Hi.
I have a problem after updating AC to 3.2.1 from 3.1.5.
When in Loiter, the the speed of the motors rev up and down/pulsate quite rapidly. This was working fine when I was using previous versions of arducopter.
I have downloaded the logs, and have looked into accX, accY and accZ. These seems to be inside the range specified in the wiki at arducopter. (Log_vibrations.bin)
Please have a look at the Log.bin, and note the difference between ThrIn and ThrOut when I'm in Loiter mode.
Anyone having any idea what the problem may be?
Replies
Rob,
You nailed it!!
And it is embarrassing! With Arducopter 3.1.5 - after gradually changing "Throttle Accel" P/I from 0.750/1.500 to 0.500/1.000, and "Throttle Rate" P from 6.000 to 3.000 - the pulsing is gone. Just a very small disclaimer - the wind calmed down during testing... I noticed earlier that Rate Roll and Rate Pitch PID values stayed the same during version changes, and assumed other values would do the same. My only excuse - I thought AutoTune took care of the most important parameters and never thought of changing the most relevant parameters for throttle. Live and learn.
After that I got the courage to reinstall 3.2.1, keeping the 3.1.5 parameters. It was getting dark outside, but I sneaked in a short test flight in the garden. The pulsing is back! Again you were spot on - IIRC the default Throttle Accel P/I for 3.1.5 were 0.750/1.500 and for 3.2.1 P/I were lower 0.500/1.000. Throttle Rate P are 6 and 5 respectively. So tomorrow I shall play with the parameteres again - trying to reduce "Throttle Accel" even more, even if THR_ACCEL_P lower than 0.5 is "out of range" according to Mission Planner. I did not get as far as changing the Altitude Hold P parameter (1.000 for both versions) today - how would that affect the behavior?
From reading the logs and graphs, the jagged ThrOut curves are still present in Loiter mode, even after the successful tuning in 3.1.5, so I guess that is normal?
Regards Jan - and thank You!
Would be interesting to see how many people complaining have badly set values. Helped with a case on the ardupilot forum, not had response back on that to see if it helped, but the gains were higher than default.
Hi again!
Bad news from last days tuning 3.2.1. I have played with tens of permutations of ThrottleAccel P&I without satisfaction. Mostly lowering both by the same rate, ie. "P" half the "I" value. The pulsing decreases a little by lowering the gains, but the AltHold is getting worse/unacceptable. Re-balanced props, checked BARO foam protection, and nothing is hanging loose on the frame.
Returned to 3.1.5 and kept default values for 3.2.1 (ThrottleAccel P/I = 0.1500/1.0000) and the the pulsing is gone again and AltHold is OK.
In my case the default values were bad in 3.1.5, also in 3.2.1, even if they were different and lower at the the first installation (0.5000/1.0000). I have looked at my tuning screenshots from my first installation (3.2.1), and they confirm my rates were default - not higher.
I think that my issues may come from my combination of frame, motors, ESCs. Looking forward to try 3.2.1 with new motors (change from 460kv to 620kv) and perhaps different props (from APC 14"x4.7" to 12" carbon fiber). This will bring ThrOut down to around 500 on 3s batteries. I also notice that the OP has a big tricopter with similar issues..
-Jan
There was another listing on this problem and it was the value of ATC_ACCEL_RP_MAX set too low. have set mine to medium 108000 and ATC_RATE_RP_MAX set to 25000.
Apparently the 3.2.1 download has very low values which cause a problem.
Good luck
I don't think those params should affect this problem.
Thanks "Henny V"!
I forgot to mention it in my previous posts - so much to remember...
From my "tuning log" (sorry for formatting and Norwegian) I changed this earlier:
"DIYdrones - Autotune-tråden, side 64, omtalar mulige endringar:
Original Ny 270315. (Lefébvre)
ATC_ACCEL_RP_MAX 0 72000 54000
ATC_ACCEL_Y_MAX 0 36000 36000
ATC_RATE_FF_ENAB 0 1 1
Standard...
ATC_RATE_RP_MAX 18000 18000 72000
ATC_RATE_Y_MAX 9000 9000 18000
ATC_RATE_SLEW_YAW 1000 1000 1000
Dette fungerte ikke. (..did not help)"
I tried... :-)
I shall keep it in mind when I reinstall 3.2.1 - maybe increase ATC_ACCEL_RP_MAX even more.
-Jan
Hehe, I noticed - thanks for your suggestion anyway! From the log you may see that the thr_mid already was set to 640 (recommended, same as ThrOut). Before your latest message I set it back to standard 520 and made a test - as we can see, same jumping throttle in Loiter... In this test I used a lighter battery so the thr_out was about 520.
I also made a video from flying the tricopter, uploading to YouTube now. I think it performs very well in Stabilize, but looks like there is something wrong in all autonomous modes. Perhaps it is lack of tuning (but have performed AutoTune and adjusted/reduced PID a little after that), perhaps because of the frame construction (scratch build), or something wrong with the controller.
Here is the YouTube link: http://youtu.be/kYiwiqPYlvY
Hi guys,
We can only help if we have good data logs to look at. So far only Rolf has posted logs. I had a look at them, and can't see much. His vibration levels do seem fine. For the short segments he's in Loiter, the motors are clearly pulsing. There is a negative correlation between climb rate and throttle output, but impossible to say if this is cause or effect. The logging resolution is not high enough to see which happens first.
I do notice that his Alt Hold PIDS are exactly the default settings. These might need to be tuned.
The talk of "bugs" is a bit premature. Keep in mind, we have over 10,000 users, and there are only a handful of people complaining of a problem.
Is this only happening on APM-class hardware? Or Pixhawk as well?
Jun, so I see that your Throttle PIDS are *not* the default, at least some of them aren't. So it looks like you have attempted to tune it which is good.
However, I do see some divergence between the Baro altitude, and the Inav altitude estimate, which makes me suspect you have a vibration problem. There is no vibration data in that log. Also, how is the barometer covered on the board you used?
I have noticed that my quad tends to lose altitude when moving forward fast in AltHolt and Loiter. I always put this down to some aerodynamic effect, but should this actually be happening? Loiter/AltHolt in hover seems ok. Never tried 3.1.5 - started on 3.2.
-
12
-
13
-
14
-
15
-
16
of 16 Next