Plasmatree PID Analyzer | SETUP & TUNING A QUAD

Uncertain about what to buy? Check out THE ULTIMATE FPV SHOPPING LIST: www.fpvknowitall.com/ultimate...
SUPPORT ME ON PATREON at: bit.ly/2Ep5Sia
THIS IS MY FULL TIME JOB. Here are other ways that you can support me: www.fpvknowitall.com/support-me/
---
You're about to learn how to setup Plasmatree PID analyzer and use it in tuning a quad.
Plasmatree PID Analyzer is here:
github.com/Plasmatree/PID-Ana...
Blackbox Tools are here:
github.com/cleanflight/blackb...
My PID Tuning Masterclass is here:
• PID Tuning Masterclass

[@53roger]

I have only been flying for about 6 months and this week installed a new FC and I went from BF3.2 to 3.4...WOW what a difference on stock PID's. After getting comfortable in ACRO, I had just got to the point were I felt I was ready to start messing with the PID's but BF really helped me out. I still have some prop wash to deal with but hey, I'm well on my way to a decent tune on my quad. This is the second time watching this video and your advice for beginners ...just go fly... is well taken. I'm going to go fly and think about PID's this winter. Thanks Josh, you 'da man!

[@plasmatree5829]

Great work, awesome you made it! Here are some comments: - As you concluded: The stock tune is already good, at least for 5" (different story for smaller quads...). Your final tune needed more D to compensate for overshoot. Overshoot is an oscillation that might be excited/amplified in propwash scenarios. - If you already have a clean but slow step, increase P and D proportionally. This keeps the shape but gives faster rise. Here: roll P40/D30-->P60/D45 - 13:00 the slow decay after the slight overshoot is caused by accumulated I-error. As you increased P less error is added to I during rise and the pronounced overshoot adds negative error and thus faster I-term reduction. - 13:30 the sharp rise in Yaw results from the acceleration of the moving mass (motor/prop) while the "slow" part is connected to the difference in drag (as the props are spinning at different speeds) - 14:30 The >500 lines are based on a smaller amount of data (look at the top graphs, >500 happened seldomly). That's why they are sometimes so jittery. Nothing to worry about. A general tip for not getting fooled: Never believe a single measurement. Things have to be repeatable to be sure, that's the credo of any science. ;)

[@JoshuaBardwell]

Unfortunately it wasn't possible to increase D much further because of noise so this seems to be the limit of this tune. At least until filters were adjusted.

[@Giantmdgt]

Super helpful thanks!, When you say 13:00, "the slow decay after the slight overshoot is caused by accumulated I-error. As you increased P less error is added to I during rise and the pronounced overshoot adds negative error and thus faster I-term reduction. " does that mean that you need to increase I? or is that more of a filter issue?

[@plasmatree5829]

It's a neutral observation (to be precise of 13:20). I can't really say if it's too much I, especially because if P and D are set properly, you probably don't need much I.

[@fotoamgamgfoto3695]

​@@plasmatree5829 hi, can a D be mcuch better than P or I? my quad has a slow response on P but overshots, but I have lots of filtering and slow loop rate as it is an older FC 2/1khz and F3 chip. I tried PIDS 50-60-30 then 25-19-21 similar to the one son your example at github for xyour 2.5inch quad, I got much less Jello, I want to tune it for less Jello but the response is much slower. I am thinking i raising PI back and increasing D to have much higher values than PI.. Or is it a no-go?

[@bicky420]

Great video as always. Honestly, you could run a 2-hour technical quad video and I would sit through it. Always learning something every video... Rock on.

[@drumrboynoid]

This is so weird. I just sat down to finish a school final on tuning a simulated PID system within LTSpice and I was just about tp start looking up some online examples on how to do so, and this video popped up. I'm a subscriber so it was just on my recommended videos. Its so strange how that stuff happens. Thanks for the video Josh.

[@youeladamas7840]

lol the synchronicity of fpv

[@jaraandres]

So far I've of the most interesting videos you made lately

[@swoopaf8536]

Just figured out why I love the sound of motors in acrobatic flight. It reminds me of the fights scenes from those old king-fu movies I grew up on. Thats awesome. Great video, will be looking forward to more on this tool. Thanks.

[@benmoeggenberg7846]

PID tuning, how I do it for combustion engine idle control. 1. Set D and I to zero. 2. Start low on P and increase until you see three to four oscillations. 3. Add just enough I to take care of any steady state error. 4. Add just enough D to reduce the P term overshoot. That is a pretty classic PID tuning approach, and I got a good chuckle with your comment that "classic, by the numbers" is not always the right answer. You are correct, this approach will get you in the ballpark, but not into your seat. Your initial screen shots with the overshoot that was hanging on for awhile looks to be the result of too much I or Betaflight winding the I when the controller is saturated. Trying to compensate by cranking P to resolve an I issue will add to the instability, which most likely is why the second video was showing some bounce/propwash. It is really beneficial to plot all of the gain terms separately and it will give you a lot better insight to which portion of the controller that is giving you issues.

[@ArchmageAU]

Been waiting for a video like this for a while. Thank you. As you said. BF default PIDs are good. Results are subtle. So just go fly.

[@jester3265]

JB......OMG!!!! Those two flights were friggin Amazing!!!! Your piloting skillz were Sick!!! I agree with the propwas oscillation showing more on the second flight, those default PID's were great. Slight propwas oscillation, but nothing a beginner pilot would worry about. All in all a nice bit of kit and another great Vlog.

[@i4dat]

Great Vid Joshua, thanks for explaining this tool. I would also like to see more videos of you making further adjustments to the tune using this tool as well as it's other features that you didn't touch on. As always, thanks for doing what you do, it is greatly appreciated.

[@markhaasen9968]

Its like you can read minds.. i just started using plasmatree about s week ago and was thinking hey Joshua Bardwell should do a video on this.. great job cant wait to see more

[@TMacFPV]

Joshua, outstanding video! Thank you! Your summary at the end was spot-on! It DOES appear to be a useful objective tool to visualize what's happening with PIDs and Blackbox and refine tuning to your specific quad. Yes, the default Betaflight PIDS work well, however, there will always be slight differences for each quad. So, if we can get a tool to better tune our own specific quad, should we desire to do so, that's great. Yes, PLEASE more videos on this PlasmaTree PID analyzer and how best to sue it optimally. Try out more D on your quad you were using and show results. i bet, you can simplify/optimize PID tuning using this PlasmaTree tool. Just need to refine the process using it. Thanks again, great work!

[@WrongWayFPV]

I’m going to try this ASAP!

[@TheDirtyHarry]

Love your vids dude...so interesting!!!

[@MrFlyfishy3]

I’ve been playing with Plasma Tree for a couple weeks. I had a lot of trouble getting consistent data. I’m new to FPV and don’t quite have the skills to do fast rolls, or snappy flips consistently. After A LOT of logs in ACRO, I decided to try in ANGLE Mode. The logs smoothed out and became VERY consistent, with only minor tweaks (ie. I + 2) showing in the graphs. Being that it was in ANGLE mode, I could also fly indoors (couch tuning!). I found that logging at 8k for 45-60 seconds in ANGLE mode, doing quick Roll, Pitch and Yaw inputs several times (ie. 5 - 10 pitch forward, pitch back, roll left, roll right, yaw left, yaw right) would output nice noise and response charts. I would make changes, per the Analysis and repeat. Once the settings looked good I took it out to fly. Little TPA here, touch of AG gain, Setpoint Weight to 1, and it flies awesome! My Armattan Rooster (2205 BNF CL_RACING F4S with DAL 5046Cs) flew great out of the Box. But, now it’s really nice! And no more Nose-Dip after hard punch-outs or hard throttle-cuts.

[@michaelpetale6872]

Loved the example explanation on it!:) car!👍🏻

[@3-danything977]

Great. Info buddy i have learn so much from u

[@ralphnicholls754]

Captain Robert & Captain Bosh thank you for your interesting and entertaining build tests. You are correct with your criticism of Dynam planes. They are gaining a reputation for terrible electrics. This is despite efforts to upgrade with Tomcat motors and Skylord esc's. Their retracts are un-reliable despite their best efforts. They are marketed as a cheaper aircraft, but an increase in price to supply reliable electrics would increase their sales of otherwise great aircraft.

[@DuckyRC]

Yasssss... 3.4 has set some interesting defaults for filtering.. and they are butterful :P

[@unitedelectrician4334]

thank you I learn something today.

[@nathan1sixteen]

Ooh, that new intro, not a fan. I actually liked the "I'm Joshua Bardwell, and you're gonna learn something today"

[@Red5FPV]

In regards to what was mentioned at around 6:15 about switching to PT1 and filters off (Dynamic filter on of course). This is what I ended up doing and my quad feels so much better at the controls now. I have noticed that the more beat up my motors have gotten, the warmer they've been getting. I'd say this filter set up is fine on a fresh build but you may want to start adding notch filters after a period of time.

[@cameronsmith5968]

I'm actually about to go into an exam for digital control systems design. I don't think any of this will be useful for that, but at least I know I can try a new tune once it's over. Also, maybe the betaflight default pids felt great because it's something you've experienced and are comfortable with.

[@Dorkyand40]

Yeah. Go fly. Starting to preach this more and more myself and oddly enough to myself. By the time it’s really tuned it’s time for a new build anyway.

[@Umsterdam]

If anyone else having problems getting PID analyzer to run , I had to "run as Administrator". Right click icon and select above. Thanks for the video Joshua

[@Politicallyhomeless957]

I like tuning it’s just a little calculus:) Nice vid JB good info

[@theflyingace4148]

Im more in collocating and centering the weight of the aircraft than tuning, but I think Plasmatree is great. It feels much more comfortable to jump in PID with it now...

[@JT-jy2sy]

The plasma one in one turn at least I could hear vibration. It is sharper. Uav tech interviewed the developer and he recommended to turn off setweight and other scaling parameters. My understanding is Plasmatree uses the real Pid error. For example P error in bbl that is already scaled , which is P error * K ( K the value you enter in the config). I have been using it a few weeks, it helps me to validate in a more scientific way that my tune is good. But again if you feel your quad is flying great imo that is the best tune. It does not matter what Joe or Bob says.

[@JustInvertedFpv]

It looked a lot less mushy on the plasmatree pids. It seemed like a much more direct connection to your stick input. My pids are more similar to what you ended up with on plasmatree with the exception of how high you ended up going on yaw and I run I at around 61 on pitch and roll. The slight propwash you got could disappear after running some packs on those settings and getting used to the throttle management and maybe some fine tuning the pids up or down slightly.

[@wlgrd7052]

Thanks UAV Tech :)

[@Sully365]

3.4 without notch filters brings my motors down at half the temps of 3.3 with notch filters. it's amazing

[@djsteve2014]

Some overshoot in a pitch or roll may feel better in the sticks, e.g., faster response, but results in decreased linear acceleration. I didn't realize how much until I began using the PID analyzer, but it's quite dramatic.

[@pablofpv]

Nice explanation, but I'm not so sure I would use it for tuning? May be as a fault finding tool? Once I get to the point where my video looks reasonably jello free, then any tuning is just a personal thing, and its easier to do it by feel, I just think trying to tune a quad this way would be a bit of ball ache?

[@RitztheDJ]

Is there any chance you could do a video on how the i and d terms are changed. And how jt does with raceflight

[@MusicOnlythebest]

How to analyze data from a black box in BFL format? My betaflight does not create a BBL file. After inserting BFL I get only an empty folder. After using blackbox_decode, it also does not receive a BBL file. Please help me Joshua. This is important to me because my pids that suits me are relatively high compared to other people.

[@fotoamgamgfoto3695]

hey both flight are much better than I can do :) is there a reason why my response time graph shows much slower response? is it related to turned on filters or the 2/1khz loop rate of my F3 FC?

[@pauljerome79]

I think it would be interesting to analyse a quad that you've already tuned by ear and feel and see what the graphs look like, then others could tune their quads using this tool, and end up with a similar tune to your hand tuned quad...

[@balint133]

Interesting, what motors are you using? I've figured with powerful 5" setups, no high P/D term necessary. Plasmatree rise times with minimal D are pretty much the same with 35 or 60 P for those, except 60 shows excessive oscillations which needs ridiculously high D to dampen and you'll likely end up having propwash. They really locked in between 35-40 P on roll and 5-7 higher on pitch. That way, it's perfectly enough to run 20-27 D on both axes, no need to fry the motors and tune stick feel with FF. I used to try to raise P but could never get to a good tune, it was either too twitchy or suffered from terrible prop wash or burning hot motors. With my most recent 10:1 thrust - weight ratio build, I simply dial down the defaults and it's like flying on a rail :) Any thoughts on this, Joshua? Am I hallucinating, or in fact less P/D is enough for high power setups?

[@LeonVuksic]

I would say after the tune vs default betaflight, less vibrations when you flip and the quad looks it flies more locked in.. but that's only my noob opinion.. I think this is great for begginers and will fly better than with the default ones

[@MkmeOrg]

Might be just my ears or the way you flew but def sounds like a bit more oscillation in the second. Maybe some D would bring it in line and make even better. Maybe original without the increase P and D is better too ;) Great video Josh

[@MCsCreations]

Yeah... I definitely agree. The default PIDs were better - but not by much.

[@idfpv6510]

Cool!!!

[@wudz_fpv]

I'll just say you really don't want any overshoot in the step response if it can be helped. The best PID's will give you an immediate jump to 1.0 and then stay completely level. It's suggested if you can get the blue and orange lines to reach 1.0 and stay level, you should increase P/D at the same ratio to quicken the jump from 0 to 1.0. My current step response is very close to this and it's the best my quad has ever flown. Overshoot in the graph = overshoot in the PID's and requires less P or more D. Also, once the majourity of the graph is level at 1.0, but there is still roughness in the lines, Iterm must be adjusted to smooth things out. If you check out the RCGroups discussion there's some good conversation going on about it.

[@dirwood4365]

Wüdz_fpv What thread? Is it a dedicated Plasmatree thread or just a pid tuning thread?

[@wudz_fpv]

www.rcgroups.com/forums/showthread.php?2953533-Betaflight-PID-Analyzer/page17&page=17

[@dirwood4365]

Thanks!!

[@AndyKayJah]

Thx for the sum up... I was suspecting something like this,especially after the second flight example had audible oscillations on sharp stick movements

[@SlipperyFPV]

Interesting tool, but I would do my head in with it, I don't like tuning at the best of times. I can't wait for a descent auto tune feature... BuF team talked of it in the future sometime.

[@prankmypants]

Yes this. Autotune.

[@johneod7821]

Auto-tune would require MANY flights to do a real tune.

[@SlipperyFPV]

John Eod dRonin has had an auto tune feature for some time, people say it's good. RF had as a mode in earlier builds. Not so many flights are required, just hovering and performing certain maneuvers.

[@basstrom88]

Arducopter can complete an autotune within 10 minutes of testing - and that's all three axes

[@mikeboehk2307]

Joshua, what were the filter settings for the default PID flight? Did you do the same filter setup you mentioned for the Plasmatree trial? I would have expected more prop wash oscillation with everything default, including the filters and since there didn't seem to be much, I just want to confirm how your filters were set.

[@JoshuaBardwell]

Filter settings were identical between the two: no notches, PT1, dynamic filter enabled.

[@daniellewis8606]

Hey Josh, any chance you could do a video on how you go about tuning your filters? I would really like to learn how to know for sure, that I either have enough filtering or that I’m not wasting unused filtering on delay. Is it easy enough to log a debug, look at a graph and know what cutoff frequency to set each filter at? Really struggling to find info on the process. Maybe even another live stream with Mark could be good I think you did an excellent job at relaying the info on PlasmaTree in a way that was easier to understand. Regards Dan

[@JoshuaBardwell]

Basically, raise the lowpass filter values, and if your motors get hot then, stop. A quad with very little vibration can have gyro lowpass up to maybe 150 or 160 Hz, much higher than default. But if you bend a prop, you can smoke your motors, so be careful.

[@poweroftruth9336]

So after tuning with the program did you turn your anti-gravity back on and your set point weight back to one ? Thanks for the video

[@JoshuaBardwell]

Yes.

[@MattVentureNL]

JB: if you want to tune out propwash; would plasmatree tool help you with that or not? The tool works great for me for low frequency rc_command tracking analysis (step response time), but propwash seems to happen more at mid/high frequencies.

[@JoshuaBardwell]

I think in order to see the step response for a propwash situation you would need to have a log consisting primarily of propwash. So I don't think this is going to be super helpful for tuning propwash. But an optimally responsive system will handle propwash well.

[@jorgeromero4680]

does this work for betaflight 4.1?

[@jasondevine6014]

I was hoping a method of using oscillating frequency of the quad from too high p gain and then calculating the I and d gains as used in other pid tuning fields would be incorporated into some software like this.

[@Robertfarabaughjr]

I would really like to see the graphs showing the stock P I D's; Plasma trees tune; and your to feel tune.

[@roystoys3239]

Is there a resource where beginner/newbie quad flyers can download betaflight setups that have tame/easy/gentle settings for learning to fly a racing/whoop/mini quad without crashing all over the place? What us beginners really really need are "loadable" betaflight configurations with beginner, intermediate, and advanced settings (so we don't have to spend a 2-month learning curve about betaflight. Considering our current situation (winter, corona) these might also be broken down into indoor and outdoor. This question is being posted for several of us who are experience RC flyers and also have flown "camera" drones, but want to challenge each other with indoor and/or outdoor competitions, but are really having problems with over-controlling, holding altitude, making coordinated turns, ad nauseum. Perhaps you could create a video addressing this issue? P.S. I have an Emax EZ Pilot which I love and flies great around my house. But it is in no way windworthy. I have ordered an Arris X110 HD for outdoors, and need to progress beyond the training wheels of the EZ Pilot. THANKS!

[@mrozu1337]

Currently, your Plasmatree tune is better at tracking your stick inputs, while BF defaults are better in rejecting disturbances. If you want to balance the both you'd really need to play with both P and D and ensure that the whole quad is able to handle higher control bandwith... endless effort and in my opinion the numbers don't matter at all if you don't like how the quad flies and handles propwash, jello etc.

[@MeanTheBean]

can you do this with flightone/raceflight?

[@N4505]

do windy conditions screw the readings up? Should I definitely wait for a very calm day?

[@JoshuaBardwell]

Wind would definitely add some noise.

[@twistinprops604]

Filter tuning is the new PID tuning

[@benaresq]

It's easy to do the test flight LOS in angle mode. It's only 20 seconds or so per test, so with LUA scripts, you can run a series of tests without even bothering to put on your goggles.

[@JoshuaBardwell]

Tell me more about this lua script. But email me because I'll never see the comment response because YT comments suck.

[@ricyjo]

Vbat PID compensation: does this require that the voltage settings on the battery tab be set so that it reads voltages accurately?

[@JoshuaBardwell]

Yes.

[@roninfpv2402]

Need more on this seems like this could be a great black box assistant/teaching tool for people and seeing a full tune from it would be nice. Also more about the full line output would be great not just the initial movements of the sticks, and something more than just upping P a bit you did not even show the end results, did you go up more on P then fly and find they were too high? Then dial p back this was just a tease really would love to see more on this Please!

[@JoshuaBardwell]

I did show the end results when I flew them back to back.

[@HiTek_RedNeK]

I'm ready for Betaflight to code an auto tune feature comparative to dRonin.

[@thespencerowen]

Do you need to have rates above 500 DPS to fully take advantage? What if your rates are closer to 300 like you recommend for racing?

[@JoshuaBardwell]

Then you won't see the orange line but it doesn't apply to you.

[@brisketbaron]

Joshua, you should teach folks how to use git. It's so convenient maintaining these fpv related repo's locally. Maybe just using something like gitkraken if ppl don't like cli (if time permits, I can tell you're busy af).

[@brisketbaron]

a couple of one-liners if anyone cares. Windows? get git. first get chocolatey (chocolatey.org). Then *choco install git*. Download? *git clone **github.com/betaflight/betaflight***. Update? (while in the dir) *git pull origin master*. Or maybe just *choco install gitkraken* ... -` P

[@Rasgore4]

could you do a video on this for BF3.5?

[@cristayy6586]

hey josh love ur home

[@GeoffreyMohr]

Any idea why my cmd box closes shortly after starting plasmatree. I have tried .51 and .52 and in both the cmd box closes milliseconds after opening.

[@JoshuaBardwell]

Try running it from the command line and the window will stay open to allow you to see the error message.

[@GeoffreyMohr]

I tried but it gave me other errors. (I can try send screenshots)

[@MrFrobbo]

If you can't set your debug mode... Betaflight Version 3.4 As a part of the filtering overhaul, the names of the debug modes available to log filtering / tuning data have been improved NOTCH (gyro data after scaling, before filtering) is now GYRO_SCALED, GYRO (gyro data after all filtering has been applied) is now GYRO_FILTERED set debug_mode = gyro_scaled

[@MrFrobbo]

How did I miss this on the video!! Cheers JB, really interesting as always and yes, I did learn something :-)

[@jackdaniels7913]

That took the first turn to tell I like the sound of no1 over no2

[@boriquadfpv7677]

Do you need to debug when using ButterFlight with Helio?

[@JoshuaBardwell]

Doesn't matter, because you can't. Helio can't expose the pre-filter gyro data because of the filtering co-processor.

[@exinfernnus]

Did you raise D setpoint weight back to 1 after analysis?

[@JoshuaBardwell]

Yes, on both tunes. Also re-enabled anti gravity and TPA.

[@dave123443210]

Did you put tpa and setpoint back before second test flight? Otherwise hard to compare flight performance. Also curious to see how to use D gain to tune with the tool PS agree with you default pids are best for most people. Exception is when u have a quad with a bit of a noise problem (especially different size quads, maybe old motors etc)

[@JoshuaBardwell]

TPA, anti gravity gain, and setpoint were identical between flights. Only PIDs changes.

[@dave123443210]

awesome thanks for clarity- appologies if i missed it in video :)

[@OrhanKarsligil]

PID controllers are normally tuned in frequency domain and it is assumed that you never hit constraints, meaning limits on your inputs/controls. When you hit constraints the problem becomes extremely non-linear and existing methods do not work. If you tune your controller in almost idle flying with little shakes and then go ahead and fly with full throttle and 100% roll, pitch and yaw input you will not see much of a difference between large amounts of PID parameter ranges. If you do not saturate your control inputs then this method would create the best frequency response in terms of set point errors.

[@briancourtice2858]

It's still a trial and error process. You look at the graphs and say, "Hmm..more P less D maybe?" Then you fly, collect fresh data, take another guess based on the changes you see, rinse and repeat. Not much different from doing it based on observing how the quad flies between changes. I guess the advantage is that you can see the graphs getting closer to an ideal tune. If you aren't an advanced pilot you might have trouble observing flight and analysing what changes your quad might need The obvious next step needed is firmware that can autotune the quad for you, or software that can interpret the graphs and tell you what changes are likely to improve the tune.

[@SirShadeSquamous]

This kinda feels like a manualized autotune. With that said, why doesn't betaflight have an autotune yet? dRonin has always had one as far as I can remember and to be honest its pretty amazing. Have you tried it Bardwell?

[@JoshuaBardwell]

No, I haven't tried it.

[@SirShadeSquamous]

I know open pilot stuff is out of the norm these days for the majority of people but if you have the time I think it would be worth it to check it out. The guys at dRonin actually have some interesting plans for their firmware in the near future which includes an LQG controller as an alternative for PID loops and a artificial motor kV limiter for running higher voltage batteries.

[@tscc]

There's a preview of the next release on their forums, it includes the LQG controller already.

[@JaimieFryer]

You can hear the difference in the motors, especially after the rolls after punch outs the plasmatree PID oscillations sounded much closer together and less amplified in most of the flight. Not sure why you'd overtune instead of increase rates and expo. Surely a quad that flies exactly as intended should have its feel changed by rates and expo instead of overcompensating in the PID controller?

[@JoshuaBardwell]

There is a difference between a quad with high rates and a quad with high P gain. A quad with high rates and undertuned will still feel imprecise.

[@glenby2u]

defaults seemed slower to react, less crisp. also less oscillations. second set were sharper to react and motor sound had that "snap". more oscillations. as always, I think it will down to what each person wants with their craft. drib vs steele tune would be much different.

[@wooler78]

To much of a mess on for me but interesting I just change a few things from the stock pids then in the field if I notice out off with the quad

[@tratjac4605]

way above my pay rate i guess i will just go fly and try to be a better pilot 👨‍✈️ 🏁👍🏼

[@theedc]

My best friend is a extremely good pilot, he is a sponsored 3d heli pilot and a world class 3d plane pilot and FPV racer he is crazy good LOS as well and also carries a jet turbine wavier holder... im brand new at this so he will help me tune and all that but i would like to do it by my self. I have watched all your videos on tuning but im kinda scared of tuning.. do you think this would be a good fit to help me be able to get a good tune by my self???!??!

[@JoshuaBardwell]

Like I said at the beginning, probably you should just run defaults and focus on becoming a better pilot.

[@denilsonguerra1]

Hello, Joshua. The FC I use logs to card. Use Betaflight 3.4 and the log files are * .bbl and not * .bfl. That way when I run the PID-Analyzer the result is not as reported in the video. How do I solve this?

[@JoshuaBardwell]

Try just changing the extension? I'm not sure to be honest.

[@denilsonguerra1]

Already tried. The error remains when decoding and does not generate all the analysis screens.

[@JoshuaBardwell]

What is missing?

[@denilsonguerra1]

Not decoding completely, the result is only two screens.

[@JoshuaBardwell]

Only two screens on the Noise display? That would be caused by debug gyro not being turned on.

[@hmarquestt]

Is possible use plasmatree with FlightOne?

[@JoshuaBardwell]

I dunno. Try it!

[@tweekFPV]

👍

[@ilikeairsoft00]

What debug mode should i use with the helio spring running odin?

[@JoshuaBardwell]

Debug notch olin Betaflight 3.3 and earlier. They changed the name on 3.4 though.

[@ilikeairsoft00]

What about with Butterflight 3.5.2? Would I still use notch?

[@ilikeairsoft00]

I used debug notch and it seemed to work but I got some really strage looking graphs for the step response curves for what i thought was already a pretty good tune so I was wondering if maybe I need to use a different debug mode for the Helio Spring FC running Butterflight 3.5.2.

[@GoatZilla]

Yaaaarp

[@carlosfpv]

They kind of looked the same. I'll probably end up checking it out at some point.

[@michaelwilkes0]

Someday someone will build auto pid tuning into the flight controller. People have tried, but its not quite ready yet. The proper way to implement that is to have the flight controller analyze the physics of the quad, but then the user tunes some slider bars of what they "want". Then the flight controller takes what the user 'wants' and adds that with the analyzed physics to produce the pid tune.

[@dlsspy]

It's been ready and working remarkably well for years, just not in the multiwii/baselight/cleanflight/betaflight/raceflight/butterflight world. github.com/d-ronin/dRonin/blob/next/flight/Doc/Autotuning%20Derivation.pdf In practice, it also has a couple inputs for whether you want it to feel a little more accurate or looser. It works both on the PID controller and the LQG controller.

[@michaelwilkes0]

neat I didn't dive all the way into the math but I glanced through it. Looks interesting. I did notice that they are tuning the filtering also, which is nice. Also, I have seen auto tune code running on a quadcopter before. You fly up about 20 ft, flip a switch, and the quad shakes and wiggles by itself tuning itself until it likes the result. It did work, but produced a pretty terrible tune. It has been several years now so I expect people will try again. So yeah, auto tune exists. I just meant it wasn't user friendly and common yet.

[@dlsspy]

There are very bad naïve implementations of autotune where people with no control theory background have made guesses as to what would work and then there are really well thought out models with control systems experts who've built good, reliable, repeatable systems. You've probably observed the former. This is more of the latter. It's only not common because people rarely choose the best things available to them. A newbie tends to go to an event and get introduced to something by someone who's only used the one thing and will tell the noob how it's the best possible thing and the noob continues the lineage. The weird thing is how these things that are worse in many regards are brute-forced up to being fairly decent. You get a enough people who don't know much about control theory or embedded systems building an embedded control system and somehow something relatively OK emerges.

[@michaelwilkes0]

yeah, agreed.

[@moendopi5430]

For anyone looking at this in 2020, it turned out that on the newest update of Windows, the blackbox-decoder won't open. Or Windows won't let it. However, I just gave Plasmatree a raw log file and it did the conversion itself and it made the graphs anyway. Additionally, I found another way to look at your logs: amir.ch/blackbox/. The main page has links to Joshua's PID tuning guide on RCGroups as well as other useful resources.

[@Nico-dj3gu]

Its hard to compare, because the quad flew super good with stock pids.

[@wudz_fpv]

The difference between the two videos (most obviously propwash) is due to the added overshoot in the second flight. The default PID's probably only needed about 1 value less on Pitch P, and about 3-5 less I on Roll. Yaw is relatively negligible and can be tuned mostly by feel. I'd be interested to see how my suggestions affect your step response.

[@JoshuaBardwell]

You're saying it's better to tune no overshoot and slower response?

[@wudz_fpv]

Yes, you don't want overshoot but you do still want a quick response at the same time. I would Imagine a general rule of thumb would be to find a balance between the two. I find that once Iterm is set, the only values you really need to play with are P and D. Iterm more or less smooth/roughens the step response. By balancing P and D you can get a sharp jump to 1.0, and a virtually flat plane afterwards. Flo suggested on RCGroups that once you get the jump to 1.0 and then a flat plane, you can decrease the response time by bumping up P and D simultaneously and at the same ratio. I haven't tested this quickening of the response yet though so I myself am unsure of the results. All I know is I am in love with the way my quad flys now lol.

[@MrSabram07]

Wow

[@ChriFux]

i'd like to see the graphs of a quad you tuned yourself and consider very well tuned as a comparison please

[@JoshuaBardwell]

Well to be honest I thought the BF defaults were pretty damn good.

[@ChriFux]

Joshua Bardwell okay, thanks

[@fsarfino]

I run full flights and this method IMO is not needed. I'll link video to stock helio pids on a 3m old build running odin with elastic pids. The results on stock for this analyzer are just nuts

[@JoshuaBardwell]

It's hardly surprising that Odin elastic PIDs confuse this tool.

[@fsarfino]

kzbin.info/www/bejne/q3XOpmaKo7OXsJY

[@fsarfino]

Don't get me wrong there are some extremely minor issues with the tune on my build but for the vast majority of pilots out there Helio is really flash and fly with stock its amazing what these firmware's can do now compared to a year ago

[@Fleche_FPV]

Hi, with BF 3.5 what is your point of view for the "Feedforward" settings in the PID board to use plasmatree ?

[@JoshuaBardwell]

GOOOD Question. I don't know the answer yet.

[@divingfalconfpv4602]

Anyone still use plasma tree or is there something new.

[@GhostRyderFPV]

Mmmmm, donuts...

[@shri7582]

is this similar to g tune of cleanflights.

[@JoshuaBardwell]

It's similar in that it is measuring the step response. G-Tune tried to automatically determine the best P gain, but it wasn't very good at it. G-Tune also did not tune I or D at all.

[@shri7582]

PID tuning is an art, it is not compulsory that pi=3.14, it might be just 3 for me or some other value for others.BTW no AI or dumb software can help me paint the best canvas which I like.

[@Ammothief41]

Thank god for playback speed 1.5x!

[@beanMosheen]

Go fly 3.4!

[@propnut1361]

Funny...I used to build and fly (and not do any tuning) and it seemed the quads were tuned OK. Not much prop wash if any, I began with Librepilot, then Cleanflight, and finally to Betaflight. The thing I find odd is that it seems that all of my recent builds (within about the last 9 months to a year) all seem to suffer from propwash. from subtle, to obvious. I sometimes think I can tell the difference now better than I used to, but the truth is I have always have been able to identify propwash as I found it happening more to "deadcat" style layouts. But now it seems more pronounced on every build. Is it just me?? :) Thanks for the content!!

[@JoshuaBardwell]

Old quads had terrible handling, we were just so used to it we didn't notice. We were not able to fly nearly as aggressively or quickly, so we didn't bring out bad propwash like we do now. If I go back and watch my old tuning videos, the propwash is so huge and low-frequency we would hardly call it that today.

[@1337flite]

#1

[@pkvillager]

Life is too short to PID tune these days unless you have a major issue.