New type of yaw control

Tagged: EAGLE TREE VECTOR, GAINS, PID, RCEXPLORE TRICOPTER V4, TUNINING

This topic has 224 replies, 20 voices, and was last updated 6 years ago by mcstan.

Viewing 15 posts - 1 through 15 (of 225 total)

1 2 3 … 13 14 15 →

Author

Posts
10 June, 2017 at 13:28 #38026
lauka
Participant
I implemented a new type of yaw control which is based on the tail motor virtual feedback. It makes the dynamic yaw obsolete. It does what dynamic yaw did but calculating the linear output all the time based on the motor speed. This should produce more accurate linear yaw output throughout the motor output range.

tri_yaw_boost defines the maximum yaw output that you can get at zero throttle. The new feature linearises the output so that the maximum output is the same at all throttle levels. So now decreasing the value reduces the overall yaw output power.

The changes are available here: https://github.com/lkaino/Triflight/tree/new_control. It’s still a WIP.

I have added HEXs for the most common targets here.

I would like some alpha testing help here :). It might require some tuning, I didn’t change the defaults. I raised the D term a bit on my copter to get rid of some wobbling at dynamic scenarios. I have only a small yard here for FPV testing, but I really liked how it flies.
Attachments:
10 June, 2017 at 15:44 #38033

mcstan
Participant

Hiya @Lauka

I just gave it ago in the garden and I have tail wag .. but it does keep pointing the same way.. which is a plus!!

the only time it does a random yaw is if you pulse the throttle which I understand seen the servo is slow and I guess it can’t keep up.. does this mean I can use my super fast servo again? lol

anyways
what are the main parameter I need to tune on the tricopter??

Do I get the feeling it’s just more than just simply doing PIDs ?

10 June, 2017 at 18:51 #38034

lauka
Participant

With these newer releases its mostly pids that you need to change, after running the tail tune.

What setup do you have?

Try increasing the yaw d or decreasing the p.

10 June, 2017 at 19:10 #38035

mcstan
Participant

Hiya

not able to get any mid air tail tunes right now.. I get no beeps or anything

I only get the crazy tailwag when is on the groud.. like the good old days when I used to fly the baby tricopter on triflight 0.5

but soon as it’s in hover it stops but still seem to be twitching but does not yaw which I was guessing that was okay.

I been dropping and up the p and d all afternoon.. I can never tell if its helped or not.. lol

going to the field on Sunday so can do some FPV with it then and change some of the values again.

I did take some black box stuff and the only one which ha a lot of movement on was the yaw P and D

still using my custom made tricopter 210 with 6ich props everything els is standard tricopter stuff from David store

11 June, 2017 at 19:21 #38050

mcstan
Participant

Hiya

well went to the club today and didn’t think it was too bad in fpv I still had lots of wobble on the yaw.

anyways my VTX seem to die during the day so that kind end my day

then a member helped me tune it in line of sight he managed to remove all the tail wobble by lowering the P on the yaw to 35 and it still had very good yaw control at this point.

the only thing it did do was when you dropped the throttle you would get a wobble on the tail but still kind seem you would keep the right heading

https://drive.google.com/file/d/0B0T-bBpbxqBRc1hKZ0daaWNSR2M/view?usp=sharing

anyways I attached the log if you wanted to see it

12 June, 2017 at 04:57 #38064

LitterBug
Participant

I’ll give this a try on my original build next time out.

Cheers!
LitterBug

12 June, 2017 at 09:42 #38068
lauka
Participant
New builds attached to this mail.

All the angle calculations in the code changed to use float. The changes are available in the branch.
Attachments:
12 June, 2017 at 11:21 #38073

LitterBug
Participant

Flashed to my original Baby with the 2100kv motors. Hope to hit the field later today.

12 June, 2017 at 22:25 #38079

LitterBug
Participant

Checked servo speed on both Baby Tris. Baby 1 that has been flown quite a bit actually has higher servo speed on 5.2 V than New Baby 2 on 6V with all new parts. Need to break him in more! 🙂

Baby Tri#1 5.22V Servo speed = 316 (with new tail routines)

Baby Tri#2 5.23V Servo Speed = 173
Baby Tri#2 5.93V Servo speed = 308 (with TF0.7Beta2)

Hope to hit the field in a short bit, but just got call from work…. So not sure how much if any air time I’ll get in today.

Cheers!
LitterBug

13 June, 2017 at 03:29 #38081

LitterBug
Participant

Gave the New Yaw Control a test on Baby Tri#1 with the 2100kv motors on default PIDs and no inflight tailtune. WOW… What a difference. I had left my worst behaving props on that frame from my last time out, and it did better than my best behaving props at the field last time. Think you have made some good progress. Will upload video later.

Cheers!
LitterBug

13 June, 2017 at 11:55 #38086

swissfreek
Participant

Any chance to get a build for a Colibri Race (aka TBS PowerCube)? And if you’ve got any tips on where I would connect servo feedback on that controller (or, more simply, a primer on how one would identify a resource to use for the EXT1 pin? A motor output? PPM input?)? Thanks for your help! Looking forward to get my baby back in the air but I’ve become spoiled with my OSD and built-in menus for VTX and PIDs…

13 June, 2017 at 19:46 #38095
lauka
Participant
@swissfreek: I attached a build to this post. The target already has all the ADCs defined:
```
#define VBAT_ADC_PIN            PC0
#define CURRENT_METER_ADC_PIN   PC1
#define RSSI_ADC_PIN            PC2
#define EXTERNAL1_ADC_PIN       PC3
```
I have no idea where the PC3 pin (EXT1) is available on the board. You could ask TBS?

@LitterBug: that’s good to hear.

I just flew couple of batteries on this and I’m liking it. There is the thing where the servo oscillates on throttle chop. Took some logs and it seems possibly be caused by the yaw_correction. I ran out of batteries, will test without the correction later this week.
Attachments:
1. triflight_new_control_2_COLIBRI_RACE.zip
14 June, 2017 at 00:15 #38109

LitterBug
Participant

Here’s video from yesterday another very windy day, even my quads were waggling… Only had time for one flight on each frame, so not able to tune the yaw. Baby Tri #1, New Yaw control firmware, Edge Racing 2206 2100kv Motors, Gemfan 5152 props, Stock PIDs and yaw parameters. 5.22V servo power, 316 Servo speed, too windy to get inflight tailtune. Some servo oscillations at initial arming that went way once off the ground. Propwash handled much better now. Feels much better in flight.

Going to work on redoing the 5V and bump the servo up to 6V. Wish I had not moved that resistor now. LOL Trying more aggressive props to get the low KV motors to perform better.

Cheers!
LitterBug

14 June, 2017 at 11:10 #38117
bhuism
Participant
Hmm, sorry to be de deliverer of bad news, but I can’t get rid of yaw oscillations with the new_control code, flashed with full chip erase, redid unarmed tail tune (tri_tail_servo_speed = 308), and verified min/mid/max of the servo.

I tried combinations:

* lowpass filtering on 50hz for P-yaw and D-yaw
* every combination of high/low p-yaw and d-yaw
* half default tri_yaw_boost (120)

All have some form of slow and/or fast yaw oscillation.

baby tri / bff3 fc / emax 2400kv / bluebird DMS-210DMH / 1500mah bat
Attachments:
1. new_control.zip
14 June, 2017 at 12:11 #38119

swissfreek
Participant

Hm, sounds like maybe it would just be easier to order a BetaFlight F3 and use that instead. So I did. I’ve been neglecting my baby tricopter for a long time now, and that makes me unhappy. Looking forward to getting it back in the air this weekend, and finally (finally!) having black box capability to share flight experiences. Hopefully not having 6V to the servo doesn’t prevent me from getting good results.
Author

Posts