PDA

View Full Version : Discussion nMine UAV Controller, more hardware options than anybody could need!!!


jlcortex
Sep 23, 2008, 09:26 AM
Hi, I would like to present my project, i know there are a lot of uav projects but i think it is a amazing balance of price, performance and simplicity!
It is based on OLIMEX E2468 board and Sparkfun sensors.

Features
4 Layer board with ground plane for high noise immunity I/O Connectors:
8 x Servos - Standard pitch 2.54mm
4 x speed input (for sensors, motor RPM, or PPM capture)
2 x CAN BUS with transceivers - Header 2.54mm
10 x Analog Inputs (8 x 16bit - 5 used for gyros and pressures) general propose 0-5v (for sonar altimeter for example)
4 x UARTs (one used for USB port) (3.3v level) for GPS and MaxStream XBEE-Pro 60mW radiolink
Onboard sensors options:
Accelerometers (two options):
MMA7260Q 6g- analog
LIS3LV02DQ digital 1mg resolution
Gyros:
Analog Devices ADXRS300 (any other possible!)
Compass (two options):
HM55B 2 axis 6 bit
PNI Corp's MicroMag 3-Axis Magnetometer
Onboard Pressures:
MPXV5010DP 1.45psi (Airspeed)
MPXA4100A6U 15.2PSI (Altitude)

Features from Olimex LPC-E2468 development board:
MCU: LPC2468, 16MB SDRAM (more is possible if necessary), 128 MB
Flash, standard JTAG connector for debugging with ARM-JTAG
Ethernet 100MBit
2x USB hosts
USB-to-RS232 converter device connected to LPC2468 UART
SD/MMC card connector
Dimensions: 90x60mm (3.54 x 2.36")

we have make all hardware access rutines in C, included SD card writing and reading with DMA!!!

it is a simple micro boards, it is ARM7 so it is easier to program than any 8 bit micros (pic, avr...). Power and memory are enought for the AHRS's Extended Kalman Filter and for another EKF for INS system.

this board can run under ulinux!! (we are not using it)

It has been designed and build for "UNIVERSITAT POLITECNICA DE VALENCIA", special thanks to DISCA departament.

Jose Luis

rich smith
Sep 25, 2008, 05:11 PM
IMO ARM7 is more powerful but not easier to program than 8 bit micros. I do all these types in my business and for example, find AVR instruction set friendlier, architecture easier to work with, and flash algorithm simpler.

Downloader is easier to implement and cable is just 3 wires. More compilers, assemblers, and disassemblers available. Both developement tools and chips themselves lower in cost than ARM7 which is overkill for most apps. But that's just my opinion.

it is ARM7 so it is easier to program than any 8 bit micros (pic, avr...).

jlcortex
Sep 27, 2008, 07:45 AM
hi rich smith,

perhaps i have not explain correctly, in my experience 8 bit microcontrollers get very complicated when you go near their limit, with complex flight algoritms i think 8 bit micros are very near to their power limit.

i really think 32 bit are the correct choice for this project, of course, it is my also opinion.

Jose Luis

jlcortex
Sep 27, 2008, 07:46 AM
Hi,
I am building a experimentation plataform it is based on FC600 from UAVP (www.lipoly.de). Movement is limited to 3 degree of freedom with a rod end, it is sure way to work until all control loops works perfectly.

I will connect motors to a powersupply in order to remove batteries for indoor testings.

Jose Luis

rich smith
Sep 27, 2008, 12:01 PM
Jose,

Yes, I agree. For some very complicated programs with floating point etc. ARM libraries can be nice. My statement related to RC software in general where integer math is adequate in 99% cases and even assembler can be used most of the time. Bad experience with Paparazzi has left me somewhat prejudiced against ARM even though I work that platform almost daily because my clients require it.

UAV apps like waypoint sequencing and RTL can be implemented without floats or high level language. I've got 32 and 64 bit math routines for AVR to do things like arctan and Kalman and although it was hard work the Paparazzi stuff was much more difficult. Getting tools running was a big problem compared with AVR and x86 so left a bad taste in my mouth.

Good luck on your project.

-Rich



hi rich smith,

perhaps i have not explain correctly, in my experience 8 bit microcontrollers get very complicated when you go near their limit, with complex flight algoritms i think 8 bit micros are very near to their power limit.

i really think 32 bit are the correct choice for this project, of course, it is my also opinion.

Jose Luis