 5 ft; copter worked well for rest of flight but should be re-inspected. Getting increasingly used to controls, and was able to land it softly a few times (throttle curve likely helped). Hover was observed at approximately 1 notch above halfway mark on the transmitter. Results Flew well, slight wobbles still occurring likely due to wind. Should not be flown in windy weather. Throttle response was quite good in the lower range (for landing) but can be a bit more aggressive for ascending. Perhaps increase mid point and second quarter points. Ending battery voltage was 11.04 V (starting under-charged at slightly less than 12.6 V) Extrapolating to 30 minutes gives about 10.27 V ending, which is 3.42 V/cell and above the '3V' cell minimum.. hopefully For a full 30 min flight, recommended to 'top off' both batteries with a slow charge to full 12.6V and balance properly Copter would sometimes yaw in a strange direction, likely due to wind? Not sure if the Stab_Yaw did much, perhaps could be increased more (did not notice a vibration) Copter would still drift sideways in wind, but this could be handled by controlling it manually One crash caused it to flip over, and two CCW props have moved from original position Both will need to be re-attached, consider better attachment method Batteries seem puffier than usual? Acrylic was cracked near one of the supports Suggestions * Re-attach both CCW props * Check batteries puffyness * Check motors are straight and properly spaced on rod after crash landing * Check all bolts again * Reattach acrylic base since it is cracked * Adjust throttle curve to be more aggressive in ascending range (hover was observed at approximately 1 notch above halfway mark on the transmitter) * The double aluminum APM mount seems to be working fine but accel needs to be re-calibrated after crash * Check that the carbon fiber rods are properly aligned to the central holder in terms of movement along the axis of the rod * Perform a radio distance test * Increase Yaw_D - hopefully prevent problem with yaw turning slightly back after spinning one way (overshoot on the command to stop yaw rotation when stick moved to zero) x Throttle rate P - may need to increase or decrease, depending on whether the barometric reading is being maintained well or poorly. This would require some sort of logging or in flight testing with OSD. Actions Changed throttle curve to 0,42,59.5,78,100 Measured radio range - without amplifier approximately to edge of basketball courts (440 ft) in urban setting Yaw_D increased from 0 to 0.01 Replaced base and tightened bolts Put in props with sandpaper to stop from spinning in place Camera and transmitter system wired in MinimOSD wired in Telemetry port (outside to inside): Ground, not used, OSD Tx/APM Rx, OSD Rx/APM Tx, +5V Voltage detector added Sensor works VOLT_DIVIDER param set to 4.12712 (corresponding to 13.8 V max) -> tested and readout is correct Flight mode 4 set to Alt Hold (in case transmitter fails) Calibrated accelerometer using PREFLIGHT_CALIBRATION command (under Actions tab) Flight 4/5 April 26, 4:00 pm and 8:00 pm Flew reasonably well, hard to control at altitude/in wind and crashed from fairly high up, cracking new base! Changed Stab_P from 3.515 to 3.25 Changed Rate_I from 0.2790 to 0.1 Was more controllable, but still crashed after trying to respond to wind Changed Rate_I further to 0.01 (hopefully this helps?) Re-charged batteries and set up everything for autotune, will try this Updated firmware to 3.1.3 Set up autotune with channel 7 Changed flight mode 6 to alt hold for autotune (gear switch on right of transmitter) Channel 7 connected to the throttle hold switch (large switch on left of transmitter) Need to re-calibrate compass and accelerometer after firmware update (done) Take screenshot of PIDs before doing auto-tune (done) Test flying in Alt Hold mode, then do the auto-tune http://copter.ardupilot.com/wiki/autotune/ http://copter.ardupilot.com/wiki/altholdmode/ AutoTune worked! It still crashed though Changed I_roll from 0.12 to 0.1 and I_pitch from 0.16 to 0.135 Better but still crashed Changed Stab_Pr and Stab_Pp from 3.9375 to 3.70 Better, with suggestion of descending more slowly made very sharp curve for throttle on transmitter, and when descending slowly there is little oscillation According to http://diydrones.com/forum/topics/wobble-on-decent increasing the rate Is may help However this effect seems to be due to wind and this would be explained by an I term that is too high: http://www.youtube.com/watch?v=LtocGBngSrA (Try increasing rate I as from video of flight crash it seems that it did not 'over-react' but rather 'under-react' and disturbance was due to prop wash rather than aggressive changes) Changed Rate_Roll_I from 0.1 to 0.2 and Rate_Pitch_I from 0.1 to 0.23 Changed Angle_Max from 4500 to 3000 (max 30deg pitch/roll) Changed Batt_Capacity from 3300 to 12000 mAh Changed LAND_SPEED from 50 to 15 cm/s Changed MOT_SPIN_ARMED from 70 to 0 Changed Pilot_VELZ_MAX from 250 to 100 Changed Stab_Roll and Stab_Pitch from 3.70 to 3.80 Attached new acrylic base and re-attached 3 props that moved Params at this point are saved as screenshot Flight 6 April 28, 6:30 pm Increased I caused some significant instabilities Ended up decreasing I, increasing D, and increasing P Final parameters worked quite well, with copter being able to recover (!) from a wind gust These are saved under Apr 28 file name Pitch stabilization was found to be problematic so the Stab_P for pitch is slightly higher, this seemed to work well but maybe could be even higher up to 3.87 (this has been changed) One CCW propeller moved, but it seems fairly stable in current position so re-attaching will not be necessary The CCW motors need to be better attached as they are shaking - rotating about the axis of the carbon fiber rod APM mount has held up well, not requiring re-calibration of accelerometer The final PIDs need to be tested in larger area flight, since that testing was done on a small field and in windy conditions - need to test gaining more speed and going through sharper turns Final throttle curve during that flight was 0,48,57,74,100 and provided reasonable control although sometimes fast descent - perhaps the Q1 point could be moved up to 49 (done) Check D setting on pitch/roll -> during flight copter would sometimes continue drifting in same direction after performing a maneuver and returning stick to center, this could be because the D term is too high and thus prevents P term from completely stopping the motion; however not sure if this is more evident on pitch or roll since they are set differently Flight 7 May 1, 4:30 pm 25 minutes recorded flight Took video, recording general flight and a crash Will try adjusting D parameter, flying in both stab and alt hold modes Adjusted D down on roll, recovered many times but one crash from high up - seemed to be on way to recovery but was oscillating a lot Perhaps try increasing D to past value fraction on both roll/pitch to stop the oscillations? (0.072*P) The Alt Hold mode was slightly too aggressive, ie commanding rapid/sudden changes rather than being 'smooth', would drop past expected altitude then suddenly pick up throttle (too much P?) - should be corrected before trying in next flight Also altitude reading changed significantly in wind so it was not used extensively in this flight Alt hold mode needs to be explored more thoroughly, as it is likely that the crash that occurred was due to trying to fly in Stabilize mode without a good view of the quadcopter (looking at it from the bottom makes control difficult, especially throttle - may have been descending too fast) In stabilize mode, after ending maneuvers sometimes the quadcopter would gain significant height - is this to be expected? My guess is that this is a property of the mode. In next flight, make note of at what point on the throttle the quadcopter descends, and use that as a guide for throttle when the quadcopter is flying high above After further consideration of advanced statics concepts, it seems that the two-level carbon fiber rod architecture does not create additional torques during maneuvers requiring roll or pitch changes, thus adjustments to the PID settings should be used to fix observed oscillations Recommended: After viewing flight video, it seems that reaction does not occur quickly enough (it rolls too much before applying power to other motor) and when it occurs, reaction is too much Perhaps this is due to not enough P (Stab P in this case because it was trying to remain flat), and subsequently both P and I term become in effect once copter turns far enough and the D term is not enough to counter both, so an instability arises Recommended to increase Stab_P proportionately and maintain Rate_P Set Stab_P_roll and Pitch to 3.9 Rate_D likely can be increased more to prevent reaction overshoot, however the effects of increased Stab_P need to be checked in flight first Actions Re-attached broken propeller Re-attached both CCW motors with both plastic holders in place rather than zip ties Set Stab_P roll/pitch to 3.9 Consider increasing Rate_D as well (see Apr 28 screenshot) -> not done at this point, will do flight testing Cleared logs for faster response Decreased Alt_Hold_P to 0.70 Also decreased Throt_Accel_P and Throt_Accel_I to 0.75 and 1.5 respectively Accelerometer re-calibrated - advised to use trim in flight rather than re-calibrate so often Flight 8 May 9, 7:30 pm It was very windy which required some abrupt maneuvers to remain in same location; however despite the wind and maneuvers no crashes occurred! The settings seem to be OK, need to also be tested in calm weather and in conditions of ascending/descending vertically on a calm day. Additionally, it was difficult to test alt hold mode due to wind, so that should be tested on a calmer day. Flight 9 May 11, 3:30 pm It was somewhat windy. Tried alt hold mode, it seemed to not react fast enough so increase alt_hold_p to 0.87 Increased rate_D to 0.009 and 0.0113 from 0.0085 and 0.0108 Still crashed in alt hold (dropped throttle too far down and hit ground) changed Imax throttle accel to 3.0 from 5.0, then increased alt hold P to 0.95 D seemed a bit much so decreased to 0.0087 and 0.0110 It fell again from high altitude - perhaps an effect of increased rate_D Stab_P increased from 3.9 to 3.95 for both roll and pitch Alt_hold settings changed to those during autotune (worked OK). with previous settings it was oscillating in altitude, with slow frequency about 1 Hz but quite large amplitude. with new settings, it did not work too well either (refer to video comments). Got a decent recovery when something happened at high altitude in stab mode (about 100 ft), however the cause of the disturbance is uncertain and should be determined from logs. Faster recovery would be ideal, perhaps the settings are a bit conservative overall. Also, instabilities were observed during later high altitude flights, although one with a slow descent (looking at the copter from a hill) was markedly better. Recommended to download logs and compare with video (synchronize with video based on highest altitude gained) to determine cause of disturbance and possible fix. Not likely that fast descent was the cause of all the disturbances - wind could be some but not all. There was something flying off one of the propellers after the high altitue 'recovery' in the video. Flight 10 May 24, 2014 5:25 pm Recommended to enable RCOUT log for better understanding of PID action; currently actual and desired roll visible but not controller output Also recommended to reduce D setting to autotune levels (in roll/pitch response, sudden stick changes cause an opposing initial actual output change, which can only be caused by an excessive D level - this may also explain the oscillations seen during flight) Decrease D - evident on Pitch, slightly less on Roll Increase P - evident on Pitch + Roll Increase I - evident on Pitch + Roll (should be last change) Perhaps increase IMAX to 500 default Recommended settings: Lock roll+pitch together Stab_P = 3.95 Rate_P = 0.1600 from 0.12/0.16 on roll/pitch Rate_I = 0.1000 from 0.07/0.09 on roll/pitch Rate_D = 0.0040 from 0.0087/0.0110 on roll/pitch IMAX = 500.0 from 400.0 Yaw_D decreased to 0.0075 from 0.01 logs enabled: ATTITUDE_FAST PM CTUN NTUN IMU CURRENT MOTORS before was: ATTITUDE_MED PM CTUN NTUN CURRENT (downloaded log 1 with above settings) Based on roll/pitch information, suggested to decrease Rate_I and Rate_D Decreasing: Rate_D = 0.0000 from 0.0040 Rate_I = 0.0800 from 0.1000 Other settings unchanged Now noticeably less stable with slow oscillations - due to no D damping?? (downloaded log 2 with above settings) Increased Rate_D = 0.0020 from 0.0000 Other settings unchanged More stable than before, still induced wobbles in both pitch and roll for some reason, although maneuvers were only in either pitch or roll Additionally it was difficult to control, and would tend to drift after returning stick to center, subsequently starting oscillations on its own (downloaded log 3 with above settings) Suggestion: Besides using above logs for PID settings, check PM log to see if something is interfering with controller speed Recommended: Reduce Stab_P (this has been said to cause a see-saw like effect) to 3.5 Set Rate_I and Rate_D to 0.0000 Enable PID log (logs ch.6 data) Use ch.6 for in-flight tuning of Rate_P; find level where oscillations after step input are acceptable (close to touching desired value on first peak) - see p 2-3 on http://pse.che.ntu.edu.tw/chencl/Process_Control/2006/12%20Tuning%20PID%20Controllers.pdf Next increase Rate_I to point where oscillations centered on desired value and die out slowly Next increase Rate_D for faster oscillationi die-out May increase Rate_P subsequently Flight 11 May 25, 2014 2:00 pm Tested adjusting Rate_P in-flight using ch. 6 with others zeroed in Stabilize mode (using log 4) Recommended initial Rate_P = 0.204, Stab_P = 3.6 Next tuning Rate_I from 0.0000 to 0.3000 (log 5) Recommended initial Rate_I = 0.066 Next tuning Rate_D from 0.0000 to 0.3000 (logs 6+7) Recommended initial Rate_D = 0.015 Not recommended to increase Rate_P further since oscillations were already present without D term With Rate_D, flying was very stable! Note that these adjustments were in relatively wind free area and at low altitude. Testing in wind might necessitate increasing Rate_I with a corresponding decrease in Rate_P in case it tends to drift. The flight was stable despite a loose motor (motor 1) which necessitated the break between logs 6 and 7. Rate_D was adjusted until the sound of ESCs trying to make fast adjustments went away, and where oscillations after a maneuver were not visible. Some oscillations were still observed when descending vertically into prop wash; limited oscillations when descending both horizontally and vertically. Final settings: Stab_P = 3.6000 Rate_P = 0.2050 Rate_I = 0.0660 Rate_D = 0.0150 IMAX = 500.00 At this point, disabled ATTITUDE_FAST IMU PID resulting in following enabled: ATTITUDE_MED PM CTUN NTUN CURRENT Note that IMU logs suggest excessive vibration levels, something to consider for future modifications After tuning, went to local park to fly - was very stable even with aggressive maneuvers (including moving stick all the way to edges) Very little wind at the time - perfect flying weather but tuning needs to be tested in wind as well Two videos (third person) of this flight are in Quadrotor folder Stable hover observed (keeping it in place for Chris taking photos) - suggesting I rate is OK? From maneuvers though, it seems that raising I a bit could help respond better to slight stick movements. Instabilities observed in fast descent as seen in one of the videos, however copter compensates for it much better than before and wobbles are not too great + die out quickly. (logs 8 and 9 downloaded of this flight. Log 8 flights are in videos. Log 9 is a short flight with battery level seemingly low; at end there was slight crash as throttle level was insufficient, perhaps low battery perhaps not enough throttle input - at this point about 30 minutes of flight had been done including all the tuning flights) Flight 12 June 15, 2014 Visually observed some divergence (of the I type) flying downwards while angled nose down in previous flights. Thus raising Rate_I from 0.0660 to 0.0900 and decreasing Rate_P from 0.2050 to 0.155 (proportional difference) at first. Looks like that I was too high. Change Rate_I to 0.07 and Rate_P to 0.1933, added some tape to camera since it was shaky Appears that camera is out of focus since taping it did not fix the problem. Flying was OK, but from logs not sure if settings could be better overall. Fairly stable control, no crashes but throttle response is a bit tough (easy to fall too quickly). Changed Rate_D from 0.015 to 0.016 due to observed oscillations - stability seemed to increase, although overshoot observed on returning stick to center (perhaps due to I term? - with the higher I term, descent while tilting had better performance however). The latest configuration seems good and was fun to fly, only crashes were due to overestimating thrust at 100% throttle or low battery effects.