Thanks Mark, love the new terminology. Definitely makes the understanding of what they do and how they relate to each other that much more clear and in layman’s term.
I don't think that is what was said or the case. What he said is that the reduced delay allows for higher PID parameters, and those increase the resonance freq ( 5:50 ).
@@someonespotatohmm9513 That's what I said. Except that PIDs don't increase the "frequency" - it's the less delay allows for a "tighter" interplay in the system thus "raising" the frequency that they're ideal at.
@@mlentsch I admit it is kinda pedantic, but rewatch the bit after my timestamp. The control frequency goes up from the increased PID values, not from decreased delay. The nosie rejection (so prop wash) should go up from both decreasing the delay and PID values (not 100% sure on the math about the delay). The difference is importand because you leave performance on the table if you just decrease the filtering.
Doing tuning right now on aura 10. Started at presets that were extremely too low. So bumped up later instantly alot. And bumped D term to get PD ratio. Then had to take yaw off slider double that. Haven't gotten to go tune again. But i did get a cli pid tube from rcshim he said to try his. He still has wobbles but i could start from there.
Once again an excellent summery of the concept! Only thing I will nag about is that when you raise D the step response curve should get flatter (slower).
Thank you for that, it’s stinking awesome to have someone put it into English.. joined your Patreon👍 I planned on watching enough of your content last winter on my time off to learn how to tune my quads, I have quite a fleet and I build a lot of them, big birds. Hopefully this winter will be the winter that I get that done
Cause that is when the great debate occurred for the slider. Why would we/I rerun for every release?. The controller is, has, and always will be a Proportional, Integral, and Derivative controller. Same control logic. Same physics on the quad copter.
@@uavtech can I highjack the comment and ask you how much has the sliders changed from 4.3 to 4.4? Every drone I try to pid in 4.4 fly worse then in 4.3 and I can't understand why. Thank you for your response and for your time!
Thank you for another excellent explanation. One question if I may. I note that you said that in Inav for the Z axis there is only the P term that is adjustable. On my 10" quad on RTH the quad oscillates above and below the required altitude, the resulting current draw plays havoc with the Li-ion battery. Will adjustment of only the P term be able to eliminate this, or do I have to just accept it as an unfortunate characteristic. Thank you.
The Integral (Ki) is not the most important setting. You must use Proportional (Kp) in all controllers. The Derivative (Kd) dampens the Kp response and allows you to use a higher Kp value than you could without it. With only Kp and Kd, a controller won't fully reach the setpoint because the closer the system gets to the setpoint, the lower the error signal gets, and it can no longer drive the system to the setpoint. The Kd only does dampening, so it can't drive the system to the setpoint. The Ki circuit samples the error over a period of time and sums the errors it sees until the sum is large enough to drive the system the rest of the way to the setpoint. So, you increase Kp until you see some oscillation, add Kd to dampen the oscillation, and then add Ki to reduce the long-term error between the setpoint and the system output.
Who said Ki is most important? However you get the balance, once you have it and are at a Critically Damped response, moving the gains up to just below the system resonance is the key for peak performance of whatever you are tuning if you want a rapid response and the most stability to following and keeping on the Setpoint. That's the part the classic tuning method you described above leaves out.
Just putting the terms in order of how they relate. i-Term oscillation is dampened by increasing P (shown here ua-cam.com/video/Sq_DFjmvVDE/v-deo.htmlsi=Tm5FpvpLbmIQps4t) P-Term is dampened by D.
If fast oscillations (VERY FAST), yes. Reduce P or increase D. For BF, the P and D gains should be around the same numbers. HOWEVER, it is more likely the I-gains which can result from them being too high in relationship to P gains, or the overall gains just being too low. To confirm which it is, set P and D to around 40 and I-gains to around 20 or 30 for an 8". Those are pretty safe values. Once the oscillation is gone, you can then work to increase the gains slowly to increase flight performance (stick tracking and prop wash response). Check out my presets for an 8", but be cautious as those assume you don't have mechanical vibration issues.
Thanks for the reply i forgot to mention i run inav also the i can see the ocilations while hovering i am not sure if that is considered as very fast ocilations
Have I understood that correctly that with lower filtering resonance frequency can go up ? If I run a resonance first flight with stock pids on a quad , do this resonance map I make be modified later changing the filters?
@@uavtech a profile of the resonance I make with a first resonance flight, pushing only throttle on all range gradually . Could be a pod toolbox freq/time graph or blackbox graph
Mr Mark bringing up a video editing game, love it! :) Nice explanation
Now we just need KISS to also have sliders :)
I need a slider to make me not a shit pilot! 🤣🤣🤣
ya like that! Point Tracking FTW!!
only a content creator can appreciate the time it takes for that stuff.
Again, Crystal Clear!!
I need to learn more about this. Thanks! It still makes my head hurt though. 🙂
excellent overview of what it all means. thank you sir!
This is marvelous, I'm getting more courage now ! Thanks !
Thanks Mark, love the new terminology. Definitely makes the understanding of what they do and how they relate to each other that much more clear and in layman’s term.
Good one - I'm understanding better, now - the filtering delay affecting the resonant "mode" - awesome! Thanks, dude.
I don't think that is what was said or the case. What he said is that the reduced delay allows for higher PID parameters, and those increase the resonance freq ( 5:50 ).
@@someonespotatohmm9513 That's what I said. Except that PIDs don't increase the "frequency" - it's the less delay allows for a "tighter" interplay in the system thus "raising" the frequency that they're ideal at.
@@mlentsch I admit it is kinda pedantic, but rewatch the bit after my timestamp. The control frequency goes up from the increased PID values, not from decreased delay. The nosie rejection (so prop wash) should go up from both decreasing the delay and PID values (not 100% sure on the math about the delay). The difference is importand because you leave performance on the table if you just decrease the filtering.
it finally makes sense... thank you VERY much~
Your knowledge is very useful for we beginners, keep sharing.❤
Thanks you so much!
Doing tuning right now on aura 10. Started at presets that were extremely too low. So bumped up later instantly alot. And bumped D term to get PD ratio. Then had to take yaw off slider double that. Haven't gotten to go tune again. But i did get a cli pid tube from rcshim he said to try his. He still has wobbles but i could start from there.
Once again an excellent summery of the concept!
Only thing I will nag about is that when you raise D the step response curve should get flatter (slower).
Thank you so much
I = Imbalance is perfect!
You need 'I' to balance / center the oscillations around the set point.
This is the easiest way I have heard it explained.
And you DO NOT want your Imbalance term doing the job of the P-term due to lack of P-gain.
It will cause wobble.
Fantastic info, Mark! Thanks a bunch! 😃
Stay safe there with your family! 🖖😊
And happy holidays!
Thank you for that, it’s stinking awesome to have someone put it into English.. joined your Patreon👍 I planned on watching enough of your content last winter on my time off to learn how to tune my quads, I have quite a fleet and I build a lot of them, big birds. Hopefully this winter will be the winter that I get that done
A good inav motor output tab idle setting and the receiver tab in DEPTH would be nice
Great job in making things clearer for me! But why every test I'm seeing is still on BF 4.3?
Cause that is when the great debate occurred for the slider. Why would we/I rerun for every release?. The controller is, has, and always will be a Proportional, Integral, and Derivative controller. Same control logic. Same physics on the quad copter.
@@uavtech can I highjack the comment and ask you how much has the sliders changed from 4.3 to 4.4? Every drone I try to pid in 4.4 fly worse then in 4.3 and I can't understand why.
Thank you for your response and for your time!
They did not change at all from 4.3 to 4.4.
@@uavtech awesome, so I can copy thr pids as they are. Probably I've messed up something else then.
Thank you so much!
Shouldn't the master move the sliders proportionally rather than the delta like what you seem to hint around 4:00?
The proportional differences between the gains remain consistent when the Master Multiplier moves up / down.
There is some rounding error, but ...
Thank you for another excellent explanation.
One question if I may. I note that you said that in Inav for the Z axis there is only the P term that is adjustable. On my 10" quad on RTH the quad oscillates above and below the required altitude, the resulting current draw plays havoc with the Li-ion battery. Will adjustment of only the P term be able to eliminate this, or do I have to just accept it as an unfortunate characteristic. Thank you.
P only to adjust the Z Velocity target. Yhe Z velocity PID loop has P
I. and D.
I would log the issue to see wsup.
The Integral (Ki) is not the most important setting. You must use Proportional (Kp) in all controllers. The Derivative (Kd) dampens the Kp response and allows you to use a higher Kp value than you could without it. With only Kp and Kd, a controller won't fully reach the setpoint because the closer the system gets to the setpoint, the lower the error signal gets, and it can no longer drive the system to the setpoint. The Kd only does dampening, so it can't drive the system to the setpoint. The Ki circuit samples the error over a period of time and sums the errors it sees until the sum is large enough to drive the system the rest of the way to the setpoint. So, you increase Kp until you see some oscillation, add Kd to dampen the oscillation, and then add Ki to reduce the long-term error between the setpoint and the system output.
Who said Ki is most important?
However you get the balance, once you have it and are at a Critically Damped response, moving the gains up to just below the system resonance is the key for peak performance of whatever you are tuning if you want a rapid response and the most stability to following and keeping on the Setpoint. That's the part the classic tuning method you described above leaves out.
@@uavtechcalling ipd instead of pid, it gives that vibe.
Just putting the terms in order of how they relate.
i-Term oscillation is dampened by increasing P (shown here ua-cam.com/video/Sq_DFjmvVDE/v-deo.htmlsi=Tm5FpvpLbmIQps4t)
P-Term is dampened by D.
@@uavtechYou did when you rearranged the terms into IPD at 1:32.. 😎
Never said. You inferred the wrong thing. 😜
Is it time to update the basement tune method? I think I remember it not using the tracking slider.
It doesn't matter how you get the balances, as long as you do.
So if i have fast pich and roll ocilations while hovering on a 8inch i should increase the d gains?
If fast oscillations (VERY FAST), yes. Reduce P or increase D. For BF, the P and D gains should be around the same numbers.
HOWEVER, it is more likely the I-gains which can result from them being too high in relationship to P gains, or the overall gains just being too low.
To confirm which it is, set P and D to around 40 and I-gains to around 20 or 30 for an 8". Those are pretty safe values. Once the oscillation is gone, you can then work to increase the gains slowly to increase flight performance (stick tracking and prop wash response).
Check out my presets for an 8", but be cautious as those assume you don't have mechanical vibration issues.
Thanks for the reply i forgot to mention i run inav also the i can see the ocilations while hovering i am not sure if that is considered as very fast ocilations
same advice, but reduce the pids based on my pid calc tool: theuavtech.com/pidcalc
@@uavtech thanks
Have I understood that correctly that with lower filtering resonance frequency can go up ? If I run a resonance first flight with stock pids on a quad , do this resonance map I make be modified later changing the filters?
Not understanding "resonance map". What do you mean by that?
@@uavtech a profile of the resonance I make with a first resonance flight, pushing only throttle on all range gradually . Could be a pod toolbox freq/time graph or blackbox graph
Inav, flying in navigation mode can cause fly away this is deep in notes just a heads up
Yikes!!
Ah, I see me :) Am I famous now? lol
Yeah, POS Hold is all you baby!. :-)