another super interesting video the part about TPA is awesome, thank you for that! using your tool now on all my quads and when I think how hard and time-consuming I did tune some of them it is so much more fun and easier doing it this way, just one recommendation would love to see a video starting with basement tuning - PD Gain and BD balance maybe with a quad where Pitch does need different values (deadcat designs 7" and bigger could give tons of logs about that topic :) - going over filters and to the Final touch I and other small tweaks :) THANK YOU
thanks. I'll be doing another vid tuning the Chris Rosser AOS 5", and probably using BF4.3 dev build where we have pitch to roll ratio slider, so I'll be looking into how to get that balance just right.
FYI: there are many ways you could effectively get a “finger print” of your copter’s step response profile and make a decision about what the optimal PIDs should be. We opted for PD gain steps because from practice it seemed the quickest route to finding the optimal, in part because PD balance in BF4.2 is by default pretty good for most builds, and there's a wide tuning window for D. When we see this slow overshoot continue to decrease as PD gain is increased, it is an indicator that the gain is closing the error gap, and is therefore not related to low D gain, but instead is acting to stop I term from winding up, which is the basis for the slow overshoot/oscillation in this case (i.e., it optimizes the 'PI' portion of the PIDs). The point at which PD gain no longer reduces the I-related overshoot we stop. Any remaining overshoot is therefore D-related, and some fine tuning of PD balance then may be necessary. Note: Recommendations may change as BF and BF sliders develop.
I wanna add to this, after a series of tests I now know this higher freq oscillation (75Hz) was mainly D related, as TPA D took care of 95% of it. Turns out P related oscillation tends to be lower freq, closer to 40Hz. Made an interesting post on BB log review site.
I’d just add it back based on feel. You can certainly look at step plot but it will undoubtedly show more overshoot but it’s not to be concerned about because FF doesn’t produce oscillation and that overshoot is not generally visible unless it’s some extreme value. Also sometimes FF causes an offset issue in the step plots because it messes with the shape of the the input
Thanks to all of you.
another super interesting video the part about TPA is awesome, thank you for that! using your tool now on all my quads and when I think how hard and time-consuming I did tune some of them it is so much more fun and easier doing it this way, just one recommendation would love to see a video starting with basement tuning - PD Gain and BD balance maybe with a quad where Pitch does need different values (deadcat designs 7" and bigger could give tons of logs about that topic :) - going over filters and to the Final touch I and other small tweaks :) THANK YOU
thanks. I'll be doing another vid tuning the Chris Rosser AOS 5", and probably using BF4.3 dev build where we have pitch to roll ratio slider, so I'll be looking into how to get that balance just right.
FYI: there are many ways you could effectively get a “finger print” of your copter’s step response profile and make a decision about what the optimal PIDs should be. We opted for PD gain steps because from practice it seemed the quickest route to finding the optimal, in part because PD balance in BF4.2 is by default pretty good for most builds, and there's a wide tuning window for D. When we see this slow overshoot continue to decrease as PD gain is increased, it is an indicator that the gain is closing the error gap, and is therefore not related to low D gain, but instead is acting to stop I term from winding up, which is the basis for the slow overshoot/oscillation in this case (i.e., it optimizes the 'PI' portion of the PIDs). The point at which PD gain no longer reduces the I-related overshoot we stop. Any remaining overshoot is therefore D-related, and some fine tuning of PD balance then may be necessary. Note: Recommendations may change as BF and BF sliders develop.
Lots of valuable info here, subbed!
Subbed! PID Toolbox is the Swiss army knife in your tuning box. 💜
Awesome!! More nuggets of info. Thank you!
this is awesome stuff! I managed to tune my quad in no time.
We definetely need linux version of 0.42 )
Impressive tuning guide! How about tuning with angle mode so it will be easier to control the drone inside ??
Did you lower your rates to make it more controllable in the confined space?
my rates are already pretty low: actual 100/700/0.80
Great video. Really good explanation of how to adjust TPA. How do you decide whether to use TPA on D-Term only (bf default) or on P and D?
if after the tune you see a clear 30-70hz oscillation that kicks in >30-40% throttle, TPA 'PD' will work.
I wanna add to this, after a series of tests I now know this higher freq oscillation (75Hz) was mainly D related, as TPA D took care of 95% of it. Turns out P related oscillation tends to be lower freq, closer to 40Hz. Made an interesting post on BB log review site.
@@PIDtoolbox is there a way to view this post without a Facebook account?
@@zuldanfpv4634 unfortunately not. Im an admin there so just join and I'll add you. facebook.com/groups/291745494678694/
It’s asking me to create a Facebook account. I quit Facebook 11 years ago lol. Thank you for offering.
15:13, you said magnitude, did you put the psd result in sqare root form?
I was referring to amp spec, then it switched to psd, which is essentially 20*log10(amp_spec^2).
After tuning do you just add back FF by feel or do you log a FF sweep the same way as the PIDs?
I’d just add it back based on feel. You can certainly look at step plot but it will undoubtedly show more overshoot but it’s not to be concerned about because FF doesn’t produce oscillation and that overshoot is not generally visible unless it’s some extreme value. Also sometimes FF causes an offset issue in the step plots because it messes with the shape of the the input
@@PIDtoolbox thanks!
@@MauroPeneda you could tune it via betaflight blackbox viewer just by measuring delay between setpoint and gyro