PlasmaTree PID Analyzer

[Kevin_Erik]

Anyone try using Plasma Tree to tune your copter?

[biggestRCEfan]

Would like to try it. Keen to know how easy it is to find the step response sequences in the recordings. It looks like a easy, systematic process to tune the pids.

[fomoco_44]

I’ve been plying with it on my quads and Mini + Baby Tri…..

[fomoco_44]

This is my Mini Tri running a high-spec BMS A207 servo:

Attachments:

1. 

   LOG00139Mini-Tri_0.jpg
   

[Kevin_Erik]

Here is my current progress with Pitch and Roll.Will work on Yaw later in the week as time permits.

Here are all the logs from this round of testing.
https://drive.google.com/file/d/1b6465YByQiHCCwqAF9k0QN1qMabMMvaQ/view?usp=sharing

[Kevin_Erik]

Yesterday’s run to work on Yaw while performing short hover flights in-house. (4S-LiPo, 1 minute @ 2k sampling)

Obviously Yaws D-Term needs to be pretty high to smooth out the transition but the Rise time (P) is limited by the overall speed of the Servo.

[Kevin_Erik]

Did a lot of work using Plasmatree to tune and thus far the results have been very encouraging. One thing of note is that these are full (5 minute) flights log taken at 4k log rate. Originally this was tuned for 6S-LiPo use so I guess the increase in Voltage makes the entire system more sensitive to PID changes. Thus the differences and mistakes tend to stand out more.

4S-LiPo

6S-LiPo

Oddly enough, Roll seems to be pretty easy to tune despite the fact that the Tail-Mechanism can induce Roll. Pitch appears to be the problem child as it’s often very difficult to get a sharp but smooth transition to “1.0”. Of coarse extra D-term could be used but I am a bit leary of the possibility of overheating the motors.

Yaw requires a lot of D-Term dampening. Sometimes more than 100 may be needed for it to start to become noticeably effective in smoothing out the transition to “1.0”. Far as I can tell, the D-Term for Yaw doesn’t change the behavior of the Motors, just the Servo. Hence having a high D-Term is possible and in many situations necessary.

[Kevin_Erik]

Seeing a rather large amount of Pitch bounce-back regardless of Terms used or filter settings applied. Curious if anyone has a solution or ideas that can help me resolve this problem.

https://farm2.staticflickr.com/1744/41849129875_de83dfa757_h.jpg

Note : “Pitch” baseline starts at 9/10/0. D-Term 17 shows twice as one is with stock filtering and the other is using Dynamic filtering. Yaw smooths out as Pitch starts to smooth out.

[Kevin_Erik]

OK, I think i have figured this out. Apart for disabling AntiGravity and other things that alter your PID’s. Methodology and consistency are very important if you want to get good results. Furthermore, muting the Yaw terms was required to be able to see good results without interference from with other axis. I ended up working the PID’s in this order: P, D then I. Apparently first having a nice sharp rise-time that flattens out at “1” is required to be able to tweak in the I-terms.

Made a few videos for your viewing pleasure.

I ended up using the following settings for Pitch: 38 / 20 / 36 (Running 6S)

For those using 4S-LiPo’s the PID’s would need to be a bit higher.

[fomoco_44]

Nice work!
I’ve been having problems with Pitch+Yaw coupling with both my Mini and Baby, usually during a 180 pirouette or free-fall, and reading back through the Triflight notes I think the culprit might be tri_dynamic_yaw_boost. From what I understand this factor/ratio directly affects the Pitch PIDsum based on throttle position and Yaw input.

[Kevin_Erik]

As per…
https://github.com/lkaino/Triflight/releases

Yaw boost is only enabled in air mode. Try again with auto leveling enabled to see if your still seeing coupled feedback between the two axis.

Question, which ESC protocol are you using?

[Kevin_Erik]

[Kevin_Erik]

Latest Plasmatree video for version .52 of the analyzer release.

[Author]

Posts