Heavy Lift Tuning Advice

[crayfellow]

I have not had much luck using autotune (APM / Pixhawk) on heavy lifters but that has a lot more to do with my extremely limited time. Generally, when the conditions are right I have a short time for some quick adjustments or a test and I have to go onto the next thing.

Thanks to @Mactadpole I have some great starting params that fly great on one class of craft. I ended up using pretty much the same values on a craft that will be used for some scientific imaging work. The frame is the rock solid, super stable HexaCrafter HX-1000 running U7-420kv spinning 18" T-motor CF blades.

For the most part it flies great, but at hover it's got a definite, persistent oscillation in Loiter which suggests a high "P". The oscillation appears to be "2 dimensional" along both pitch and roll axes, which of course makes sense.

Can you guys check the params and recommend how much of a tweak I should make without losing pitch/roll response and crashing this work of art? I am guessing I should lower the "Rate Roll" and "Rate Pitch" P values, but I am also guessing I'd need to lower the I's as well - and when I do, should I lower those super high Stabilize Roll and Stabilize Pitch P values?

View attachment 27037

There is so little info out there on tuning heavy lifters, I am highly motivated to make some sort of tune sharing online resource of some sort. It was a huge help to have a jumpstart of some working values on a similar craft. But I am thinking even just having something to eyeball when running autotune would be great for a lot of fliers.

@jdennings or @R_Lefebvre any chance you guys can weigh in?

Cheers!

 

[violetwolf]

Have you tried it on stock PIDs? I almost always fly APM Pix straight out of the box with no adjustments other than the ALT Hold which usually ends up around 1.6 or 1.7

Vertical oscillations can be barometer issues. Make sure you have it covered with black foam or black construction paper. The stock shielding is not always adequate.

Maybe save your current config and reset to factory and see if its better? Then a little expo on the tx to smooth out your stick commands.

 

[Pumpkinguy]

have you shimmed each motor referencing off the imu using a smart phone level app? Andrews arms, due to the mandrel manufacturing process have a slight twist, no more than 0.04mm across the length from my experience. This seems tedious but is one step closer to perfection. It showed improvements in my data logs 100%

 

[Hexacrafter]

This aircraft was built up by HexaCrafter ready for FC by client. The motors were all shimmed to perfect at our shop. All it needed was a FC, Batteries & Radio by client. Sorry to say we are not much help with tuning the PX4 FC, but disclosed this before the purchase. We currently only use and recommend A2,WKM and SuperX, and our experience with Pixhawk is very limited. This aircraft is our HX-1000 specialized hexa, and is part of the "X" series of frames that we use for Commerical applications. It is a step above our "standard" Kits and is usually sold as a complete build and RTF and not in kit form. Without sounding too over the top, it is one of the best aircraft I have designed to date. The "sum" of all my years designs & experience. The HO-1250 Octo and HQ series Quad/ X8 started as "X" series before being offered in kit form. Sorry. Go of topic here. Once properly tuned, his aircraft should serve as a solid Commerical platform.

 

[Mactadpole]

I've been working a little with Crayfellow on tuning. I have had really good luck with autotune on these biggish copters followed by a few tweaks. You may need to start with one axis at a time, or do pitch and roll first then worry about yaw later. It definitely depends on your flight time. I've personally had really good experience with autotune on 3.3 and just changing the AUTOTUNE_AGGR parameter.

• AUTOTUNE_AXES allows control of which axis are to be tuned (useful if the vehicle’s battery life is not long enough to complete all 3-axis). “1” = tune roll, “2” = tune pitch, “3” = tune roll and pitch, “4” = tune yaw. Add these numbers together to tune multiple axis in a single session (i.e. “7” = tune all axis)

I also start with AUTOTUNE_AGGR at 0.08 and see what I get. I then move down in increments of 0.005 (so first step down is 0.075). If its too "tight" or oscillates I then go down and autotune again. If its too "sloppy" I go up.

Once I have it to my preferred "sharpness/crispness" I then set RC_FEEL_RP to between 70-80% and add in some expo (details on expo at end of post). It controls how aggressively the copter reacts to small changes in the stick inputs.

I was then working with ATC_ACCEL_RP_MAX and ATC_RATE_RP_MAX in 3.2, but these parameters seemed to have disappeared in 3.3. It seems these are the values left to tweak according to other posts:

ATC_ACCEL_R_MAX
Maximum acceleration in roll axis

Range: 0 180000
VALUE MEANING
0 Disabled
72000 Slow
108000 Medium
162000 Fast
Increment: 1000

It is not recommend going below 36000 and 54000 is a nice number for everyday relaxed flying. Most agile frames, running 10" props, are capable of 100000 to 120000 (on high powered micro quads this could easily double again).

and ATC_ACCEL_P_MAX
Maximum acceleration in pitch axis
Range: 0 180000
VALUE MEANING
0 Disabled
72000 Slow
108000 Medium
162000 Fast
Increment: 1000
These values should be similar to ATC_RATE_R_MAX

I read somewhere recently to decrease the following parameters by 10% after several autotune session and work up or down from there if needed on big copters: ATC_ACCEL_P_MAX, ATC_ACCEL_R_MAX, ATC_ACCEL_Y_MAX, RATE_PIT_I, RATE_PIT_P, RATE_RLL_I, and RATE_RLL_P

Here is what I am running for expo in the Taranis at the moment.
Aileron, Elevator, and Rudder Expo set by three position switch SG
40% for all in middle switch position
and 60% for all in up switch position
I mostly fly in the middle position (40%)