Draft Fybarless Heli Setup Process
Posted 14 December 2011 - 03:56 PM
Thanks for sharing.
Posted 15 December 2011 - 03:18 PM
At the moment I am working on the RC input and see some hope in results but not yet a repeatable or predictable response.
Hope to post more later today.
BY the way the board works fantastic on the multicopters. Been having great fun with it.
Posted 17 December 2011 - 04:22 PM
The jump at the bottom of the travel can be eliminated by making sure the Accessory 0 Neutral value on the input setting screen is set to same value as the channel minimum. During calibration neutral is normally set slightly higher than minimum value for the channel even if the throttle stick is at bottom. THis is because you will probably find the channel minimum when the stick is in the corner at bottom stick as opposed to the center at bottom stick. The jump issue is very reproducible and the fix seems to always work.
The change in swash cyclic behavior having two zones can also be eliminated. Having the two zone issue manifest is harder to reproduce but it happened several times during multiple test setups. Good news is it can be eliminated by putting the throttle neutral to channel minimum. As with the throttle this does not happen in calibration so you have to do this manually. I have several saved setups that appears almost identical. On some setups the cyclic issue is present and on some not. I have not identified the trigger for this behavior to start but as noted above, if it does present itself it can be eliminated. When it does occur the position of the throttle neutral determines where in the stick travel the border of the two zones will be present. The swash zone at the bottom is not affected by PID settings so don't set the slider to the max position. This will result in uniform swash behavior throughout the travel of collective but it is incorrect. Swash movement will be the same throughout the collective range but will not have the PID control loop active. I tried reducing P to zero and still got the same swash behavior. With Throttle neutral set to minimum swash behavior is controlled by PID settings.
The asymmetric collective movement where positive collective pitch range is much greater than negative collective can be fixed as well. It appears to be a artifact of the the limit of max manual input at ~90 of input stick. I think this was done to allow correction control even when at full throttle on multicopters. However, the effect of this is to case the collective travel to stop around 80 - 90 % of full collective stick movement and because the positive collective output is mapped to a smaller input range the positive collective range ends up substantially more than negative. This is when a correct manual mechanical setup is done and neutral is set to 0 degrees pitch. The fix is to use the data browser and set Channelmax for Accessory 0 channel (number 4 for me) under ManulControlSettings in my case to about 2200 microseconds. This moves the embedded ~90% cutoff point to the correct pulse width of about 1900 and gives full stick control and proper range at the same time. I did this thru several setups and got repeatable results. By the way this also affect tail positive and negative pitch range and stick control. The same fix works on the tail. Hopefully this limit can be removed at least for non multicopter applications.
One last issue I am working on is the limiting or cyclic when at max or minimum collective much as a cyclic ring setting would do. For max collective I do not have a issue on my setup when I set pitch range to +-12degrees. this is about 59% on the basic setting where pitch and cyclic range are controlled. At 100 % it does occur at max pitch. However at minimum pitch it is still limited. Most of the time this will not be a big issue unless you are doing some hard 3D or using max negative collective with max cyclic. Still working on this.
But overall with this fixes I am ready to give it a whirl again with some confidence now of success.
Any ideas on the minimum issue are welcome.
Edited by Pellicle, 17 December 2011 - 05:55 PM.
Posted 18 December 2011 - 09:59 PM
Posted 19 December 2011 - 09:31 PM
I'm currently getting a T500Pro up and running with the CC board. It formerly had a 3GX on it and was flying quite fine. Can you share your latest working configuration? I always like to build on someone shoulders rather than starting from scratch if possible.
Posted 22 December 2011 - 04:52 PM
I have a JR transmitter system so 1520 nominal center pulse.
Digital servos with 333Hz capability on cyclic and tail.
The setup is standard +-12 cyclic
For testing I have the flight modes setup to allow manual on the cyclic for FM1, Rate for FM2 and Attitude for FM3.
Tail is Axis lock on all 3.
You will need to do your own radio calibration. This will wipe out some of the manually input changes noted above.
Also the Rate settings I have now are preliminary. I have had to bring them down quite a bit to get rid of oscillations.
I am not happy with the impact on handling agility however. It seems manual agility and stabilization adjustments are intertwined so faster handling = higher stabilization gain and vice versa.
It is actually a relatively nice day here in Ney York so I plan to do some actual flying today with it and see if I can tweak some more.
Posted 22 December 2011 - 06:26 PM
Thanks for sharing. I will be consulting your uav file.
I received my CC board yesterday and now trying to get familiar with the GCS entries. So far all I have done (since last night) is do initial setup and emulation on the bench while trying to come up with a step-by-step procedure (from setting up the radio, wire connections to CC config) which a non-technical user can just follow without knowing all the technical stuffs. Hopefully it can be done.
Would be nice to know how your test flight go.
Posted 22 December 2011 - 10:00 PM
Kp on Aileron and elevator are raised to .0022.
To prevent oscillations at this level Gyro Tau has been raised to .04 from .005 default.
I still want to increase Kp higher. Look at adding Ki back in as well as the effect of minimal derivative.
But no crashes today!!! A definite improvement. A few flights and videos. A few test hovers in the shop at varied settings and some bench testing of Gyro Tau impacts.
Handling is much better at these setting compared to the previous file at least for my setup.
Further explanations later but a good testing day with progress.
Main focus today to get reasonable handling response by maximizing Kp. To maximize Kp we had to find ways to make the CC less prone to oscillation. Previously we were getting oscillations at .0013. Mounting on gel seemed to increase to ~.0018 and increasing Gyro Tau to .02 to allowed us to go to .22 but with very slight oscillation on sharp cyclic moves. Increasing GT to .04 eliminated all trace of oscillations at .0022.
Ran out of test time but next step is to see how high we can get Kp at current setting. Then experiment with a little Kd.
No stings attached for Kp testing!!!!
Posted 02 January 2012 - 07:14 PM
Posted 08 January 2012 - 05:28 AM
Posted 08 January 2012 - 05:33 AM
Posted 10 January 2012 - 05:03 AM
No significant work done yet on Attitude settings. It did self level when in attitude mode but the speed was a bit slow. Will look at this after working on tail parameters more.
Trex 500 1 07 01.uav 8.94KB 10 downloads
Posted 10 January 2012 - 08:02 PM
Still not the ability to decouple handing agility and stabilization or multiple gain settings as in most systems but that will take some other changes and hopelfully will appear as teh platform has time to mature.
I have been pushing for this for a long time, very difficult to express how important this is to heli pilots. What I might do is ask Oleg to implement it in one of his branches, it is a must have feature. However, I stopped arguing about it a while back and let the core team get in to helis more and then they will see the need for it.
Posted 11 January 2012 - 03:14 PM
Still not the ability to decouple handing agility and stabilization or multiple gain settings as in most systems
Meaning you can't adjust the response from your inputs and the the response from external inputs like wind, right Pellicle?
Posted 12 January 2012 - 05:21 AM
I will continue to improve this setup and update the results on the stabilization tuning page http://wiki.openpilo...ning+Helicopter
Trex 450 w/ JR DS3500G tail servo , linkage ball is 7.8mm out from center, 61mm KBDD tail blades
Old settings: Kp .006 / Ki .005 / Kd .000003
New settings: Kp .004 / Ki .012 / Kd .000010
1 11 2012.uav 8.91KB 7 downloads
Posted 12 January 2012 - 06:18 AM
Posted 12 January 2012 - 11:22 AM
I see in your .uav file that you have the LowThrottleZeroIntegral set to TRUE. Do you notice any issue during autorotation when the TH is engaged? Maybe I'll try that when I have another chance.
Posted 12 January 2012 - 06:35 PM
Posted 13 January 2012 - 12:26 AM
Looking the the Kp and KI values you have I would assume you are working on the Rate Stabilization (inner loop) coefficients, correct? So does it mean you just working on the Rate Stabilization mode at this time and have you tried making use of the AxisLock for the Tail already while hovering?