Newsgroups: comp.robotics
Path: utzoo!utgpu!watserv1!watmath!mwtilden
From: mwtilden@watmath.waterloo.edu (Mark W. Tilden)
Subject: Re: Small Hovering Hobby Robots
Message-ID: <1991Jun14.204833.9470@watmath.waterloo.edu>
Organization: University of Waterloo
References: <00949D78.A594B600@vms.csd.mu.edu> <26776@ttidca.TTI.COM>
Date: Fri, 14 Jun 1991 20:48:33 GMT
Lines: 209


The following are some articles I've kept on the subject.  Might be 
still worth a read.

-------------------------------------

To: outram@cernvax.cern.ch
Subject: Re: model flying saucer - help/advice needed.
Newsgroups: sci.electronics
In-Reply-To: <1746@cernvax.UUCP>
Organization: University of Waterloo
Cc: 
Bcc: 

In article <1746@cernvax.UUCP> you write:

>                             ___ 
>      TOP VIEW :            / O \      where 'O' is an engine.
>                           |O + O|
>                            \ O /
>                             ---

Limit it to three motors and install a microsequencer (rom based)
to do high speed dynamic stabilization control for you.  Nothing else
is fast or light enough.  Three motors gives you better lift area
while simplifying the control problem to tripod balancing.  The actual
balance mechanism should be two loops of plastic tubing half filled with
murcury (the other half with heavy oil).  Paint the surface of the tubes
half way with high conductive paint. Stick an insulated wire in contact
with the murcury.  Feed both outputs into a capacitance meter circuit
and you have a damped stabilizer of reasonable accuracy.  Far cheaper than
building your own gyroscope.

>Has anyone made one of these things before ?  What difficulties where
>encountered ?

Yes, I made one out of carved styrofoam as a frame.  Biggest problem is
finding the ideal lifting fans.  As for control, drive all fans from a 
single high-power electric motor (used in rc cars) driving triple
magnetic clutches.  You'll have to make these.  Hope you have a Unimat 
lathe handy.  The motor, as the batteries, must be mounted dead center.
Using 100mA batteries, you can expect a maximum 2 min flight per charge.
Longer if you want to run it by 'wire' of course.  You will also want
a speed regulator for the motor or control instability will result.

Your biggest problem is weight and control.  Too much weight and all
you'll have is a floppy hovercraft.  Control is also the major bitch.
You require very high speed/accurate feedback to keep the thing from 
flopping hard.  One of the things I did was install a deep murcury switch
which only registered if the machine flopped > than 90 degrees.  When triggered
it replaced the balance circuit control with signals to level the craft at
all costs.  You won't have to slamm the thing into the ground too many times
before you realize how handy this is.

I'll tell you now, such a thing really isn't as fun as you might think.
It is very difficult to control and if your motors aren't precessed exactly
the damn thing is always spinning on you.  This won't matter if you can 
find a gyroscope you can lift but otherwise you'll have hard times ahead.
The only way around it is to leave the lift motors always stabilizing the
outfit and install control vanes around the outside of the craft which
allow retro-rocket like control.  Works, but very boring, noisy and unstable.

Anyway, that's my two cents.  Good luck with it.

--------------------------------

Newsgroups: rec.models.rc
Subject: Re: Flying saucer - the beast emerges !
Summary: 
Expires: 
References: <1765@cernvax.UUCP>
Sender: 
Followup-To: 
Distribution: 
Organization: University of Waterloo
Keywords: 

In article <1765@cernvax.UUCP> outram@cernvax.cern.ch (Nicholas Outram) writes:
>   The craft will be powered by FOUR rc variable speed engines to
>provide lift and small amounts of side thrust for horizontal mov-
>ment. Two more small electric motors provide spin stability and
>rotation capability.

I suspect your craft will be far too heavy.  Might want to
reconsider three lift motors for better lift vs. effective area. 
As you're using a processor, fast calculation of the centroid 
of the device will be no prob.

>   Horizontal stability controlled by Mark Tildens 'mercury gyro'
>idea. ...

>-> I still need help of how to sense the rotation of the craft <-
>->  Something like an electronic compass ??????                <- 

Just now I've had a better idea which might kill two birds with
one stone: rotation and stability.  Place 4 double contact mercury
switches on a shallow bowl, the hub connected to a small stepper
motor.  The outside edge of each switch raised to a yet-undetetermined-
angle.  Rotate at a fixed speed (defined by uP) until the all switches
*just* register closed at the outside edge.  Any tilt will register as 
a open in the direction of lift and any rotation will be detected as
uniform open contacts compared to the ideal spin motor value (this 
value being predetermined by experiment).  Connections to the switches
made by pickups on the outside of the dish.

Granted rotation detection can only be made in the counter-stepper-spin 
direction but if the craft is designed to spin against this direction 
anyway, then the problem is solved.

Granted also the signals coming back might require their own processor 
for correct filtering and control, but it's a neat idea.  The range
of values of rotation speed and switch angle should make up for the
inherant errors generated by the switches.  Wonder if it's
practicable?

>-> Also how do I convert small capacitance values to a form    <-
>->  easily input to a computer ?????      (0-1nF)^             <-

Make it part of a temperature-insensitive transistor based oscillator with
a square wave output.  Feed this signal into the strobe of a 8-bit latch 
(positive edge) and a delay line which will reset the 8 bit binary
counter which is feeding the latch (at positive edge + a tad).  The clock
source for the counter must be a Xtal source timed to be approx 128 times
faster than your transistor oscillator.  Thus, if your t-osc frequency
goes up or down, you have a 128 bit window to play in.  A simple extension
to this circuit will allow you to multiplex multiple capacitive inputs
to the same output latch.  As your lower 2 or 3 bits of your data will 
be random error, replace these with a binary value corrosponding to 
which capacitive tilt meter you're referencing.  These will, of course,
be driven by their own counter, the clock input from the t-osc.

A five chip solution. 

Is all.


------------------------------------

Re: Flying saucers:

>In one of yor artciles you made said that you have been able to construct
>such a beast.  Could you be a little more detail about these magnetic
>clutches?   You see, I'd like to make a platform with three rotors, and
>I need some way to vary the thrust that each puts out for steering control.
>My instaneous idea was to use three motors, and vary the speeds, but that
>would be far too heavy.  Second idea, use one central motor, driving var.
>pitch rotors (mechanical nightmare and heavy).  It would seem that
>using one motor with three "clucthes" would be ideal, so any ideas in this
>area would be greatly appreciated.  Thanks, Jack Brooks
>6843BrooksJ@vms.csd.mu.edu

Yes, that's pretty close, I think.

Magnetic clutches are available from a variety of industrial
suppliers but the best thing is to find one at a surplus shop.
They are small, rotary bearing shaped devices with two wires coming out
of them.  A good substitute is to find a source of pancake walkman motors
and direct drive the shaft.  If you short the leads from these motors,
the back-generator effect causes a drag proportional to the efficiency of 
the motor.  Pulsing this effect with a single transistor does a good job
and is very energy efficient.  Fast too.  Only problem is the 
additional weight of the motors.
                                                                    
My original flying creature was a flat thing made out of styrafoam.  When
all was said and done it was not really very exciting and took up far 
too much power.

If I were to do another one, it would look more like this (side view):


Magclutch and paddle  -                            #------=       
Fan vanes and mount   -           = ----+---- =====|===== ----+---- =
no-slip band drive    -                 |----------|----------|   
drop frame and control-                         \_ | _/         
                                                  \|/         
Single hi-ef motor    -                           [=]        
                                          
                                                                    
A single, low high-efficiency motor drives three lifting vanes by a no-slip
band drive (nylon O-rings).  The same shaft which drives the vanes 
also drives a free-rotating magclutch which is controlled by the onboard
intelligence.  I found that only a very small reduction in the available 
airflow to one lifter vane was sufficient to steer the device without
damaging it's lifting ability.  Saves on control vanes too.  By mounting
a small 'paddle' on the magclutch, you could limit airflow to any of the 
three lifters by a simple feedback control circuit.  Also, if the paddle
position can be controlled from side to side of a single vane tube, you can 
also control roll and direction.  Unlike my original device, I now know 
that if the majority of the weight is 30 degrees below the lifting 
points, the device is self-stabilizing, so that helps.

Neat thing about this design is that with the new 17% efficient solar-cells
available, it could be entirely self-supporting.

Could be.

I'm sorry, are you building a flyer, hovercraft, model or full-size version?
This idea is really only good for a model of some sort.

Is all.


Mark Tilden.

-- 
Mark Tilden: _-_-_-__--__--_      /(glitch!)  M.F.C.F Hardware Design Lab.
-_-___       |              \  /\/            U of Waterloo. Ont. Can, N2L-3G1
     |__-_-_-|               \/               (519) - 885 - 1211 ext.2454,
"MY OPINIONS, YOU HEAR!? MINE! MINE! MINE! MINE! MINE! AH HAHAHAHAHAHAHAHAHA!!"
