My cinelifter oscillates when throttle higher than 40%, tried autotune and manual tune multiple times and no improvement, gyro is super clean with almost no noise above 50hz.
I did some searching and see a github post mentioning that this could be caused by incorrect thrust expo, and I’ve read that coaxial build need less expo than a normal build, and someone (If I remember correctly, andy) said 0.5 is probably good for a coaxial build.
However, my drone is an acro quad so I’m not sure if this applies to me and I don’t have the ability to test the real expo, even manufacturer of my motor say that they don’t have the thrust chart to provide.
I’m using iFlight E80 4in1 esc and iFlight Xing2 3110 1250kv motor with 6s battery.
Anyone could share their thoughts?
Do you mean 8 inch props?
This would be normal, regardless of whether it’s a flat octo or coax/quadX8
MOT_THST_EXPO,0.62
The general rule is:
- set too high you can see instability at low throttle
- set too low you can see instability at high throttle
But this assumes the expo is your only problem
If you can provide a .bin log file much more can be checked.
Thanks for offering a look at my log.
Here is the log
https://drive.google.com/drive/folders/16V_ufwsa8lZu9umK_Ta7IGD0R0rGVCWG?usp=sharing
Apart from the oscillation on my X8, I also need some help regarding my 7 inch tune: I just can’t get it to perfect, it flies great but there is always some D-term oscillation, when I lower the D term, it fells terrible but when I raise it, esc starts to heat up but flies great. It will be nice if u could take a look and privide me some help.
For the octo, I haven’t done a lot of tuning on it, but it already feels ok to fly but there is oscilation in the footage.
And the log from my octo is weird, it always shows 1970 when downloading, but after downloaed, the date is corrected, it only happens to my octo and not the other quad. Octo is using Mamba H743 and quad is using NeutronRC H743 which uses Matek H743 target.
I’m thinking if my two drone are experiencing the same problem but one is more severe than the other… Maybe it’s just something I did wrong, like for the gyro low pass filter, both are set to 300hz, because when I run FFT with magnitude option checked, the dynamic notch seems to eliminated all of the noise
8 inch
I would start with these:
BATT2_MONITOR,9 // for ESC voltage, or 0 for disabled
BATT_ARM_VOLT,22.10
BATT_CRT_VOLT,21.00
BATT_LOW_VOLT,21.60
BATT_FS_CRT_ACT,1
BATT_FS_LOW_ACT,2 or 3
FFT_ENABLE,0
INS_GYRO_FILTER,51 // or 100 more likely
LOG_BITMASK,407294
Pitch and Roll tuning are quite different, from a point of view of PID values. The attitude control is similar and quite good.
I found some small oscillations mainly in Roll, I would probably adjust down the Angle P since it is so very high and increase the Rate P and I a bit since they are quite low and see how that goes.
ATC_ANG_RLL_P,28.0 // could even go a bit lower
ATC_RAT_RLL_D,0.00266
ATC_RAT_RLL_I,0.08
ATC_RAT_RLL_P,0.08
Pitch could probably go the other way: Angle P increased and Rate P and I decreased
ATC_ANG_PIT_P,22.0
ATC_RAT_PIT_D,0.006
ATC_RAT_PIT_I,0.18
ATC_RAT_PIT_P,0.18
I can see when you change to Loiter mode there is oscillations induced in Roll, obvious in the Rate and PIDs. I see you’ve reduced some PSC params to reduce the agressive position holding. I suspect this is tied to the super-high Roll P value. Next I would adjust the GNSS settings so it’s giving a more stable reading, GPA.Delta is all over the place
GPS_GNSS_MODE,65 // or 9 if your unit can definitely do Beidou
GPS_SBAS_MODE,0
Give it a go, see what you think.
7 inch
somewhat similar to the 8 inch quad.
Angle P’s very high, Rate P and I very low.
Once again I would reduce the unnecessary logging
and put this back to “normal” around 57 or even up to about 150
INS_GYRO_FILTER,300
I suspect 100 would be about right after looking at the FFT, to keep your responsiveness. I think 300 just means the Gyro filter is practically doing nothing.
With both the quads, you should be able to see oscillations during ascents or descents if the thrust expo was a problem - they both look clean in that regard.
Pitch and Roll are different is mainly because my drones aren’t standard X structure. 7 inch rp could be similar but 8 inch is much wider and short, there is also imbalance with the payload.
8 inch batt2 monitor is just batt1 with different amp/volt setting to give me a reference of bottom motor’s current, due to the lack of the second current sensor.
Both drone’s angle rp are set by autotune, I haven’t touched angle pid, thought it wouldn’t affect acro flying, I will try tuning it now.
I prefer not to have battery failsafe so I turned it off.
Didn’t notice GNSS problem, will take a look.
For the gyro low pass, since I came from betaflight and that’s what they always do. I’ve read that ap is much more sensitive to noise since it’s build more for reliability. Will try a low pass at 50/100/150 and see how my fells.
Thanks in advance, I will start tuning now and get back to u.
I still believe the oscillations that occur in specific cirmcumstances, like when you switched to Loiter, and some of the tiny oscillations while flying more level/sane are due to the PIDs. I think you could safely try mine and see how it goes, and it’s easy to put back your original values too.
In fact changing the gyro filter alone and rerunning Autotune on pitch and roll should give slightly different results to what you had before.
I think INS_GYRO_FILTER,100
would suit both of yours - that frequency is below most of the noise and will retain your responsiveness.
You can also set sag-compensated voltage, so the reported value is what the voltage should return to without the current drain - BATT_FS_VOLTSRC
That way you can still have the safety of failsafe actions when you get busy and dont notice the battery voltage, or the battery ages or goes bad, but you dont have the false triggers when current spikes with big climbs.
That’s a good idea to use the second battery monitor with a different scale.
The GPA.Delta rate doesnt necessarily mean you get bad position data, just that the timing is all wrong because the GPS unit is being overwhelmed by the number of constellations. So like everything else it’s better to be correct than left to chance
Tuned 7inch today, lowered gyro low pass to 100hz and rerun autotune, new pitch pid is double of before and roll pid are slightly lower than before. Didn’t notice any responsive issue, still as sharp as before.
Then I did some manual tune, lowered rp angle p and increased rp rate pid, and it now flies much better than before. Only some small oscillation when doing sharp turn, some oscillations when taking off from un-leveled ground and about 1 or 2 degree of nose dip when doing throttle punch. Esc and motor temp is normal and high throttle oscillation is gone. It turns out the problem before wasn’t caused by thrust expo.
Still could get some improvement but I’m so happy it flies so good now.
Those oscillations are what I consider as acceptable now, it only occurs when I try to make it happen, during cruising and light freestyling they rarely occur.
What I will do next is do some more testing and optimizing yet I wait for some more batteries to come so I could start tuning the x8 (it now takes off at around 5kg with camera, lens, dual 6s 6200mah 60c batteries and only flies about 7 minutes and then takes at least 40 minutes to charge the battery) And I will do what you suggest, to set up a battery failsafe,
Massive appreciation to you, your advise helped a lot.