Tuning adventures.
Thought I'd update a bit. I've been flying around mostly with the default PID's, and JoshuaBardwell had posted in another thread that his Proportional gains were upped from the default 3.5 to 5.7:
http://forum.flitetest.com/showthread.php?14248-Cleanflight-vs-Baseflight&p=152710#post152710
I believe his was on a Quad, so I figured it'd be very different on my tricopter. Having flown a bunch with the default PIDs (with the yaw slightly lowered from a high default P setting of between 8-9 to 6) I decided I wanted to play around with tuning next as I feel I can recover having flown in wind, and mixed orientations.
I started out with the process detailed here:
http://www.multiwii.com/wiki/index.php?title=PID
With having the props on, carefully, holding the tricopter in my hands and throttling up until it's light which worked out to just under 50% throttle. Since I was starting with the default PID's, I just started on the pitch axis and increasing the P gain for pitch until it got really hard to pitch forward. I then pitched it back and forth to see if it would oscillate and upped it even more until I got oscillations. It was a little hard for me to tell if it was oscillating until I noted the changing pitch of the motors spinning and also saw on the MultiWiiConf GUI that the PWM values being sent to the motors were hopping. I then lowered the P until it stopped doing that. Normally, you have to lower them just a bit, but because I wasn't identifying the oscillations, I kept going up to high initially, so I had to lower them quite a bit.
I repeated the process for the roll and yaw axis, and eventually ended up with these settings (I and D values were from default):
Next, I charged up two 2200mAh 3S packs and brought them out. Figured with the cold I wouldn't need more flight time. I was right. I also didn't want to lug my computer out, so I searched for a phone app. that would do this, thinking some enterprising individual would have tackled this. Sure enough, I searched, discovered that most people use a BlueTooth transceiver connected to the Flip/MultiWii and use a mobile app such as:
https://play.google.com/store/apps/details?id=com.ezio.multiwii
Unfortunately, I didn't have this bluetooth adapter, but I did have a OTG USB cable for my phone which I use to connect to the 3DR telemetry radio for my APM use. I tried it out, and it worked! I took this out, despite the wind, and tried tuning it. Well, that was somewhat interesting. I'm not familiar with the multirotor tuning process at all. I read up on it, and even watched a youtube video (RCExplorer/David Windestal) but putting it to practice was another thing. On lift off, I immediately got some oscillations, but I couldn't determine which axis was causing this. I tried some forward flight, despite the oscillations, and it seemed to damp them out, but there was still some pulsing in the motors. It could have been the wind that was blowing around 5mph, was my thought. I flew it backwards, and it seemed ok. I also tried the roll axis test by rolling it left and right, and back to center. This time, the motors sounded smoother. Based on this, I assumed it was the pitch axis that had too high of a PID. I flew it back, landed, connected my phone via USB, and lowered it by .1
I kept this up until I got to 5.4 for the pitch P. The roll at 6 seemed ok. I was still getting some pulsing, and when hovering close by, in front of me, noticed that the tilt mounted rear motor was kind of going back and forth with the pulsing, even though when I yaw it, it doesn't oscillate. Well, I land and start lowering that P gain as well. I ended up with a yaw P setting of 3.0 and an I of .015.
By this point, I had used up both batteries, and was freezing, so I returned home to think about it.
Since I had built the tricopter with the FliteTest 13-370 Tilt mount, which will hopefully be replaced by a tuff tilt mount when they start stocking it in the store, it had always had a bit of slop on the tilt due to the design of that mechanism. I started wondering if that might have been responsible for the oscillating tail motor, so I inspected the mount and found the slop was due to the piece that acts as the arm the servo push rods use to tilt the motor mount.
Well, I added some hotglue to it to see what effect this will have on my next foray out -- it did remove the slop, and now the only slop comes from the flex on the push rods:
I just got back from flying it with this modification. It's still windy out there, but I think that's what's causing the motor pulsing. It's not really oscillating when flying forwards or side to side. I somehow confused motor pulsing with oscillation.
Before going out, I searched more, and ran across this thread:
http://forum.flitetest.com/showthre...rameters-for-Electrohub-spider-quad-w-MW-Flip
There was some great information there. While I'm not quite ready to make custom motor mixing values, I saw what Dan recommended about reducing the D values to 0 while tuning. The defaults from RTFQuads were quite high, so I went ahead and reset them to 0 while setting I to a lower value.
I'm now flying with:
Roll PID: 5.5, .0010, 0
Pitch PID: 5.0, 0.010, 0
Yaw PID: 3.0, 0.010, 0
Frankly, it seems pretty similar to what I had before, but maybe with bit more floaty in that it drifts a bit more, but it's still windy out.
