Just did some retuning anyway. Went through the Autotune processes for Pitch Feed-Forward and Pitch Rate-D and Rate-P gains. This time, I got a little larger FF gain. Rate-P has been pretty consistent at 0.061, at least if the 35% of max gain rule is followed. Rate-D typically ends up somewhere between 0.0030 and 0.0040. The heli was underdamped with these rates, so I set the P-gain to 0.061, and manual tuned the D-gain up to 0.0062, which cleaned up the response. At these rates, the Pitch axis is about twice as aggressive as the roll axis. Maybe it makes sense to scale back all of the pitch axis gains together? Or just reduce my FF gain back to about 0.20 (where it was).
. . .
Yaw Tune Check, (Could maybe do this at a higher frequency):
Some things maybe worth noting that I observed:
Pitch axis commands tend to have greater effect on Roll at certain frequencies. Same with Roll effecting Pitch, due to the gyroscopic procession effect.
I learned not to auto-rotate while in Auto-Tune mode. Didnāt hurt anything, its just how I like to land, come in a few inches above the ground then shut off throttle for a smooth landing. When I did this in Autotune, the heli immediately hugged the ground. This is probably expected behavior, but for guys who fly helis a lot, it is common practice to set down with throttle off.
For RC heliās, there is a lot of coupling between the pitch and roll axes both in the command and the response to rates. I am still working on trying to have the autotune one day tune that out potentially using the rotor phase lag.
Yeah, you should unlearn that habit, especially for modes that use altitude hold in the vertical axis. You donāt have direct control over the swashplate. Autotune is especially bad since we discovered the bug. It does not handle throttle off the same as other modes designed to use altitude hold.
ā¦
I think you should go back to the Pitch FF gain of 0.209. The gains at lower frequencies are higher with 0.24.
The tail tracks really well, but it could maybe chill out a little bit. Itās twitchy on the ground, and sounds like itās working more than necessary to track position during flight. What is the best way to āsoftenā the tail response?
I actually like how it handled the way it was after manual tuning. Things felt natural. My Rate-P gains were much lower, Rate-I was higher. Makes for a natural overshoot with slower error response, reducing any ringing. Feels good in manual flight, but I suppose the tight control is necessary for a stable autopilot?
I think you can back off on the rate P gain some. You could try 0.4 and see how you like that. That would help with the comment of working more than necessary. Iām not sure what you mean about it being twitchy on the ground. are you making inputs or the tail is just moving around?
Are you still referring to the yaw axis?
For all modes, you want the aircraft to track the desired attitude very closely. It is important in autonomous modes because the position controller is relying on the attitude controller to set the requested attitudes in order to have the desired velocity and position of the aircraft to be tracked well. I think this is different coming from the RC world because you, the pilot, are expecting it to respond well to your inputs. You donāt care about some command model and whether the actual response tracks well to it. You just want it to be crisp and responsive to your inputs.
You can change the tuning to your liking. It is just important to understand that the desired response must be tracked well by the aircraft. Autotune is a work in progress and I currently donāt have a way to set the intensity of the tune. it is pretty much looking to get the most out of the aircraft without causing oscillations, however this can be challenging with helicopters and requires filtering to do it well. Thanks for trying autotune and sharing your data. It helps me learn about tuning different aircraft and a 900 size heli is definitely one of the bigger ones.
I am not making any inputs. The tail just is responding to vibrations and wind. It oscillates a little on the ground just because it is fighting the ground.
I am referring to all axes. This was the general difference between my manual tuning, and the auto-tuning.
Thank you for your support on this! Iāve been opening up my college books on Control Systems trying to get back into the āswing of thingsā. Thereās a lot to understand though in applying it. Itās a topic I enjoy, but definitely needed your experience to get through the tuning.
I did some Auto flights today. That may contain some useful data as well:
For future reference, this tuning was for the RHE1E29XW E1 900 PLUS KIT (Tri-Blades Rotor Head) helicopter from Align. All flights and tuning were done with 2x 6S, 16000mAh Li-Po packs. No payload was attached during tuning. And a light camera gimbal payload has been added during flight.
If ever I fly in Manual mode however, the pitch axis has way too high a gain. The PIDP Proportional gain is currently set to 0.054. To improve the manual response, I need to bring this gain down about 10 times!
Looking back in our conversation above, it looks like I did alter that pitch gain from what the auto-tune arrived at. Iāll go back and return the gain as the auto-tune had it.
The flight log for the manual flight is here
During this flight, I was adjusting the P-gain of the Pitch. Brought the P-gain down to 0.01 and it still wasnāt quite low enough to remove the oscillations, though it was much better. Could probably go down to 0.005.
My latest auto flight is here
Pitch seems to be tracking very well here.
Probably not the best way to deal with this. Return your gains back to what they were.
Looking back at the conversation, I pulled and looked that the post tuning frequency sweep. The pitch axis has a pretty big hump in the frequency response explaining why you are seeing some overshooting in the response. Here is my advice. First even at low gain, I think the FF gain is too high and overdriving the response. So drop the Pitch FF gain down to 0.18.
Next to help remove this hump, I would suggest we use a filter. I would suggest you change the Pitch FLTT filter to 2 hz. This should help provide a better response without the overshoots.
Thank you Bill,
I made these changes as you explained, it seems to have improved the pitch response. Auto modes still cause the heli to swing back and forth slowly a little bit. Iāll probably run through the steps to tune pitch rates once more when I have time, ensuring that the FF is set appropriately before hand.
