Proof that 2-cell scale planes work!
I put down the flyer at the start of 2001 when I eventually got small (<25gram) fixed wing r/c planes to work. I have left this page here for those who are interested.
Be warned I have beend fairly convinced (by heli people) that using the ADXL202E 2-axis level sensor will not work :( It is simply the wrong sort of sensor, it does not give the gyrosopic reading that is required. It may be possible to mount a unit on each arm and probably a fith in the middle, but this is starting to look expensive (and complex). Analog Devices do have a new gyro sensor on the market but it is not a cheap item.
The easiest solution would be to strip down 3 GWS gyros - maybe a bit expensive?
I suspect that the mixing for this is just about possible with the 20Mhz PIC.
I have had this interesting suggestion emailed to me by hurdeR@pro.hu who found it in a yahoo group ( I can't find it!):
``For who is interested, ``I made a draganfly-roswell clone myself and solved the gyro problem ``with a very easy solution: ``The idea was to take 4 liquid-metal switches and mount them at the ``far ends of the arms,with an inclination of about 2 degrees. ``So when the arm drops below zero degrees the switch makes contact. ``Now with little electronics a circuit checks for inputof the ``receiver to see if the movement is caused by the operator of the ``transmitter. ``If yes,no correction is made.If no,then the power of the motor will ``be increased in steps until the switch breaks the contact. ``This is been done on all 4 arms ,so onces it is in the air i dont ``have to make any corrections at all. ``This is a low cost solution. ``If anyone is interested i will post the plans and schematics in the ``file section. ``Anyway i am interested in sites or persons that have been doing some ``experiments or have build one themselves,to exchange information.
This solution is well within the scope of a PIC, it is so simple I might even get interested again.......
For the real thing see http://draganfly.com/products_4eh.html
Basically 4 motors pointing up, opposite motors are counter-rotating.
This project was started around October 2000 and abandoned at the early prototype stage. The design uses an AXDL202 2-axis level sensor to detect tilt of the flyer, combined with a gyro to detect yaw which these there is a fighting chance of controlling it. The control algorithm is in an 8-bit 18F64 PIC running at 10 MHz which handles all the mixing.
The following is a block diagram of the schematic
The original code is here.
The original prototype used 4 geared down scalextric motors. To get to the astounding height of 3 inches 17Volts was put up them, causing overheating after a couple of minutes- but it nearly flies. However this was enough to show that the algorithms worked although gain is a problem - too much and the system oscillated, too little and the sensors may as well not be there.
Geared Kenway N-20 motors available from Bol Selmans site are the choice for propulsion. An IR control version may well follow but since the original was based on the 4 channel GWS receiver then this will be used in the new design. I now have a new GWS that will be stripped down to minimum weight. The intention is to use surface mount electronics, the following is the parts list:
ADXL202E 2-axis level sensor, I've not found a source in the UK for this (analog devices sell direct in the US) - I'm using free samples try analog devices (even better!) the other parts are listed with farnell part numbers.
PIC16F84A-20SO 343-5106 p558 £4.66 IRF7456 353-8096 p375 £1.40 Need 4 20 Mhz resonator 648-190 p601 £1.72 Resistors 0805 size 100R 109-306 p170 £0.026 (min 50) 1K 109-312 p170 £0.026 (min 50) 680K 109-329 p170 £0.026 (min 50) Diode LL4148 739-182 p401 £0.054 (min 10) Capacitors 805 size 0.1µF 644-160 p8 £0.109 (min 10)
IRF7456 FETS have been chosen since they seem within spec. and are
fairly cheap; the SO-8 layout gives the opportunity for swapping with other
devices.
The schematic is below, note that the gyro takes care of the yaw
and is simply in-line with the receiver output. Provision is made for David
Tait style in-site programming. An RS232 output provides a means of diagnosing
the code.
The new design uses 5ms sampling on the ADXL202E, and I've changed the filtering from 10Hz to 50Hz (To save on buying more capacitors!). The code above as not been modified for the 20 MHz PIC or to take into account the above changes.
The following board layout was originally a bit tight around the level sensor, I've increased the pad size accoringly. Print at 600 dpi and it should come out as 31.22mm by 28.16mm!!
Here is an overlay of where the components go as showm in the following picture (before the FETs go on).
In-circuit programming adapter
Here are some links for PIC programmer software and hardware and compiler that I am using.
11/12/2000 The latest code gettting closer!
12/12/2000 The first test rotor is under way!
13/12/2000 Tests on the scope indicate the the FETS are being driven OK less than 10mV across the FET when switched on. The algorithm is a bit out, it only gives full thrust on full throttle, full left and full tilt to the right! The 300Ma batteries are giving well over 2A (shorted with a 1R resistor) on there first charge!
13/12/2000 Tests on the scope indicate the the FETS are being driven OK less than 10mV across the FET when switched on. The algorithm is a bit out, it only gives full thrust on full throttle, full left and full tilt to the right! The 300Ma batteries are giving well over 2A (shorted with a 1R resistor) on there first charge!
29/12/2000 Photo of the completed chassis with four of Dr. Chris's rotors::
Position of the GWS receive underneath the PCB:
The problems so far:
Not quite enough power to lift the batteries this may be solved by cycling them? For now I'm using an umbilical cord.
With the motors running, the level sensor does not appear to work, the flyer flips over.
Even with the level sensor disabled the PWM of the motors is not working correctly - is this supply ripple???
3/1/2002 Looks like it is supply problems, Ive just disabled the level sensor and temporarily powered the motors from a bench supply. Managed to just leave the ground to about a height of about 5mm for ½ a second!! - It is difficult to mix manually!! I'll rig up something more satisfactory and do some more tests.
Some idea of the size, electronics on loan for another project!
