I have set up and tuned a 500 electric using CUAV Pixhack v3 with success earlier this week. This gave me the courage to progress to an 800 size gaser and Pixhack v5. I am following the Wiki to tune but the tuning end with a forced termination as the heli started to oscillate violently.
I completed the rudder tuning and VFF. The heli was manageable and it seemed that the rates were tracking nicely, apart from being a bit tail heavy. However, when I started to adjust the D gain to 0.003, according to Mission Planner, the craft started to oscillate heavily.
What am I missing?
Is it possible that the D gain can be 0?
Is it possible that the scaling between Pixhack v3 & v5 is different?
I am attaching a link for the log file. Any help will be appreciated.
@StefT. Most of the time, P and D gains will be small. D gains can be 0 or less than 0.001. You can enter values out to the 4th decimal place if need be. The P gain is typically less than 0.05 on most helicopters.
Be sure to set your VFF appropriately. The I gain is where you will see the aircraft more closely follow the desired pitch and roll angles, however your ILMI must be non zero. Less than 0.1 is desired for that parameter.
Hope this helps.
I have successfully completed the setup and 1st flight. I still need to add the head speed governor and fine tune the PID loop but so far I am extremely happy.
A great thanks to the community and everyone involved in developing the code. I was flying a DJI Ace Waypoint unit and was “thankfully” forced by the lack of support from DJI to migrate to Trad Heli last week. It is something I wanted to do for a long time but just did not had the courage to do.
Would be great to get some more feedback on what went better and where can we improve, so if you are inclined to give such feedback, documenting your migration would be wonderful.
I have governor code for ArduPilot so you don’t have to add an external governor. The binaries are still built on Copter 3.6.3. But I can build updated binaries on 3.6.5 tomorrow. All you need is a hall effect sensor to plug into your CUAV v5 controller.
Thanks. I have 3.6.4 installed and already installed a T-Rex nitro hall sensor with the wires crossed. I am planning to set it up and test today. Will 3.6.4 be ok? You also mention to set it up on aux 5. Can you confirm that A5 is aux 5? I am having difficulties to find the right documentation for v5.
The V5 is the same as any other controller for the rpm input. Yes, the A5 pins is the Aux 5. The governor is not in the main code yet. But there is a thread here when it was introduced for user testing and it explains how to set it up.
3.6.5 is still a point release candidate #2, but I think it will be going to official point release soon. There is a link there to where you can download the binary with the governor (still built on 3.6.3). But I will update that today to 3.6.5 and make a post on that thread to indicate that the binary is available on the latest point release.
Yes, I have the sensor installed and it is working, except that I will need the correct binary. I will wait for your update of 3.6.5rc. I am looking forward to test it.
@ bnsgeyer. I almost finish tuning the helicopter after some minor issues related to the hard landing. She flies well but I am struggling to understand the ILMI and IMAX.
I need some clarification regarding the IMAX value. I have set the IMAX = 1 and followed the instructions. My PIDP.I seem to be very low and only peaked once at 0.03 and them mostly at 0.01 or below. According to the wiki page “Set IMAX for 0.1 above the maximum value”. Does this mean 0.13 or 0.04? Also, it seems that Pitch and Roll are split after reading the following “You could do the same for the roll axis but typically 0.4 should be sufficient.” Where do I check for Roll? Do I also use the PIDP.I as a reference?
I also find it difficult to understand how to tune the IleakMin (ILMI). What values do I need to look at? Is it the Roll vs DesRoll but in the Stabilized mode to check how closely they follow? I have read your explanation and reference to Rob’s analogy referring to the cup of water, but I can’t seem to find more data or methodology to properly tune the ILMI.
@StefT glad to hear that you were able to work through most of the tuning using the wiki.
So the IMAX really only applies to forward flight with ground speed above 5 m/s. So when you conducted your flight, did you go into fast forward flight? You will need to do that to properly set IMAX for pitch. For hovering flight and ground speeds below 5m/s, ILMI will be the parameter that will govern the max integrator value. this is due to the integrator leak that is turned on below 5 m/s. So the I term or PIDP.I is not allowed to build because it is immediately leaked off. the ILMI parameter is used to not allow the PIDP.I term to leak below that the ILMI value.
I would recommend setting the IMAX = 1 and ILMI =0.1 for a test flight. For the flight, hover and do some slow flight maneuvers and then fly out to your max speed that your are comfortable with. Come back and pull the data. Look at PIDP.I and for hover and low speed set ILMI 0.02 above the highest value and for the high speed flight, set IMAX to 0.1 above the highest value.
For roll, you would look at the PIDR.I value in the logs. We give a standard value for IMAX because you typically won’t be able to judge that well with doing a flight. So we just say to make it 0.4. for ILMI, set it to 0.1 and go do some low speed flight. come back and pull the PIDR.I signal and see what the highest value is and set ILMI to 0.02 above it.
Bill, thank you for the clarification. It is finally coming together for me. I did a test run on the 500 electric today to practice and I think I have got it. It is a pleasure to work through the Wiki as there is a clearly defined process to follow. No guessing and wasting time to tune. Great job on making this an enjoyable process. There are still a few technical terms I want to work through and understand but at least both helicopters behave very predictably.
Glad to hear that you found the wiki useful. @ChrisOlson and I worked hard to get it to that point. if you have any suggested changes, feel free to let us know.