I would like to ask advice on how to decrease pitch/roll oscillations when stopping/braking. I tried to remedy this by setting atc_thr_mix_man = 0.9 but that does not seem to work very well.
Attached download link ( ) of a logfile from a short testflight to test/show this behaviour.
It is a flight in althold, takeoff with a few roll/pitch wobbles to see if rate tracking in hover is good, then flight sideways at about halfstick. If I then center stick I see a quite significant oscillation both visually and in the logs in rate.r / rate.rdes. The same effect happens in pitch ( somewhat later in the logfile ). During this flight, atc_thr_mix_man was set to 0.9.
Bit of context on the copter that is maybe not in the logfile.
fairly large system ( 32â props )
tuning and above mentioned flight at âlowest massâ of 7.5kg
manual tune so far ( weather here is not cooperating, lots of wind )
The wiki autotune page mentions setting atc_thr_mix_max = 0.9 after a âgood tuneâ, to âThis can reduce the pitch overshoot sometimes seen (especially on copters with large propellers) in AltHold if the vehicle suddenly slows after performing a fast forward flightâ, which seems correlate with what I am seeing. I tested this with atc_thr_mix_man = 0.9 in above mentioned flight, but the oscillation is still quite visible and a bit concerning, especially since the system is only flying halfstick.
I think the system rates track ok in hover, i.e. when I look at rate.x vs rate.desx , but I am still testing if the tracking is ok during higher airspeeds and so would not classify this as a âgood tuneâ yet. Also Iâm aware of the quite high vibration levels the system is experiencing, which is being worked on⌠but not down to acceptable levels yet, i.e. there are still vibrations/oscillations in the system so I am hesitant to set atc_thr_mix_max to 0.9. yet.
What I would like to ask:
Is there a different way to reduce these oscillations? / am I missing anything?
Is there a difference between atc_thr_mix_max & atc_thr_mix_man other then the flight modes? I am assuming max â all modes, manâ acro, stab & alth.
I would fix that first to get those down before you get too far into tuning.
I think there is a lot of un-needed filtering applied but until the mechanical vibrations are lowered thatâs going to be difficult to nail down.
I would suggest running Quick-tune on this (after the vibrations and filters are fixed). Quicktune works well for big platforms and also doesnât mind windy conditions.
We know about the vibrations, and about the clipping, and are working on a solution ( frame with better stiffness ). Until that updated system is tested we are attempting to fly, tune and gain experience with the old oneâŚ
We tested mounting the fc on different silicone/rubber standoffs, which did not help much. On the flight in this log the fc was mounted on a âvibration damping mounting setâ from ( Vibration Damping Mounting Set â Unmanned Tech ) which we hope helps a bit⌠and it does help a bit but not very much.
What filtering is un-needed? We dont want to filter without good reason but followed a filtering howto ( and used the webtool which was very helpful ).
Will look into quicktune, so far I have mostly worked with autotune and manual.
The problem with all the vibration is itâs getting in the way of everything else.
Youâre building a house on the wet mud while you wait for the concrete to arrive. Itâs not going to work out well.
Typically unless you have an internal combustion motor the wisdom is to hard mount the flight controller or at least use the least amount of soft mounting possible. Soft mounting can make it difficult to get a good tune because the controller is being numbed by soft mounts. Ideally, you need to get the mechanics of the frame sorted first before you get too far ahead of your self.
As for the filters, with all the vibration there is too much for the filters to manage. You have a lot of harmonics active and thatâs going to load the processor and delay the filter response. Again, with the vibrations managed youâll find you need less filtering.
Also consider that when you fix the airframe (more rigid, less vibration) youâre going to end up changing the filters, and tuning. So assuming you donât break anything, youâre not really going to gain a lot fighting with this one.
You got the logging wrong for the notch filter. You donât want batch logging and RAW_LOG_OPT, itâs one or the other. And, you donât need to log all gyros. You have a bad case of âcheck all the boxesâ
But, this is incidental. As @Allister noted, there is no point in advancing or even flying this craft with that frame.
I agree with all of this, and I do have a bit of experience with e.g. internal combustion engine driven craft ( hence the soft - mount fallback ). That is also why there is a new frame on the way⌠which will take a bit of time.
The idea was to gain a bit of experience with the old setup while the new one is being worked on. A part of that experience is logging a few minutes of airtime with a bit of airspeed. This is when the braking - oscillation issue arose I wanted to ask advice on.
If that issue also arises on the new frame build, and I have no reason why it would not if the wiki description is accurate, I might as well attempt to troubleshoot it now. So any thought on how to troubleshoot this is still very much appreciated.
Aha! Good point, must have forgotten to switch off RAW_LOG_OPT.
We often fly two flights, one with batch logging on, the other with raw imu logging on. The batch logging to evaluate imu behavior&filtering on the field ( without fast internet ). The raw imu logging for the newer filter review tool, which is very handyâŚ
filter setup is based on the 7â long range build series which I still think is a good setup guide.
Youâre unlikely to achieve any decent results with the mechanical vibrations and flex that frame has. Just canât beat the physics. Itâs likely the oscillations are a function of the frame and youâll only get so far with parameters.
I appreciate the desire to go through the process and to learn. So with that in mind I would try fix up that frame the best you can. Maybe sacrifice some flight time and do something to stiffen it up. Then re-do the filters once the vibrations are lower. Then try quick tune and see what it gives you.
We already tried to stiffen the frame using braces connecting the arms laterally and to the center frame. That does help, a bit, but not sufficient to get the vibeâs << 30m/s^2. Also checked if props and motors are well balanced etc⌠and they are as far as we can tell, so excitation should be fairly low.
The new frame is designed to have a lot higher stiffness, but that will probably not decrease slow-turning prop vibrations as these are well below arm resonant frequencies. Guess weâll see in a couple of weeks.
