Debugging the tricopter mini racer

[RCExplorer – David]

Finally got around to updating to Triflight 0.4. Had time to get 1 flight in before dark.
Feels like I got more drift yaw drift, and roll drift! in this version than 0.3. Will try switching back to 0.3 tomorrow to see if there is a difference. Would be strange if there was?

Here is the blackbox log

On request from Lauka I’ll also try playing around with the I term a bit. Looks like I got some vibration on the craft as well. I’ll try to eliminate that as well.

Awesome work guys!

[lauka]

I’ve been going through the logs from @Terje and @David. In both cases the drifting seems very similar.

I fail to spot any issues in FW logic. The issue is most likely in the tune, namely too low I gains.

Looking forward to hear results from @David after upping the I gain. Hopefully I can try out myself on weekend.

[biggestRCEfan]

Pardon my ignorance @lauka, but how do I read the log file, with what SW?

[lauka]

Easiest one is cleanflight blackbox explorer, it’s a chrome plugin.

There’s also command line tools to turn them into csv.

[Terje]

There even is a md about it on cleanflight github 🙂

[biggestRCEfan]

Thanks guys!

[lauka]

Quick update: I have been working on an IMU heading based yaw PID. First impression is that it removes almost all drifting (the IMU itself drifts a tiny bit under vibration). There are still some drawbacks that need to be sorted out before a release can be made, but with the little testing I have done it seems very promising :).

[Terje]

Just a quick warning about using cheap pololu buck converters with the BMS-22HV servo: Don’t 🙂
Crash and burn baby, crash and burn…)

Attachments:

1. 

   pololu.jpg
   

[swissfreek]

That’s very exciting @lauka! Thanks so much for taking our comments to heart and taking immediate action. Now I might have to get my wife to ship my mini Tricopter to me so I don’t have to wait six months to fly her again… 😉

[swissfreek]

Looks like the order was more “burn and crash”, @Terje, you wild experimenter. Either way, sorry to hear about your woes. Hope there wasn’t too much damage, or too much burn AFTER the crash!

[biggestRCEfan]

@lauka, got the blackbox set up. At least it is recording. Question: Triflight is running a loop time of 1000. What is a good blackbox_rate_denom value? Currently set to 2. Also, how do I know the blackbox FW for OpenLog installed correctly? I saw somewhere that the contents of the Config file is an indication, but in my case it was set at baud 9600. I did set it to 250000. Will do a flight in the morning and see what gives.

[lauka]

@biggestRCEfan: 1:2 is a good logging ratio, you can try 3:4 also. You can see in the blackbox explorer if you get too much dropped frames for your liking. I think for the blackbox FW the config file has only two values on two rows (baudrate value on first row and text “baud” on second).

Today was the first FPV session in weeks for me, blew up my naze board on the first flight :D. Changed the board and did some LOS testing for the heading lock feature as it was getting too dark for FPV.

Here’s a video showing how it works so far, I’m pretty happy with it. You can see in the beginning that there are still things to fix on take off, the locked heading was way off and copter quickly snapped to the target heading. You can also see that when yawing aggressively there’s small bounce back when it snaps back to the heading. When I cut the throttle the copter loses authority (no airmode here yet), but when I raise the throttle back it snaps right back :).

I have no idea yet if this works on other copters, will test on my V3.5 after I’m motivated enough to change the naze board to that as well.

[Jan Paul]

So, I’m out of ways of fixing now.
I tried for about 2 hours of tuning 0.4b with no improovment at all. I toyed with all ralted values in the software but the craft sayed unflyable. The yaw was loosing orientation all the time, in all flight situations.

So reluctantly, I rolled back to Triflight 0.1. Now my miniTri is stable again. The yaw is still not really locked in, but waaaay better than I could get it in the newer versions.

So, seems like I have to skipp some versions. But Ill be back as soon as thinking of tuning multicopters dosn’t infuse me with rage anymore.
And count me in if you manage to implement thar air mode! I will tune the s**t out of anything for that! Hopefully I’ll have blackbox and video running to help than.

[lauka]

@androidfennek: Couple of questions for you:
Was the yaw responsive at all, was it satisfactory when you were commanding yaw?

By losing orientation do you mean that it failed to maintain the heading, i.e. yaw was drifting?

Did you try increasing the I gain of yaw?

Did you use the Triflight setup guide to set up the FW?

[Jan Paul]

@lkaino: Thanks for the repyl, amd sorry, I guess I was to vague there.

-Was the yaw responsive at all, was it satisfactory when you were commanding yaw?
Yes, yaw was responsive. I had good yaw authority in both directions.

-By losing orientation do you mean that it failed to maintain the heading, i.e. yaw was drifting?
Correct, it was loosing heading. While hovering, Yaw started drifting after a seemingly random amount of time. Some times it hovred stable for 5 to 10 seconds, than lost heading. In fast forward or backward flight yaw drifted hevily to one side. (I cant remember if it was always to the same side). I also noticed slight roll drift, but that could have been in my head, trying to overcompensate for yaw. Tring tailtune mode didn’t get me any berrter behaiviour.
Btw: normally I have no flightmode selected, so i guess i’m flying in rate mode.

-Did you try increasing the I gain of yaw?
Yes, I have experimented with the p and i gains. Tried everyting(i gain) between too low and high enough to get slow oselations. Also upped i a little for roll to stop it from drifting. It got roll more stable, but sadly no change in behaviour for yaw.

-Did you use the Triflight setup guide to set up the FW?
Yes. To make shure I reflashed(inclusive reset) the bord and carefully did it again. Sadly there was no change in behaviour.

I will defenitly try again some time, but the wether this weekend was to good for testing. I needet some time in the air, an got that!
As sad, I will document my next trys on video and hopefully with blackbox.

Dont break your head over my stupid edgecace, if you spot nothing aparrent in my report. I can fly for now. Maybe more data will help.

[Author]

Posts