New type of yaw control

[bhuism]

I left tri_motor_acc_yaw_correction default, maybe the default should be 0 then? I’ll set it to 0 from now on.

I see what you mean about the read values, I have no idea why that would be, in the betaflight configurator (and in flight) everything looks+is fine 🙂

[lauka]

Default is zero. That’s weird, the log is showing that it is in use.

[Minidarren83]

When I flashed mine the default was at 20

[lauka]

So it seems. Guess I forgot to set it 🙂

[bhuism]

default is definitly 20

diff showdefaults

shows:

#set tri_motor_acc_yaw_correction = 20
set tri_motor_acc_yaw_correction = 0

[fomoco_44]

I managed to get 2 flights in today before the kids got in the way!
With the V6 build I was able to get a good tail-tune done in the back yard prior, with the servo oscillations only present at idle conditions.
First log was with mostly default settings, most obvious was the yaw correction set to 20:

https://1drv.ms/u/s!AtWLDB57HPnrhDgMFF1O5trZiflK

My impressions from this flight was that the yaw felt much more locked in than previous versions!…I still had some loss of control during rapid throttle-cuts but not as bad as before. And I also had the uncommanded yaw during extended forward flight.

Second log had the yaw correction set to zero:

https://1drv.ms/u/s!AtWLDB57HPnrhDkDGc-CF5NsMAcI

This flight had the tricopter more unpredictable, with throttle cuts having it almost fall out of the sky.

[lauka]

@fomoco_44: Thanks for the logs, the yaw seems to track the rc_command quite well.

Based on the logs, you don’t seem to have airmode on? It would help a lot with the throttle cuts.

I also found one interesting place in the LOG00004 at 2:36 when you’re cutting throttle. The copter starts to pitch forward quite violently, the I terms are reset when throttle hits zero as the air mode is not on. Pitch PID forces the tail motor to zero even with the tricopter mixer trying to add output to the motor. This means that the motor mix goes to negative. I will make a change to force the motor correction even when the motor mix is at negative. This means that pitch PID will have to work harder to not allow the copter to pitch forward, but I hope it will be able to do this with the air mode on.

[lauka]

I have pushed new changes to the branch, but will not make a build until I have tested it.

The weather is not looking too good for flying..

[fomoco_44]

Great!!…in the meantime I’ll turn on Airmode(usually do, must have been too tired to remember!) and get a couple of flights in.

[bhuism]

Quick flight with just some rolls and little bit of LOS with the head of new_control branch:

(Triflight 0.7 Beta 2 – new control test v6 / BETAFLIGHTF3 3.1.7 Jun 25 2017 / 11:24:00 (fc9f38d))

https://drive.google.com/open?id=0ByiRgpuS3FxsWHA5dnQ5SjlnMHc

I’ve added yaw_lowpass but I think that made the uncommanded yaw moment worse. Also even alittle bit of yaw during a roll is killing, expect 45 to 90 degree uncommanded yaw at end of roll, when you throttle up to catch the tri.

Also ending of yaw’s is soft to my taste, and not locked in. I see that d_yaw is lower that earlier versions, maybe I’ll try to up d_yaw later.

# diff
# Triflight 0.7 Beta 2 – new control test v6 / BETAFLIGHTF3 3.1.7 Jun 25 2017 / 11:24:00 (fc9f38d)
name baby
resource MOTOR 1 B07
resource MOTOR 4 NONE
resource MOTOR 5 NONE
resource MOTOR 6 NONE
resource MOTOR 7 NONE
resource MOTOR 8 NONE
resource MOTOR 9 NONE
resource SERVO 1 B09
resource PPM 1 NONE
servo 5 1091 1941 1515 40 40 100 -1

serial 0 32 115200 57600 0 115200
serial 2 8192 115200 57600 0 115200
aux 0 0 0 1300 2100
aux 1 28 0 1750 2100
aux 2 26 0 1300 2100
aux 3 31 1 1300 2100
rxfail 5 s 1000
set rssi_channel = 7
set digital_idle_percent = 1.500
set motor_pwm_protocol = DSHOT150
set frsky_vfas_precision = 1
set bat_capacity = 1500
set ibat_scale = 247
set gyro_notch1_hz = 0
set gyro_notch2_hz = 190
set gyro_notch2_cut = 90
set deadband = 1
set yaw_deadband = 1
set tri_tail_servo_speed = 307
set tri_servo_feedback = EXT1
set blackbox_rate_denom = 4
set osd_units = METRIC
set osd_cap_alarm = 1500
set osd_time_alarm = 0
set osd_vbat_pos = 2437
set osd_rssi_pos = 2433
set osd_flytimer_pos = 2443
set osd_ontimer_pos = 374
set osd_flymode_pos = 12
set osd_vtx_channel_pos = 2456
set osd_horizon_pos = 200
set osd_current_pos = 408
set osd_mah_drawn_pos = 2451
set osd_craft_name_pos = 289
set osd_pid_yaw_pos = 384
set osd_power_pos = 345
profile 0

set d_notch_hz = 180
set d_notch_cut = 80
set yaw_lowpass = 90
rateprofile 0
rateprofile 0

set rc_expo = 50
set rc_yaw_expo = 20
set thr_expo = 80
set roll_srate = 75
set pitch_srate = 75

[lauka]

You’re hitting the I term disable limit there at the end of the rolls.

Try increasing the
#define TRI_SERVO_SATURED_GYRO_ERROR (75.0f) in mixer_tricopter.h.
to 100.0f and pump up the I gain a bit for next flight.

[LitterBug]

Finally made it out to the field last night after a busy weekend and waiting for the wind to calm down. Was still rather windy and gusty for all but the last pack. Didn’t get a chance to try the inflight tailtune due to the wind, or do any tuning. Uploaded the log files here.
Both baby tris were pretty waggly.

Cheers!
LitterBUg

EDIT: Video from BabyTri #1 late in the day when the winds were calmer. Should match up to LOG00017.txt in the directory listed above.

[fomoco_44]

Ok, so I turned Airmode on permanently and set the yaw correction to 0, logs here:

https://1drv.ms/u/s!AtWLDB57HPnrhDo6fM_mrLmkpB2J

I noticed that the low yaw authority was great, but i was still getting uncommanded yaw during FFF.
I’ll try tomorrow with yaw correction back at 20.

For reference, the set tailtune set the thrust factor as per below:
tri_tail_motor_thrustfactor = 81

[fomoco_44]

@lkaino
Latest flight with the yaw correction set to 20, no other changes
Finally managed to get the youtube uploads working here:

log sync @ +22.689 approx.
https://1drv.ms/u/s!AtWLDB57HPnrhDsSTdfcq2jWhDAx

What I have been noticing for a while now is that the BB logs aren’t showing the gyro roll that is being induced by the uncommanded yaw. There are several spots where I’m flying straight ahead, then there is an uncommanded yaw and induced roll, but the roll gyro doesn’t reflect that??

[LitterBug]

Should be headed to the field tonight. Going to start by tuning out the Roll oscillations. Have it on both now. I switched to Gemfan 5152 props on Baby Tri1, and it now has the same roll oscillations that baby tri2 does with the beefier motors. Will start with dropping Roll P/D. Had just dropped D on baby tri 2 last time out to the field, and that didn’t quite tame it.

BBlog shows pretty clearly that Motors 1/2 oscillate 180 degrees out of phase with each other.

Cheers!
LitterBug

Attachments:

1. 

   RollOscillate.jpg
   

[Author]

Posts