PID Tuning Principles: Step #5 - Yaw Axis Tuning
Vložit
- čas přidán 28. 08. 2024
- A review of how to approach Yaw tuning. We review how Yaw axis is different from Roll and Pitch (no thrust) and why you need really high i-Term and P/FF terms to track Setpoint. We also talk about and show why D-term is not needed on Yaw.
Tuning guide at: tiny.cc/uavtech
Playlist of the series: • PID Tuning Principles
------------------------------
Want to support me? / uavtech
or buy a beer: paypal.me/pool...
NEED HELP?!?
UAV Tech Discord: discordapp.com...
Betaflight BlackBox Log Review Group: / 291745494678694
Facebook: UAVTech1
Seriously, this series is the first of innumerable PID tuning videos to make sense and allowed me to make headway. Thanks for bringing this together!
Awesome!
how can you dislike a video of someone clearly and correctly taking time to teach people important aspects of a hobby. i mean its only one dislike but still blows my mind. Thanks for these videos
Know it alls get offended. 😁
You very much filled the space that JB left. (I know he's still got it but he certainly focuses more on entertaining now than technical details I'd say... which is great but BF 4.X still remains a mystery to many). Thank you for the work!
I tuned my new Freestyle Quad step by step to this Video series. Im soo happy now my quad feels great . Thanks a lot
Amazing Job Bro
So the other day I decided to become one of your patrons. I tried to go custom but it didn’t work! so I went with the highest one, which you well deserve!!!! you rock Mark!😎
Thanks for the support!!! Much appreciated!!!!!!!!!!
Thanks for taking my fear away from Blackbox and helping me understand the tuning process. It makes complete sense to move a dial when you actually understand its effect on the quad behavior. cheers!
thank you for this very very good explainations! best pid tuning guide so far! 👍
Thanks a lot Mark, very well and clearly explained chapter! :)
man this is great stuff thx !!!!! love you dude even if my wife hates me after I watched all the PIDs tuning videos in a row .....
That makes two of us!
excellent!
thank you this playlist was very informative, and thank you for sharing your drive folder with all those years of research so well condensed : )
Thanks so much for sharing! So far this has been such a detailed and comprehensive PID tuning guide, awesome work explaining everything! PIDs and Filters have been demystified for me. - subscribed to your channel and tuned for future videos:) cheers
Legend
Your videos have been super helpful.....thanks for your contributions......👍
Great vid my guy. As always.
Great video as usual
Thanks again for theses great video.
It is also a comment for CZcams algo
Appreciated!!
Thanks again Mark for another awesome informative video. Quad never flew better! Is the new FF already in the nightlies?
Not yet, but it is in Jflight: github.com/jflight-public/betaflight/releases
Allriiight
Thanks.
Yeeeh yaw!
Just finished tuning one of my quads following your five steps, betaflight 4.0.5. This has to be the best I've ever had any of my quads flying. I can't get the gyro to completely follow set point but I would say my quad is now 99% there. Thanks for the video series. I assume for betaflight 4.1 since everybody is recommending to keep all the filters on the only thing I would need to do then is to in the pits and not mess with the filters?
You want to move the filter slides up. See my RPM video, also the filters details video i released today.
Can you help me trouble shoot please? Frame is a six inch Kabab Flowride. Motors Emax 2306 RS. I refitted it with F7 CL racing stack and motors 2 and four will loose power at about third throttle and then continue but at a much lower current making any kind of Acro maneuver causing you to drop out of the sky. If a motor is bad could that cause this? The ESC and FC are brand new and the build was clean. Could it be defective ESC? I’ve never experienced this on any quad ever. I can send a video is that helps. Thank you for any ideas 💡 🙏!
I need someone help bad my yaw is turned up on my remote I finally got my quad to arm outside of beta flight but when I give to much throttle it just takes off and I can not turn the motors back down they just go like full throttle and won't turn down
My quad (flywoo lr 4 inch with 450mah 4s battery) jumps like a meter or so when i do fast Yaw movements like in this video at arround 00:54 mark. It is ok only if i move the Yaw stick slowly. Any idea how to solve that problem.
Thank you sharing thesee valuable information. :D
Hey Mark, If I decrease filtering of D term through sliders, I get roll oscillations when performing yaw moves, are there any suggestions what to do?
Don't decrease filtering?
how do you deal with altitude changes with yaw inputs? I understand that yaw'ing spins up and/or spins down two motors so I just kinda learned to live with the altitude changes but looking at your vid yawing left and right, your altitude seemed fairly constant. i've been having to actively manage my throttle and it makes for inconsistent flying.
Correct. Drop throttle.
Sounds like FF boost is needed 😏
Is that what we are calling Chris's new code?
why all my quads are getting throttle jumps if I gave full yaw stick, also when I center the yaw stick. It happens only on very fast throws. it jump up 3 to 4 meters.
Physical constraint. Motors spin up faster than they slow down.
In tuning yaw, do keep in mine rarely do you just give it full raw rapidly without another movement mixed in.
Are you giving full stick deflection when yawing the first minute of the video? I'm on Betaflight 3.5.7 and I'm staying on 3.5.7 lol but I'm in the process of tuning a 6inch and I've got a bad yaw jump just curious if you have any ideas of what I should do.
Yaw noise.
Hi! Thanks you for this work. What is constant up and down on gyro trace mean? While i fly like on 30-40% of throttle its always there, frame resonanse or maybe bad gyro itself?
Look look like this? czcams.com/video/GrUvTi2b7aA/video.html
Do I need to turn off anti gravity, so I could isolate only pids?
Yes and No. I do not. AG only activates on fast throttle moves.
Props in or props out have any noticeable effect on this?
No
Hello as usual, thank you for the advice you give us. I tried the setting and I have to say I like it, I only have one question: with pids so high, does battery consumption increase? sorry for the grammatical form but I use the translator
In theory, but consumption has so much more to do with how hard you are flying; or not.
@@uavtech Hello. I think you're right I tried it and it's just like that. It's that the setting is so good that it leads me to push harder.
Hello Mark with the Eachine TS 215 V3 and Betaflight 4.0.5 all hve all corecly done and flashed him , have maked all from your Vidios and when i starded him he was going directly high about 50 m and comes back with a noise of a bee that two times so i dont have control with t Transmitter so when he comes back in 2 m about the groud i cut him and he falls down . The Omnibus F4 SD with this version the FC is the whole bottom plate . The motors have had 105 degres i was rely shoked , and i have 9 Drones more and the all are going perfectly . End of the song i flashed him back to 3.5.7 and he goes like a devil i write you that i have had everytime the problem with this Drone . I dont understand until today ???????
Sounds lik you have it sorted. In the future, you don't need to flash back. You can just apply 3.5.7 defaults to 4.0.x and get the same results as 3.5.7.
Your issues is low end noise that 3.5.7 defaults has more conservative filtering to cover up. github.com/betaflight/betaflight/wiki/Community-Presets#revert-to-bf-35x
I know we have talked about this mulitple times now. Other comments, FB, Discord. All over this Eachine TS 215 V3. Throw it away. LOL.
I have very strong bounceback after fast rotation on YAW axis and sharp stop. Its only in BF 4.1 . There is NO strong bounceback in BF 3.5.7 .
Its only on YAW axis. Pitch & roll is good !
What can i do ? What settings to change to eliminate strong bounceback on YAW ?
lower P or FF, that's about it. Not a lot of authority on Yaw to control such issues. I only get them if i do a full rate 360 on yaw, which i never do in flight so not an issue.
Any chance you can attach a cli dump. I watched the whole series but I’m not so techie. Also would all the same info apply to 4.1?
yes. applies to any firmware and PID loop.
hey uav tech i just got the iflight succex mini twing f7...im confused about the product page description at the very bottom theres a notice""NOTICE:
For betaflight 4.0 and up you need to flash the base firmware stm32f411,and then copy the CLI for individual fc board.
CLI code here for SucceX F7 TwinG and Mini F7 TwinG
im a noob so someone please explain.....thanks in adavance
you can flash the EXF722DUAL (see www.screencast.com/t/sPoHVcyaq). They are just jumping ahead to the Unified Targets that are coming with 4.1.0. The unified target for the TwinG is teh STM32F411 and then you post the Config File and then setup from there. For now, just flash the EXF722DUAL. BF devs are making the Unified Targets much easier to config up through the Configurator with the 4.1.0 release coming in October 2019.
Ok because when I tried the stm32f411 I couldnt get back in the fc..so I used stm32f7x2 and it worked plus my rssi is working right now...had a hard day yesterday trying to get rid of a real bad yaw twitch ..come to find out gyro 1 is bad ..yep DoD ..are these gyros bad or am I just unlucky
@@byrden1985 , mine have been fine.
Wow... Glad I've gone to falcox
LOL. Falco X still has a PID controller. Smartass. :-)
Plus kind of flying blind for making filter changes (no logging). I kind of look at it as if the presets fly good for your, sweet, but if not it is a lot of guessing and less tools to address specifics issues (frame resonance, etc...). This said, the BlackBird preset flies sweet on my Nova frame.
@@uavtech lol yeah I know, its just that it seam every revision of beta its like this and that have changed so forget everything I told you before... And yes I know all the other are like that also, but let me live in my fantasy world being new to FalcoX for a little longer...😳
So I have a question about tuning yaw. I get crazy oscillation when I yaw spin, to the point the quad either rockets into the ground or to the sky if inverted or not. Is that just too much i term? All defaults 4.0.4.
No, you probably have a lot of Yaw noise. Check your yaw traces and don't forget about the yaw filter if you can't fix it mechanically.
@@uavtech the traces look clean even on raw gyro, but the motors are going to 100%, hence the rocketing up. Also it only happens on zero throttle.
@@mlfpv7025 , you have a log? You have motor reverse set right? I-term is high in 4.0.x on yaw, not noise has NO impact on Yaw. P-term yaw is lower than ever before on yaw.
Check motor reversed toggle and also kwad orientation in Configurator.
@@uavtech Hi, got the same behavior since 4.0.x on several quads, I'm running motor reverse too, I dont realy understand why this happens on quick yaw stops. Gyro is well filtered on all axis. Also It's way more prononced on my low kv 6" than on my high kv 5".
@@uavtech Yes I have a log. You want to take a look?
I'm confused. You say that feedforward is faster than p-term because p-term needs a pid error and feedforward reacts to the sticks. I understand that, but once you move the sticks, don't you instantaneously get the pid error because the setpoint moved? Or is there something (maybe stick interpolation) slowing down the setpoint reacting to the sticks? And if the difference is caused by stick interpolation, does that mean that feedforward is reacting to raw stick values instead of interpolated stick values? Or is there something else going on? what is preventing the setpoint from instantly reacting to the sticks and why is feedforward not also hindered by whatever that is?
FF is measured of the speed of you moving the sticks. So the faster you move, the more push it drives. P-term does not work that way. It only drives based on PID Error (the bigger PID Error the bigger the P push). The main problem with P-term only is you can't have it high enough to provide enough push to track Setpoint super well. If you would make the P-gain high enough to push hard to track Sepoint you will get extended oscillations in prop wash. There needs to be another term that is seeing the change occur and beating P-term to the punch; to the push. In classic PID control, this is called FF and is based off measuring differentials BEFORE they cause PID Error. That is what FF does in BF. See czcams.com/video/FW_ay7K4jPE/video.html for details.
@@uavtech
Ok, I watched it. I think I know why I was confused. I thought that somehow feedforward was calculating the setpoint error faster than the normal controller was calculating setpoint error, which is silly because its just a few add and multiply commands. But that is not what is happening. Instead, the feedforward is predicting where the setpoint is going to be in the future based on (as you say) how fast the stick is moving. The stick isnt actually there yet, and the setpoint isnt there yet, and in fact its possible that you could instantly stop moving your thumb and this guess would be wrong. So the feedforward is basically telling the system to react as though the setpoint has moved a lot, even though it has only moved a little. In the rare cases where the setpoint does not move a lot, and really only did move a little, well oops. We will just have to let the pid controller undo our bad guess. Does this sound about right?
I fly a lot of line of sight, and this new 5 inch I just built does some weird stuff... When I do a constant yaw, right at the beginning it pumps up and shoots up about a foot..... then when I stop the yaw it pumps up even more and shoots up like 2 feet....
I guess I never noticed this happening on my 3 inch because the motors were smaller and low RPM lift wasn't as drastic so I didn't get any of that shooting up. I think the fact that theyse 2306 motors have a lot of low end torque that it makes this more drastic..... But MAN is there a way to at least reduce this? Can I get rid of it altogether? Even fenagling the stick in a way to cut the throttle at the end of a continuous yaw spin KINDA works but it kinda goes nuts........
Edit: but no I just remembere I DID have this issue one time with my 3 inch quad damn... I remember specifically I was like WHAT THE HELL.... and I honestly don't remember what I did to change it.... I THINK it might've been simply all about centering the gravity better... kinda makes sense because I do have a huge bulky antenna sticking out the back of the quad -.-
You can turn off Airmode to mitigate, but it is a mechanical limitation i don't know how you can currently get around. Yaw is motor prop and bell inertia driven - two spin up as two spin down - and motors don't spin down as fast as they spin up, which causes the jump.
Let me talk with Joe Lucid (RPM filter inventor and a big LOS pilot) about an idea I have to resolve through Firmware.
@@uavtech Cool thanks... It wasn't happening to you nearly as drastically as it happens to me. I TRIED RPM filtering and it just wasn't working for me, I started getting these extremely strange artifacts and flight characteristics. Almost like the RPM detection lost it's focus every once in a while and the quad didn't know what to do. It's hard to explain. It just wasn't working for me... I set it up all correctly etc but W.E I'm fine without it. ;>
Maybe I'll give it another shot in the future but I saw the same type of thing in one of JB's videos. It's the one where he puts RPM filtering on one of his "problem quad". And claims it got better... but you can see in the video it's not better.... But more importantly you can see those weird artifacts I was talking about experiencing after using it. The artifacts are even audible. One clear example is while flying LOS I was doing back flips and most of the time it would work fine, but every once in a while I got this HUGE... like... ENORMOUS bounce back that you can SEE even flying LOS which is like.... something you don't see very often flying LOS so I was like... Nope I'm going back to 8k 8k dshot 600 ;D
BTW I use idle up on my quads and never use airmode... I'm glad of it now lol... The thing is if I don't do a QUICK stick movement to yaw it doesn't happen as bad... it's only when I make quick stick movements......
You can kinda hear what I'm experiencing in your video here, at the end of a full yaw spin you hear the motors kind of... upping and downing (for lack of a better term) when the yaw stops to fix itself... but mine is so dramatic it almost gets me stuck in a tree it's so bad.
@@uavtech lemme elaborate further about this as well. When I do this yaw spin ALL of the axis gyro and set point lose each other terribly.... the setpoint stays flat with pitch and roll btut he gyro flows up and down over it like a wave of water.
Is this normal? This is a new quad new motors etc.
@@upplsuckimcool16 bounceback is NOT because of RPM. It is just not possible. Log it both ways and you can see for yourself. Regardless, there is nothing wrong with sticking to no RPM too.
I would need to see some DVR and logs to see what is going on with your tune.
@@uavtech It's not constant boucne back... it's an artifact. It's so dramatic I would even hesitate to call it bounce back... it's just a straight glitch in what is supposed to be happening. It's more like an enormous twitch like a bad motor.
integrated yaw next?
Thinking how to run RPM filter on BLHeli_s ESCs.
Any advice on how to deal with Yaw iTerm windup? I've a copter with winglets that transitions into forward flight. However when turning, I'm seeing the Roll bounce back. (Yaw controls Roll once nosed over)
More info on the Quad: rotorbuilds.com/build/27673
Thank you in advance!
More P-term on yaw, or lower i-term windup in the CLI to 50%
@@uavtech Thank you so much for the response / advice!
As you've recommended, I've added P via "gain" sliders and set iterm_windup = 50. Along with that I've mounted the motors with 5% dihedral to allow for more Yaw authority in winged flight. (Parrot-Swing Drone does something similar but it's more like 10% dihedral) The iTerm relax is set at RPY, Setpoint and "15" which should be close to the defaults. Once the kids are in school (Monday) I'll run a few test flights and post a video with a link. Unfortunately, the JHEMCU Play F4 Whoop doesn't come equipped with a logger. Though I'm considering installing an Openlagger depending on the outcome of the next few flights. (Assuming I have a port free)
Edit: Subscribed!
Update: Tried to fly this morning but BetaFPV M02 VTX is dead. In this case I "believe" the antenna came lose somehow and burned the transmitter. Hate BetaFPV products as they always fail or are of poor quality.
will likely swap out the FC and VTX while ditching the 5V regulator that's no longer needed. (SucceX F4 V1.1 1S 5A AIO Whoop Board with VTX) Should save 3-4 grams as well as simplify the wiring quite a bit.
I would set RP for i-Term Relax. Yaw depends on I-Term quit a bit for authority. But try it both ways.
School already!!
@@uavtech I reside in Germany (retired military) so the school schedule is a bit different from the States. Last week was their last school day and they're now on holiday for 6 weeks!
Anyhow the increasing / decreasing Yaw Gains didn't help no matter what values I tried. Will next try out integrated_yaw to see if ditching Yaw iTerm avoids this issue entirely. That's assuming Yaw will retain some form of stability control. Which would be really helpful for maintaining Roll stability in winged flight.
Latest flight: czcams.com/video/Ji4h01kVouo/video.html&ab_channel=4DCResinSmoker
Side Note: Its surprising just how little information is available online for integrated yaw. Next to nothing on CZcams either.