Yep, more data would be great … as I kind of cheat to get a kind of realistic curve 
1 Like
Yeh, the challenge is that it changes a bit depending on the prop, motor and ESC. I suspect that things are getting much more consistent now though.
1 Like
@khancyr has done some updates and added some graphs to the wiki.
Thanks!!!
1 Like
Like the graphs makes the wiki a lot clearer
1 Like
People are flying much more with props with very high static thrust on these smaller copters. I have recently switched from 2 blade 4045BN to 3 blade 4052. Will this make a significant difference to the curve? Also if it’s PWM -> thrust that’s motor dependent as well?
Yeh, the curve is basically based on the efficiency of the system. That is why it is so hard to give a single value for a given prop size. In general, smaller diameter props have lower efficiency so they have lower expo values. I suspect that modern props and motors have better efficiency than the older versions but we are also pushing them much harder. Finally many of these props will stall out at their top end in static thrust but not so much when they unwind. We can’t even measure that unless we do our thrust test in a wind tunnel!
So I tend to work with the static thrust values and remove thrust roll over in the upper end if it sounds like propeller stall but in flight it sounds good. Other times it is just the rpm of the prop gets too high and it happens no matter what the airspeed. In that case the expo drops.
In general the numbers given are a good starting point and will be pretty good for most people.
Thank you @Leonardthall for sharing this guide! Lot of useful info in there!
I’ve seen you give some insights also on PSC parameters for Z. Great, thanks!
Can you please explain how to set up the other terms?
Why POS I term is missing?
1 Like
As for the setup of the XY POS parameters the only real guidance I can give is if you have set up your aircraft to be very slow in roll and pitch then you may have to halve all the PSC_XY parameters.
To answer you question on the I terms. You only need one I term in each control system to take out constant offsets. This is best handled in the lowest loop.
Perhaps to answer your question directly. We don’t need an I term on the position controller if there is an I term in the velocity controller. The only way there is a constant Position error is if there is a constant Velocity error. So we add the I term to the velocity (the inner most loop)
2 Likes
Amazing work here Leonard!! This is an area of the wiki really needed a touch up. Excited to see the improvement people see in there birds with all the effort that has been put into this. Let me know if there is anything I can do to help the cause.
-Ed
1 Like
Really amazing work, I re-tuned all my copters following the new wiki and I saw substantial improvements.
1 Like
Hello All,
This is a great effort and a huge help for beginners like me who generally jump too quickly into AutoTune, without setting-up the required parameters for starting a tune and end up in CRASH
Good news is that I have rebuilt the whole thing and getting it ready for first flight 
I have generated Thrust vs PWM data for four KDE-3520XF-400KV motors, KDEUAS55HVC (55A+HV) ESCs with T-Motor 18x6.1 props on an RCBench Series 1520 thrust stand. Its just one data point on the curve but I hope it will be useful to verify the approximate MOT_THST_EXPO curve.
https://www.dropbox.com/sh/3glldns2s184pnn/AAB4sx0tYDtVABXVcI3icuy8a?dl=0
I would be glad to perform the same on other Motor/ESC/Prop combinations that i have and share the data with community if it carries enough value to justify the effort 
P.S: Can you please guide on how to use this data to get the MOT_THST_EXPO value?
1 Like
@Leonardthall on this thread MicroArduCopter, 3" props, Omnibus Nano, Success! I read that also EK2_ABIAS_P_NSE is an important parameter to tune in order to improve stability on small (3 inch) copters. I would like to know what do you think about it.
I can’t provide any thoughts on this without seeing the issue myself.
Sorry, I have not had time to write up an explanation of this yet but there is a pretty big conversation about it here:
There is a bit of drama in that thread but I went to a lot of effort to explain how I choose that value and where there are some tricky things to deal with.
OK thanks the same.
I saw your work on system identification with chirp signal on the PR, it is really awesome.
2 Likes
Thank you! great and detailed description of the tuning process!
Can you describe the logic behind suggested filter hz values? Any relation to motor speed or build natural frequencies?
Any info on FF parameters use / setting with multi-copters?
The filter frequencies are really about the agility of the drive system and inertia of the aircraft. I have built very agile 30 kg racing multirotors that need filter settings similar to a 10 inch quad for example.
These instructions are focused on an optimal performance tune. These filter settings are much more complex if I am doing a noise limited tune but I have no idea how I could give instructions on a noise limited tune. (where noise levels are the such that the PID’s can not be tuned up to an optimal tune without the noise swamping the motor outputs).
Classic multirotors do not use FF. the FF is for aerodynamic surface controled aircraft like heli, single and coax copters.
Thanks for the great wiki page, started fine tuning our large hexacopter with 30 inch props with it this week. Changing the MOT_THST_EXPO param seemed to have the biggest impact as it was set at default 0.65 to start with. Tried stepping it up to .75 initially and then to 0.8. There was little notable difference between 0.75 and 0.8 but both had major improvements to control ability from the default setting. Makes flying feel a lot safer.
Current rig is running:
30 inch props
T-Motor U8II KV85 motors
T-Motor Alpha 60A HV ESCs
I saw in another post that T-motor ESCs might work well with a low EXPO value, but its running so well on the higher value I’m hesitant to try it.
1 Like
That would have been me. On some of the earlier t-motor escs they did internal compensation that messed things up. I have not seen it for a long time now though.
Good to know! Thanks for the update.