SMALL - espritmodel.com SMALL - Telemetry SMALL - Radio
Reply
Thread Tools
Old Feb 17, 2013, 11:11 PM
Will fly for food
ilektron's Avatar
United States, CA, Carlsbad
Joined May 2011
532 Posts
Only had a little time to string it up and spin up the motors. The yaw control isn't great. Only had PD. Think I'll add I and see what happens.

The Flying Cube - Clothesline Testing (4 min 24 sec)
ilektron is offline Find More Posts by ilektron
Reply With Quote
Sign up now
to remove ads between posts
Old Feb 18, 2013, 05:48 PM
Registered User
Joined May 2010
430 Posts
Turn off the D term, it will cause you a lot of headaches, the D term is looking at the slope of the change if you where to looking at a plot of the movement (error table), your craft is moving too fast. The D term works well on slow changing systems, i.e. the flood gate controls on a hydroelectric plant.

The P term tries to push back the on the dislocation of the disturbance in an equal "proportional" amount.

The I term looks at the sum of the errors left over from the P term, so you only need a small amount of I term summing. I.e. the I term looks at the total area contained within the curve of the error table.

Just use the P term at first, that is all you really need for rough tuning to get some control.

It sounds like you need to lengthen your yaw control vanes to increase yaw damping, they looked under-sized to me in your photos. Just tape on a couple of foam scrapes.
corocopter is offline Find More Posts by corocopter
Reply With Quote
Old Feb 18, 2013, 06:33 PM
Registered User
Joined May 2010
430 Posts
BTW, where is you CG location currently?

If it is below 23% then you will have a very unstable craft. Combined with any amount of D term and it will dart about as in your vid.
corocopter is offline Find More Posts by corocopter
Reply With Quote
Old Feb 18, 2013, 07:17 PM
Will fly for food
ilektron's Avatar
United States, CA, Carlsbad
Joined May 2011
532 Posts
I'm thinking of adding a servo and splitting the longer vane to give it better yaw. I only have 4 channels on my fcb, but I can bypass throttle and have it come off of the TX directly.

The CG is below 23%, probably at about 33%. Unfortunately, w/o adding weight or redesigning the thing, I won't be able to move it forward. I'm going to give this design some more effort before I concede that my system is uncontrollable.
ilektron is offline Find More Posts by ilektron
Reply With Quote
Old Feb 18, 2013, 07:21 PM
Will fly for food
ilektron's Avatar
United States, CA, Carlsbad
Joined May 2011
532 Posts
Oh, and hopefully in the next couple of days I'll be able to play around with just a PI on a calm day.
ilektron is offline Find More Posts by ilektron
Reply With Quote
Old Feb 18, 2013, 07:40 PM
Registered User
Joined May 2010
430 Posts
Quote:
Originally Posted by ilektron View Post
I'm thinking of adding a servo and splitting the longer vane to give it better yaw. I only have 4 channels on my fcb, but I can bypass throttle and have it come off of the TX directly.

The CG is below 23%, probably at about 33%. Unfortunately, w/o adding weight or redesigning the thing, I won't be able to move it forward. I'm going to give this design some more effort before I concede that my system is uncontrollable.
You could add a 4 light weight sticks (bamboo skewers) on top, tipped together in a pyramidal shape and tie you battery pack up high thus you could move the CG to any location with very little weight gain.

You only need two large vanes off-set for yaw control. It maybe better to enlarge your existing two yaw vanes and use the weight savings on the skewers and CG mode.

You need to get the CG further upward to above the MAC this will also give you better pitch and roll control. i.e. a great moment arm for these controls, but the CG location above the MAC is CRITICAL

As you said "keep trying on this device to learn more for the next version".

You'll get it to hover if you take these recommendations.

What about your lateral CG location? Did you check it ?

I enjoyed the video, thank you...!

Kelly
corocopter is offline Find More Posts by corocopter
Reply With Quote
Old Feb 18, 2013, 07:55 PM
Registered User
Joined May 2010
430 Posts
Just split the long vane and add the strips of soft steel from a tin can as mentioned, no additional servo is required.The off-set needs to be fixed at all times; set to the required off-set required during hover flight. The other two vanes that are under servo control will have sufficient travel then to make smaller changes in yaw direction required for normal flight control changes. Use the"non-servo control" off-set vanes to overcome the motor torque only......This is also very important and I can detail this later, but I go over it several times with Sam in the Jap Drone thread.
corocopter is offline Find More Posts by corocopter
Reply With Quote
Old Feb 18, 2013, 08:12 PM
Will fly for food
ilektron's Avatar
United States, CA, Carlsbad
Joined May 2011
532 Posts
I think I will try moving the CG around with some hacks just to see what I can change. Something I don't get, though, isn't the MAC is meaningless in a hover, right? The only aerodynamic forces exerted anywhere are on the control surfaces, meaning that the MAC is quite aft.
ilektron is offline Find More Posts by ilektron
Reply With Quote
Old Feb 18, 2013, 09:10 PM
Registered User
Joined May 2010
430 Posts
Quote:
Originally Posted by ilektron View Post
I think I will try moving the CG around with some hacks just to see what I can change. Something I don't get, though, isn't the MAC is meaningless in a hover, right? The only aerodynamic forces exerted anywhere are on the control surfaces, meaning that the MAC is quite aft.
Good question:

If you only had a stick frame with 4 vanes then MAC does not matter, the further the CG is from the vanes the better for control power and there are many other fine points worth consideration, but that is the basic idea for a stick frame device

But,.....

You have a HUGE amount of lifting surfaces, at least 70% above what I would consider reasonable...!

These lifting surfaces react to the airflow entering the craft via propeller inflow and during the slightest amount of translational motion (drift). These reactions occur (lift) around the MAC of the lifting surfaces, if your CG is below this reaction point the craft is Unstable. The craft must rotate around the CG, but the air flow will create intermittent forces (lift) center at the MAC causing the craft to gyrate to and fro. If you move the CG above the MAC these forces are dampened out and the craft is stable, this is exactly the same for an airplane in normal flight.

Your craft is going to move about regardless of your desires, especially while learning to hover, thus MAC matters a great deal.

If you greatly reduce those side plates then you'll have a nicer device to fly.

Sometimes "bad looks" work better....... and not just in the movies.

Kelly
corocopter is offline Find More Posts by corocopter
Reply With Quote
Old Feb 18, 2013, 09:36 PM
Registered User
Joined May 2010
430 Posts
Yet more to consider:

The CG should be exactly on the center-line of the propeller rotation. This is also very important. The gyroscopic forces of the propeller should be located on the CG. This follows the same argument I just mentioned about the CG/MAC location.

The gyroscopic fores of a propeller can be quite large with respect to a hovering craft since there is little airflow over various surfaces to off-set these reactions. An airplane set up as a tail dragger will very easily demonstrate the point. A sudden burst of power while setting on the ground will twist the craft into a violent ground loop.

A VTOL craft is very sensitive to all of these forces due to the lack of airflow to dampen out there resultant motions in hover.

Propeller /CG location is VERY critical, go look at the Flying Can UAV vids again and you'll see a very well designed system with the CG/MAC/Prop correctly placed.
corocopter is offline Find More Posts by corocopter
Reply With Quote
Old Feb 18, 2013, 10:10 PM
Will fly for food
ilektron's Avatar
United States, CA, Carlsbad
Joined May 2011
532 Posts
Quote:
Originally Posted by corocopter View Post
If you greatly reduce those side plates then you'll have a nicer device to fly.

Sometimes "bad looks" work better....... and not just in the movies.

Kelly
The most important requirement that I have for this is that it remain a black box. More specifically, a cube. I agree that I need more control surface area. I think I'll try the waffle fin like on MOAB of some of the larger Soviet bombs as not to disturb the outward appearance.
ilektron is offline Find More Posts by ilektron
Reply With Quote
Old Feb 18, 2013, 11:33 PM
Will fly for food
ilektron's Avatar
United States, CA, Carlsbad
Joined May 2011
532 Posts
Here is a sketch of what I want to do, create lattice fins. More torque with a shorter chord.
ilektron is offline Find More Posts by ilektron
Reply With Quote
Old Feb 19, 2013, 01:07 AM
Will fly for food
ilektron's Avatar
United States, CA, Carlsbad
Joined May 2011
532 Posts
Alright, so my actual implementation isn't quite up to my drawing, yet. Perhaps next time I'll do a proper lattice fin. Hopefully tomorrow I'll get a moment to play with the gains and see what I can get.
ilektron is offline Find More Posts by ilektron
Reply With Quote
Old Feb 20, 2013, 12:44 AM
Will fly for food
ilektron's Avatar
United States, CA, Carlsbad
Joined May 2011
532 Posts
Looking at my code, apparently both I and D were 0 for the video. Forgot to comment out something. Anyway, changing P and I to be controlled by channels 5 and 6 on my tx.
ilektron is offline Find More Posts by ilektron
Reply With Quote
Old Feb 23, 2013, 04:13 PM
Will fly for food
ilektron's Avatar
United States, CA, Carlsbad
Joined May 2011
532 Posts
Alrighty, I think the cube will end up flying quite nicely. I need to put something on it to mark orientation until it flies essentially hands off.

I implemented the PI rate control and tested it out. Overall, I think it looks like it will fly great. I need to reduce the zero point deflection of the elevons to compensate for the torque of the motor. I overcompensated and it only spins one way. Doesn't have enough motion in the other direction.

Video finally uploaded,
The Flying Cube - Part 2 - Testing added area to control surfaces and tuning PID gains (2 min 36 sec)
ilektron is offline Find More Posts by ilektron
Reply With Quote
Reply


Thread Tools

Similar Threads
Category Thread Thread Starter Forum Replies Last Post
flying cube! BigBadWolf_TX Power Systems 13 Apr 10, 2002 11:59 AM
Antenna coil from E-Cubed J Morgan Power Systems 8 Jan 31, 2002 07:48 AM
Floating Cube? Not quite! windsurfer Parkflyers 3 Dec 04, 2001 07:22 PM
Floating Cube simitar Parkflyers 15 Oct 18, 2001 12:03 PM
Floating Cube miker Parkflyers 12 Jul 10, 2001 10:48 AM