I’m getting this vertical bouncing, most times when I fly in Stabilize Mode and seemingly even more frequently when in Loiter Mode. I’d really like to get rid of it before I try to QuickTune and AutoTune. Have never tuned any drone before because thus-far, I’ve only built 5" and below, using BetaFlight Presets.
In this video, it flies ok, but is still very unresponsive to throttle input (pitch, roll, and yaw seem very responsive) https://www.youtube.com/watch?v=z6txoZ_vUV0
But in this flight, the problem starts 42 seconds in, then RTH’s on its own and I turn on LAND mode. https://www.youtube.com/watch?v=8oE4p3GPMcI
And here, I show how it happens in Loiter Mode https://www.youtube.com/watch?v=suTxeErZ8ko
Firmware is 4.7.0-Dev (I think I got it from the Latest folder, which MicoAir told me to do.
The frame is a Tarot FY680 (the cheapest/simplest version) Generally pretty clean build, but I hadn’t put a couple of screws into one plate which I’ve put back in last night and had a couple of loose zip-ties which I’ve also fixed after the flight. 6x 3115 900kv motors running DShot600. HQProp 10x5x3 props. 6S voltage. Photos of the build here Daniel Yusim on Instagram: "Autonomous Hexacopter This one was honestly a pretty tough build."
From the symptoms seams like you have not tuned the altitude controller yet.
Either Google for that and follow the ArduPilot wiki documentation. Or install and use the ArduPilot methodic configurator software. The software will guide you step-by-step and you will learn a thing or two.
Please don’t use modes other than stabilize until you sort this out, follow what @amilcarlucas said with tuning. Chris has some good videos but there are errors in some of the tuning steps that he takes. I know that because I followed his steps to the “t” and kept running into issues.
An untuned copter in loiter and other modes greatly increases the possibility for you to buy parts.
I tried the methodical configurator up until the quicktune flight. I’m worried that I’m messing my params up because I don’t understand all the settings and it won’t allow me to delete some things from the suggestions. (they just pop back up at the bottom of the list)
It seems my vertical controller is way off. It could also have been an issue in loiter because of the learn-and-save procedure trying with a default MOT_THST_HOVER of 0.2 when I’m seeing much lower values from this log. (I think this one was a flight in Stabilize Mode) https://imgur.com/a/QHkDegt
It looks like MOT_THST_HOVER should be around 0.06? Though my MOT_SPIN_ARM is 0.08 which doesn’t make sense because it arms fine without flying away.
(Idk, should I set my MOT_THST_HOVER to 0.06, lower MOT_SPIN_ARM, and recalculate my vertical controller PIDS using methodical configurator? Than, I’d probably try to re-upload my MissionPlanner params and manually set up the new throttle/altitude controller values to retain the main MP Params)
Select the “normal” gui complexity at the top of the component editor window. That will allow you to deselect the upload of individual parameters.
The first flight is meant to be done in stabilize and then switched to althold (if it is working correctly) and you let it run in AltHold for at least 30 seconds. If your vehicle is not overpowered it will learn the correct hover throttle. AMC will read that in step 20 and correctly set the altitude controller gains.
If your vehicle is overpowered ( HOVER THRUST < 0.125) you need to manually set the hover throttle. Then the AMC will read that and correctly set the altitude controller gains.
Which values are you thinking that AMC is set incorrectly?
normal view seemingly let me delete things, but not all of them for all views. In some cases, it makes the param pop back up at the bottom of the list. I now see from a video linked on their GitHub that I can “skip parameter file(s)” and maybe skip right to the part I need. I’ll probably try it out again tomorrow morning.
It seems my vehicle is overpowered. The CTUN.ThO seems to be just above 0.05 from the graph linked in the imgur link
I have a lot of custom RC options set, which are different to the template I selected and the first page I filled out (or the defaults it wants to change to) for example, I’m using ELRS and have it set up for Yaapu Telemetry, Dshot with filters based on RPM data, am using a PCB FC and not a pixhawk, etc.
I did not realize that when I ran it, but here’s what I did and the drone just flew beautifully. (ready for tuning!)
Saved original Param File, ran Methodic Configurator, screenshotted the new params for PSC_ACCZ_I and PSC_ACCZ_P, re-loaded my original param file and loaded it to the FC, manually updated param suggestions in MP, wrote them, and flew.
Param Changes:
MOT_THST_HOVER from 0.2 to 0.055
MOT_HOVER_LEARN from 2 to 0
MOT_SPIN_ARM from 0.05 to 0.04 (now I know why it felt so floaty when landing)
MOT_SPIN_MIN from ? to 0.043 as suggested by AMC
PSC_ACCZ_I from 1 to 0.11, as suggested by AMC
PSC_ACCZ_P from 0.5 to 0.055, as suggested by AMC
And I decided to simultaneously test some filter suggestions Chris Rosser recommended.
INS_HNTCH_OPTIONS from 0 to 6 to use multi-source RPM data from each motor at the sensor speed
INS_HNTCH_BW to 13 because my original is 80 and I’m still not sure how to tune this. AMC suggests something really low like 56 at one step, then changes it back to 80 (maybe because I do something wrong while rushing through AMC)
I think a lot of my AMC troubles are because I’m using a template which has a lot of differences from my own drone and because I didn’t input everything right during the first steps. Anyway, the settings it suggested seem to have fixed the problem!
edit: Just looking at it now in the filter review tool, I think I should revert my HNTCH settings back to the originals. The noise looks worse in the new one. On my next flight, I might increase the Gyro Lowpass Filter cutoff frequency.
and I might lower my throttle expo significantly. On ArduPilot Thrust Expo it looks fairly close to the curve at 0.05
You have the Notch filter reference at 0 so it’s not tracking anything. And I doubt you need the 3rd harmonic and probably not the 2nd. I guess you are not using Estimated Post-Filter.
Ok, so if I’m understanding this correctly, I can leave my other HNTCH settings as they are for now (BW at 13.3, OPTIONS at 6) while changing my HNTCH_REF from 0 to 1.
And change INS_HNTCH_HMNCS from 7 to 1?
I guess I should adjust my hover thrust and give methodical another try. I’d really like to re-adjust the throttle controller values without affecting any other params or calibrations.
Should I be looking at the Estimated Post-Filter view in Filter Review Tool instead of Post-Filter? Estimated seems to always give a much quieter chart.
Looks like I’m currently set up for DSHOT300. Do I just change MOT_PWM_TYPE from 5 to 6, or is there anything else I need to do, like in ESC Configurator or more params to change? (I did that step before traveling twice for a couple of emergencies, so I don’t remember that part very well)
Not sure if I’ll have time to fly it today, but the changes I’m thinking of* for my next flight are:
MOT_THST_HOVER to 0.09 (not sure why CTUN.ThO was so much lower before and higher now)
PSC_ACCZ_I to 0.18
PSC_ACCZ_P to 0.09
HNTCH_REF to 1
INS_HNTCH_HMNCS to 1
MOT_PWM_TYPE to 6
Thanks! The weather’s not great now, but tomorrow, I’ll test it with the aforementioned changes, plus
MOT_THST_EXPO to 0.05 INS_HNTCH_FREQ to 66because I see a noise band starting around there
and INS_HNTCH_BW to 11because Chris Rosser says to divide INS_HNTCH_FREQ by the number of motors when using multi-source for this.
Rosser refuses to correct misinformation. For example, here is AP’s own documentation regarding bandwidth when using multi source:
If you set the multi-source option (i.e. a notch per motor) mentioned below, the bandwidth should not be half the frequency. It should be greatly reduced as more notches cause more phase lag (i.e. latency), thus the bandwidth needs to be reduced to maintain a reasonable phase lag, else oscillation and a poorer tune will result despite the more accurate filtering compared to throttle-based filtering. The suggested starting point is setting the _FREQ/_BW params to 4/1 ratio instead of the default 2/1. This is because for notch filters, a wider bandwidth causes a greater phase lag per notch. You can also use the Filter Tool to check the phase lag for your chosen filtering settings. If your phase lag is higher than it was with the throttle-based notch, then you can try reducing the bandwidth even further, balancing the phase lag and the amount of noise in the system.
I was trying to get a good fit to the blue line on this page using my current params. ArduPilot Thrust Expo
Before, I had 0.6 which I guess I’ll keep. Lowering it seems to increase throttle resolution at low throttles, so I might reduce it more gently, like to 0.45 or 0.5 if I don’t like the next flight with 0.6
