Skywalker build

[Jorn]

Tom, thanks for helpfull inputs!

For the safety-settings, my subjective short list for conditions to react on would probably consist of:
- manually being able to abort mission/rtl, by flipping a switch on the tx
- max distance from home
- max flight time
- low bat
- min agl

With regard to my wish for a separate tab in the config prog/config menu, for setting safety walues, and your reply; I would have preferrd to have both;
- default (conservative) values to be defined in the confign prog., and possibility to explicitly override in mission planning. Meaning; if not explicitly overridden, conservative values would be used. (Even if this might introduce some complexity).

One other thing; there is a setting for "min throttle" in the config tab. As I have experienced/understood it; If this value is set to anything but null, this means tx must be on all the time - and throttle manually set to 0, or the throttle goes to the "min value". (Even when tuning the ap at home). If I have understood correct, I would prefere the throttle was allways set to 0 - when in auto and takeoff not detected.

[Tom]

Hi Jorn,

I will take your suggestions into consideration for release 0.9

When being in autopilot mode, the throttle will be a value between min throttle and max throttle, depending on your target altitude. E.g. if your first command is CLIMB, the throttle will be max throttle.

Depending on your configuration in the Servos tab, your transmitter must be ON at startup or not. If no transmitter is detected, it will default to Autopilot mode.

The problem with your remark is: how do you detect take-off?

[Jorn]

I maided the Skywalker today. The plane is very stable and easy to fly by nature, and with huge space in the canopy for equipment. It even requires to be filled up, in order to have it balanced. I had to stuff it with two of my heaviest batteries (2 x 3S 3300 mAh) + 2 A-type 1,5V batteries to get it in balance!

Maiden flight went very well. After trimming out the plane, I switched to stabilize, which appeared to work perfect. I could not notice any wobbeling og strange behaviour of any kind. So I switched to auto, and flew a single circle, before landing in manual mode.

All in all very happy with the performance and looking foreward to do more testing.

Tom; about how to notice take-off, I can't tell. (I thought I read somewhere this was allready done - for setting the home pos).

[Jorn]

Today I made another flight test in auto, checking out a traditional aerial surveying pattern. The flight went very well, but there is obviously still more tuning to be done.

Generally speaking, the main problem is the plane was acting too "lazy" when turning around at the end of the flight lines, (turning radius about 55m). From the flight log, I see that the roll values in the turns most of the time was about equal to the max roll value, set in the config prog. Hence, I probably need to increase this value. Mayby to 45deg to start with. Also, the convergence to the flight line was pretty slow, (using the FROM_TO wpt programming style).

Any suggestions with regard to tuning?

By the way; for this kind of flight patterns, a possibility to write wpt's relative to previous wpt's would make life easyer. => to the wish list for next fw rev.

Also, I can't get the simulation to work properly, the plane just spirals around between the wpt's. (Same settings/tuning parameters as in the real flight was used).

[Tom]

Hello Jorn,

Wind is an important factor. When turning downwind, the circle radius will always be too large. When flying a survey pattern, the flight lines should be perpendicular to the wind. This will make sure the turns are always be with head wind.

You could always change heading2roll PID, which will increase the target roll angle with regards to the heading error.

A new airframe is always tuning work! The default parameters are optimized for flying wings which could be why a more sluggish plane needs additional tuning.


The simulation mode should be executed with the factory "default" settings (burned to flash!).

[Jorn]

Tom,

By studying todays flight logs, I think I found at least part of the reason for the "huge" turning radius and "sluggish" behaviour. Trottle was actually set to manual for auto mode. Hence, throttle/flight speed was the same as when leaving manual. (Caused flight speed to be maybe 50% higher then it should have been, and constantly negative pitch).

Next I will probably try out the chdk trigger. However, there is a pwm-pulse parameter to be set in this command. Is that really relevant when using chdk?

Jorn

[Tom]

Hello Jorn,

Indeed the CHDK trigger_start will only put a logical 1 on the servo output. The other parameters are not used. servo_trigger_stop will put a logical 0 on the servo output. Usually I use continuous trigger when CHDK reads a "1".