How to use the PlasmaTree PID Analyzer
HTML-код
- Опубликовано: 4 фев 2025
- How to use the Plasmatree PID Analyzer as a tool for PID tuning your kwad. This video covers how to download, install, run and interpret the results.
Github Project Link: github.com/Pla...
Project RC Groups Link: www.rcgroups.c...
FB Blackbox Groups Link: / 291745494678694
=================================================
Want to support me? / uavtech
Facebook: / mark.spatz.79
A great run through of the 1,2,3 process of PlasmaTree -- ruclips.net/video/nXfvbvimkaw/видео.html
Boy, this rabbit hole is getting deep:-)
I was lucky and stingers tuning video came out just a couple weeks after I built my first quad!
Fantastic video! You confirmed that I was "getting it" when working with this software. That Wikipedia animation was key for me. Keep up the good work!
Kroma , thanks!
This was perfect. I have been playing around with pid tuning finally after 1 yr and this made it so much easier!!!! Thank you.
Make sure to give special thanks to Flo on the RC groups channel He developed it. I just featured cause it is a great tool for an easy 2nd opinion on your tune.
Thank you. Been looking for a measured way of tuning PIDs that does not rely on "feel". This seems a good approach.
Did some plasmatree plots and could not be more proud of my tunes. Lol. Thanks uav tech!
Oh man, this is just awesome. Thanks for bringing this awesome tool to my life !
Paweł Spychalski , yeah buddy. Good stuff.
Also check out: ruclips.net/video/nXfvbvimkaw/видео.html
The way StingerSwarm explained it, is the way I learned it too.
My first look at the log-data in the explorer is only stick input and gyro output, as you explained.
The next step is looking, if oscillations on P or D get through to the motor output and then, with a little delay, back to the gyro. If they do, the value is too high.
I'm usually playing it save and allow a little overshoot, I rarely move the stick like a robot :)
There is another way which unfortunately will not work on copters: I and D is set to zero, crank up P until it oscillates. Then measure the frequency and enter this 'f crit' and the 'P crit' into a set of formulas and you get the P, I and D values which will make it a perfect step response.
M0rdH0rst , wow. Why won't that works on kwads (the formula method?). Do u have a link or something to docs on that approach?
Your talking about
Ziegler-Nichols method in the wiki article, right?
I've only found the German version of it, I don't know how google would translate it ^^.
rn-wissen.de/wiki/index.php/Regelungstechnik#Dimensionierung_durch_Probieren_.28Empirisches_Einstellen.29
I've tried it, but the oscillating copter scared the sh*t out of me and it wasn't really close to stable oscillations.
Yeah, that's it. But I guess, it would be quite destructive to a copter ^^
I've used it on power and temperature regulators, but not once yet on a mechanical system.
M0rdH0rst It's my understanding, that we want a little overshoot, to give a little boost and overcome inertia and get the quad rolling, correct?
Great explanation using the gif / 51 images. Thanks for the video!
Oh man I love this stuf! Thanks for showing me and explaining it!
Hisnis fantastic brother. Can’t wait to give it a go. Great job
I am scared to death that if I start with the Plasma Tree, I'll spend all my time fiddling instead of flying - but I really like the way you explain incredibly complicated subjects.
Don't chase the rabbit too much! Have fun out there!! :-)
UAV Tech - Exactly! - I’ve learned so much from you about Blackbox and from others about tuning, that I’ve reached a point where I can “guesstimate” a pretty good starting tune, validate and adjust it, look for motor heat and flight characteristics that match what I want - and adjust accordingly - my goal is not a beautiful log - it is a fun and beautiful flight.
UAV Tech - UAV Tech - Exactly! - I’ve learned so much from you about Blackbox and from others about tuning, that I’ve reached a point where I can “guesstimate” a pretty good starting tune, validate and adjust it, look for motor heat and flight characteristics that match what I want - and adjust accordingly - my goal is not a beautiful log - it is a fun and beautiful flight.
The force is with you!! ;-)
UAV Tech - and with you it is, young Jedi!
Thanks man, very helpful. Helped me understand things a little bit more :)
Very informative as always. Thank you ;)
Thanks just lost my Drone, but as soon as I build a new one I am gonna do this :D
Roee Mazor , sorry to hear about your rig, but way to get back in the saddle!
Awesome video very informative 👍
thanks man..def checking this out
I love how he says "allright!" :-D
Is this a summary of all the recorded flights in the log, or is it just using the last flight recorded in blackbox data?
Last i believe. But there is a better tool now. ruclips.net/video/e4m1wEcc_Yk/видео.html
@@uavtech I actually found your video for that right after my comment and it’s 100x better. Just the visual in the step response tool alone is amazing to see. I have the quad 95% of the way there, time to take days chasing the last 5% lol
Sans aucune hésitation on like :)
Great video! Thanks for sharing!
John E Fly , no problem. Thanks for watching. :-)
Thanks for this; I was wondering when you were going to hit this tool you've been using in other videos. Again another tool that steers you in the right direction for what to target next in chasing the rabbit :)
Can you comment on how the filtering and tuning of a quad affects the power draw of a typical 5 inch build? Is it even something that is measurable when things like throttle input and aggressive moves play a much bigger role in battery life.
Interesting topic. You do see the biggest amp draw spikes in a roll or flip. Of course the weight - and associated P gain - play a factor. Will draw some of my attention to this and if I find anything interesting to report, will throw it up.
Does this only work with betaflight pids or would it work with a flight one log?
Not sure about FlightOne. But should work for Cleanflight, Beta, Butter or iNAV. I have not tried all and someone else asked the same on the RC groups thread. www.rcgroups.com/forums/showthread.php?2953533-Betaflight-PID-Analyzer/page2
Looks like OK for iNAV. I know it will work for Butter and 98% sure Clean will be OK too (they are all forks).
Just give it a go! If you don't mind, post back here for what you get (or on RC groups page).
If flight one logs are compatible and has the same internal P-scaling it should work. Just try it and report ;)
Hey Mark, how about an update on this Video about the new stuff from version .51 plasmatree?
Hey brother, do you know if this can do inav blackbox files as well?
JJRotoGeek , yep. Works on iNAV.
Damn i wonder what im doing wrong, I drag the log file into the window, hit enter, then type JJ for the pilot name (or whatever) and it makes the folder but does not put anything in it.. i know im missing something stupid. The inav files are txt files..
Plot name :) I put mine inav txt - worked just fine. First try!! Files asa big as 2mb take around 10sec to calculate. 6mb file took around 25secs.
could you do an update for version 0.52. New commands have been added, only the folder itself generates to me and nothing more.
will do.
Will this work with flightone
As far as I'm aware, YES. But make sure to tune based on Flo's recommendation which is using level mode and do sharp fast stick deflections. Everyone does full flights, which you can do that later, but for starters follow the tuning recommendation of doing LOS, in level mode, with short fast stick deflections. It limits noise and provides more meaningful, easier to interpret, results. I have not messed with FlightOne yet, but it is on the radar.
@@uavtech thanks buddy
You made it! :-)
Are you sure you are describing the sequence correctly or am I misunderstanding something? (see comments after 7.46 min. in the video)
Yep. We just don't do the Kp half thing in accordance with: "A fast PID loop tuning usually overshoots slightly to reach the setpoint more quickly;..." For Kwad, unlike a valve or something, we can tolerate an overshoot. A valve cannot as it would slam and could cause mechanical problems or hydraulic hammer.
I'm asking because in the vid you state P to oscillations, then you state to crank up D for smoothing and then I for flowing. Is that correct? :-)
I-term for "floating". So if the kwad is drifting, you need more I-term. P pushes to get the thing moving, or stop (the main controller -- like a spring on a car), and D is to dampen (kind of like the shock absorber on a car -- except more advanced).
Ok, I know how to get the graphs, now please show me how to get the correct values for P, I ,D, there must be some software to calculate the correct values, not just by trial and error crashing the drone 100 times as a beginner.
Http://tiny.cc/pidtuning is you best bet.
Pretty much hate the tendency to make user software for quad tuning only for Windows and Mac. Where is Linux version? Especially when backend stuff is done in Python already.
KrotowX , talk to Flo. Source is there too.
Got decoder sources (C sources + makefiles) and so far I believe that it will be possible to compile them under Ubuntu/Mint Linux too. Noticed that Betaflight configurator and log viewer have 64-bit Linux binaries as well. Now the only thing which I need is BLheliSuite32 Linux version.
On Linux you do not even need to install python since it's already there. Just add all necessary packages (Anaconda has everything you need I think) and run the .py!
I would like to also provide stand alone applications for mac and linux but all of this consumes time. I'm still in the process of finding the right balance. Getting feedback, suggestions and whishes of users helps with this. ;)
Runs on mac, you just need to be comfortable with the command line. Working on making a gui for it github.com/Plasmatree/PID-Analyzer/issues/22
i dont think a 5" quad 300-400g w/o battery will fly good with low pids like 20-30. i tried to get that beautiful graph and ended up with something around 15/15/18 (PID) on all axis , but it flew like crap. it just couldnt handle hard maneuvers.
and if you fly aggressively, you will be gettin jagged graphs every time, because it is the real world with gusts of wind, resonances, 'mid throttle oscillations'
, voltage sags, electrical noises and other stuff
The example from the .gif are from tuning a 2.5" quad.
5" will have higher pids! Your observation regarding a nice curve during ideal conditions but crap real world behavior fits my experience. Your P and D should be as high as possible without frying your motors.
On my 3" 1407 build there is nearly no difference between slow flight and fast flight.
On a highly overproped build these two scenarios will differ for sure!
The reason for mid throttle oscillations will be addressed in the next update ;)