From a day to another, no software change made, my heli started to don’t hold loiter anymore doing toilet bowling and even in alt hold is not stable.
Hardware changes made: removed play on heli head keeping everything smooth. Even moving the swashplate with blades attached feels super smooth.
Now I’m definitely not new to ardupilot, I have done usual compass calibration, in flight compass calibration and done mag fit. Issue persist. The feeling sometime in flight is that servos are not strong enough. It is a trex 450 FBL (no DFC) with hs-65mg on swash powered at 6v (heli is standard 3s Liion battery).
you can find here few logs from today (14-10-25) and compare to the old one in the same folder
Log from earlier in the month shows desire roll and pitch more or less in line with actual roll and pitch. Log from yesterday as completely off, like there is a delay.
I connected bec to vbat2 to monitor bec voltage and it is quite consistent.
I’ve analyzed this problem in detail over the past years. To be short: Yes, stiff rotor systems do have less than 90° of phase shift. But only as long as the fuselage can’t follow at the desired rotational speed. I call this “static situation”. This is the case, when the fuselage is strapped down in a wind tunnel or when the control input is just there to counteract asymmetric forces like an asymmetric weight distribution. If it’s a normal control input which leads to a rotation of the fuselage (“dynamic situation”) every rotor has 90° of phase shift after the fuselage reached the final rotational speed. Regardless of its stiffness.
For our model helicopters the time it takes to reach this final speed is very, very short, so we always have 90° of phase shift. (As long as we’re not doing crazy things with asymmetric weights.) There are some helicopters which still need an adjustment but this is only due to mechanical reasons (mainly in rotor heads with three or more blades).
So unfortunately, phase shift is not the problem here.
@Lorenzo_Pessah I just compared the parameters of your very last flight and the last good flight (2025-09-27 18-06-24.bin). There are many parameters that changed, including AHRS_ORIENTATION. Was that an intentional change? Maybe you should just load the old parameter list to your board again and try a new flight.
Between end of September and 1st of October I upgraded the FC from a speedybee F4 to a similar Matek h743 sold by triumph autonomous system.
I have done several flight with the new FC (including 15m autotune if the tail in loiter mode) before the issue appears in the most recent flights.
I’m currently investigating hardware as at current stage I’m pointing my finger to a new hs65mg servo.
The very old once are super robust while the new one looks doggy as it vibrate a lot at startup, while being extremely underpower for the first seconds. I have replaced it with a completely different servos and pending a test flight as soon as I can.
Hi Lorenzo!
Hm, the hardware change makes spotting the cause of your problem more difficult here. Do you have the parameters of the last flight without toilet bowling? What changed since then? Or didn’t you change anything at all before the problem occurred? In this case, I think the most likely cause is indeed your servo issue. With a 120° swash plate a faulty servo could cause toilet bowling, I guess…
I just added few flights from yesterday. same folder linked above
I replaced the servo first and still same condition. I then started to play with pids (long log) and I got some questionable results when I reduced P for positioning which to be honest I never had to change in many many previous UAV plus before it was working ok.
last log I mechanically changed the phase but still nothig. Now if it is mechanically related, doing an head upgrade to a genuine align DFC might resolve. but I still thinking that i’m wasting my money. Other change I have done before is setup better the notch filter using notch one for the main head and notch 2 for tail and relative harmonics.