I need auto pilot help!!!

[Eddy]

Here is my website, cpl.usc.edu/eschuste
click on bedstead video, I need to make this stable, any ideas.

[Unterhausen]

how are you controlling it in that video? Is it just open loop? Do you have sensors? Obviously needs some damping.

[Eddy]

It has gyros on it, what can I use that is better?

[Unterhausen]

to tell you the truth, I was really impressed that you got it as stable as it is, that's quite an accomplishment. It looks like it has enough control authority and you are going to have to add a better control algorithm.

Accelerometers would help. It's possible that a second roll gyro would help you get rid of the bias and noise from the one you have. Difficult to know without more info.

[Eddy]

How would you use two gyros together?

[vespa]

Wow! Very impressive video!

If you are using your own microcontroller it's pretty trivial to add gyros but I assume you are just using hobby gyros, in which case combining outputs is not possible. You may likely get adequate short term results with some wings for damping, 3 really good r/c heli gyros, and a good pilot.

Basic autopilots like the Kestrel or Micropilot will probably suffice but maybe only for takeoff, as they could easily become confused in flight. If you can manage the cost and weight, the Cloudcap autopilot will give you best results as it is a full kalman filtered INS system, meaning that it uses very sophisticated prediction algorithyms to correlate your 6-axis motion with the plane's location in 3D space.

[Eddy]

Kestrel? Who makes this auto pilot? Kind of suiting seeing how this is for a Harrier.
I am currently trying to get a hold of the cloud cap guys, but this may be a really expensive way to make this happen.

[vespa]

The Kestrel is made by Procerus (http://www.procerusuav.com/) and is about the same $5k-$7K as every other autopilot. I'm not sure what you consider "expensive" after building a turbofan with 4 thrust vectoring nozzles. ;)

What is your method of control? Looks like you're keeping the thrust vectoring fixed and running a some bypass air out to 4 little nozzles on those tubes? How fast is the control response and what gyros are you currently using? I wonder if you could simply reduce your system's response time and crank up the gyro gain?

[Eddy]

Reduce the response time by adding the wing and maybe extra weight below the engine? I am using a 2 axis futaba gyro.

"I'm not sure what you consider "expensive" after building a turbofan with 4 thrust vectoring nozzles."


LOL
That is why I am looking for a not to expensive way to do this, this project is putting me in the poor house!!

[vespa]

Keep the c.g. as close to the thrust plane as possible to avoid coupling of translation and rotation. What is your control system? Can you make it faster/stronger? Without the gyroscopic and flybar stabilization of a traditional heli, you will need a very strong control loop.

[Eddy]

I am using servo controlled valves at the end of each pipe. I'm using futaba gyros to try to stabilize it. Are you saying I should have the cg of the bedstead
just under the engine's main nozzles? I am not sure I understand this right.

[mlbco]

Eddy,

I can't tell from the video just what controls you've got, but getting something to hover that has independent authoritative controls is not that hard. First you need thrust, yaw, roll, and pitch actuation that is authoritative and not highly coupled. Coupling can be caused by gyroscopic effects (i.e. spinning turbine mass) or thrust misalignment relative to the CG. The primary stabilizing feedback for the high-frequency dynamics is attitude plus rate feedback on all rotation axes. Are you using attitude (i.e. heading hold) gyros or rate gyros? Have you run out of available gain? If you are maxed out on gain, your control authority may be too low. Are there any significant (>0.1 second) lags in your system between input and response? If you have gas thrusters and valves you should have little lag, but if the system isn't designed well there could be a significant delay between input and response.

I have built several hovering UAVs and all you need to get the basic stability is attitude and rate feedback on 3 axes. A good set of heading hold gyros should do the job if all the other issues I mentioned have been resolved.

Here is a link (and photos) to a hovering UAV I'm working on that uses a similar control strategy but has my own autopilot installed instead of off the shelf gyros:

http://www.rcgroups.com/forums/showthread.php?t=653652

Good Luck,

Steve Morris
MLB Co.

[vespa]

Yes, a high c.g. may seem counterintuitive but removing any "pendulum" stabilty allows the active stability to have total control. For example if the c.g. is too low and the pilot pitches forward, the resulting forward acceleration will cause the craft to pitch further forward due to the pendulum effect. The opposite is true when trying to cease motion, and the net result is a craft that wanders randomly, always wanting to diverge in some random direction.

I would suggest the fastest digital servos possible for those valves, something below .9sec and try to improve the flow rate and linearity of the valves. The more responsive you can make it, the better the gyros will be able to assist. You should be able to increase the gyro gain to the point where the plane rocks nervously. Maybe you could "suspend" the plane with tight longitudinal cables to fix pitch, yaw, and altitude so that the pilot can focus only on roll while you experiment.

What is your bypass ratio and how much of it goes to the control valves?

[Eddy]

Steve, thanks for the help also, I believe I bought a dieselized motor of you a few years back, you sell them right? Nice job on your VTOL

I am using .09 Speed (sec/60°) servos, I could go to 5 cells and increase that to .07 seconds. I am maxed out on the gyro and have really long arms on the servos and the drive on the transmitter turned way down. I have servo amplifiers that increase the servo drive by 2 times what the gyro is sending, but when I turn the system on the servos max out to one side until the gyro has gone for the 3 seconds to activate itself?

I bought a thing called helicommand, it is amazing on a heli but I have not had any success on the bedstead with it yet, I am also maxed out on gains.

The turbo fan has a 3:1 by pass and roughly 5 pounds of air through the entire puffer system. The fan is also counter rotating to the core to reduce gyroscopic forces. It has puffers on it for roll pitch and yaw. I am just using rate gyros the heading hold is not good to use for this as it does not hold a the puffers in a way that will allow for a trimmed level flight, it may be ok for yaw.

With the servo amplifiers and helicommand I did have the bedstead rocking violently but had to cut the tests short that day, I wonder if I was on the right track?

[mlbco]

Eddy,

Can you record any of the dynamics data on-board? It sounds like you might be lacking control power or there is a big nonlinearity in your puffer controls. Do you measure a linear increase in applied torque with linear deflection of your jets? How much angular acceleration can you generate?

I know that heading gyros seems inappropriate for the tests you're doing, but once you find a way to properly initialize them they should offer greater stability. Rate feedback will slow the dynamics, but won't make it stable.


Steve

[vespa]

I agree with Steve that heading hold gyros will perform far better. Your helicommand unit may operate this way but if not, I suggest a pair of Futaba 401's.

What is your valve system? Do you have two central diverter valves, one for pitch and one for roll? Or 4 separate butterfly valves? How is yaw controlled?

[Eddy]

This how my rear puffer looks, the pitch and roll puffers look the same as this without the yaw puffer in it.
How do I record the dynamics data on-board? With the heli command I did see the bedstead get violently rolled back and forth. I would imagine I would have the control power needed or is this not true?
How do you use heading hold gyros? They are off center before I even take off, I am sure the bedstead would roll or pitch before the gyros had a chance to straighten it.

[vespa]

A regular gyro just provides "dumb" damping and in the process, damps the pilot's controls just as much as anything else. This really limits the usefulness of the gyro because the more damping it provides, the less control the pilot has.

A HH gyro is just the opposite - it does exactly what the pilot commands, nothing more, nothing less, ignoring wind, trim, etc. and just keeping the attitude where the pilot told it to be. The drawback is that it will drift a bit, so you will need to "center" it shortly before takeoff and again before landing - you could easily tie the reset function in with your thrust vector control so that it is automatic.

Another idea for increasing your control response may be to experiment with the neutral position of your puffer valve. You can get some strange nonlinearities in those long tubes and the turbulent valves so it may prove more responsive with the neutral position being 25% open or 75% open instead of 50%. I suspect a 25% neutral would provide the most response.

[Eddy]

I will try to map out the puffers over the next week, your guess at 25% is good, just from my experiments I have found that is what works best and that is where they are right now but still really hard to control, hopefully mapping them will tell me more.

[Eddy]

I did a few tests and found the puffers do work linearly, there is a slight fluctuation when other nozzles are opened.
I will also try moving the cg flush with the bottom of the nozzles, or should it be on the centerline of the motor?

[vespa]

I think it's more important to have the c.g. centered on the fuselage profile area than on the center of thrust and in practice you won't be able to achieve either, though Steve and others have proven that you can never have the c.g. too high. You may also want to experiment with wings and outboard mass as these will significantly slow the planes movement. But really, a pair of 401 gyros are the key.

[Eddy]

I have GYA 352 gyro, do you think that is different from the 401's?




Here's more info on it.

http://manuals.hobbico.com/fut/gya352-manual.pdf

[vespa]

Good! looks like the 352 is the same as the 401 just without some of the high-speed features that only the most responsive helis would need. Just set it up so that it goes to HH mode whenever you engage vertical thrust and you should be much happier - just don't sit on the ground for too long with it in HH mode as you can expect around 30 degrees/minute of drift. They don't say what the maximum gain of this gyro is, and I wonder if it is lower than the heli model, but you may want to test it to see what angle corresponds to full deflection - if it's something large like 45 degrees than you will never be able to utilize the full power of your puffers unless you are just about to crash.

[zaguruinzasky]

How can the GYA352 be used when we have a servo for each aileron?

[Eddy]

A "Y" harness to each servo

[zaguruinzasky]

and how do I manage the asimmetric endpoint of each aileron if connected with a Y cable?

[Eddy]

Not sure I understand your question

[zaguruinzasky]

oh forget forget, I was confused about my setup.. all clear now.

[zaguruinzasky]

actually I'm still confused ahah, I mean, my right aileron goes up 8mm and down 20mm, and same for left aileron. That's what I mean with endpoint, if I control both servos with a Y cable how can I set the end point for each of them?

[Eddy]

Are your servos centered properly? Or do you have something programed in your transmitter that is doing that?

[vespa]

You should only resort to a crude hack like endpoint adjustment as a last resort when you cannot adjust the linkage properly or when the surfaces serve two different roles with two different requirements (i.e. elevons where you want aileron differential but not elevator differential). In Eddy's case and most any other, differential is done in the linkage so a y-cable is fine.