I have not posted up anything in a month so thought I should give an overview of what I've designed. Unfortunately I have hit a LARGE snag in that BatchPCB is basically broken for international orders as no shipping options appear and so you cannot buy anything (its been like this since January). 5 months of waiting to produce these (no international shipping) and I'm pretty fed up, frustrated, shocked and losing faith so here for at least a quick look are my current plans.
First off I've been designing a quad to be smaller than ArduCopter but still hold everything so I can develop my camera stuff more quickly. It changed from basically a clone of ArduCopter to a slightly bare version to its current incarnation which is basically board holders with arm holders. The entire design has mounts for everything in the ArduCopter kit as well all the extras (pan and tilt, optical flow sensor(s), sonar, magnetometer, telemetry). As I progress with the build (or hopefully finish it with ease) I will update on here but the picture above is the progress so far including the 360° pan and 120° tilt bracket with sonar mount.
Now for my designs that I seem to be perpetually waiting for. All pictures are animated GIFs that ning apparently hates.
Optical flow boards
- ADNS - 2610 with ATMega328 for translational and I2C communication with APM
- ADNS - 5050 with ATMega328 for translational and I2C communication with APM
- ADNS - 5050 with ATTiny85 for translational and I2C communication with APM
- ADNS - 3080 with ATTiny85 for translational and I2C communication with APM
5050 seems to be the best sensor for our needs compared with the other two as the 2610 isn't available in the UK and the 3080 is 20 pin rather than 8 and only marginally better.
4 of these on my quad should hopefully stop it hitting walls. I'm hoping to create a library for these so they can be called for navigation easily like the rest of the I2C hardware for APM/ACM.
ACM distribution boards (4in² each).
2x two layer boards
- Power distribution below
- ESC distribution separating out the BEC 5V supplies to use for other things
Nothing new here but anyone fancying a smaller quad might find it useful.
ArduSlave boards (pictures aligned over I2C pins to compare size).
- I2C controlled ATMega328 with PWM outputs only
- I2C controlled ATMega328 with all pins broken out
- I2C controlled ATMega2560 with all pins broken
I am hoping to get the PWM only board done so it can be used as camera control circuit over I2C. The idea being to offload the computation to the 328 so the APM can do what it does best and keep your plane/copter safe while the ArduSlave will sort your pictures. Of course the other boards could do this and much more given the extra channels.
Synthetic aperture radar system (I did say I was working on a different imager).
- Bottom board is the analogue section for it with power regulation, ramp generator and amplication leading to stereo jack to plug into a laptop.
- Top board is the RF section using SMD devices instead of coaxial devices to save space.
- Combined board to show how it fits.
2.4GHz with 240Mhz bandwidth in a FMCW configuration. Record the "audio" and process with MATLAB.
I intend to record the audio onto an SD card after finding a suitable ADC as described in the picture. I already know this as a coaxial system works but as I'm using SMD I've probably borked something plus having a laptop in an Easystar isn't easy so once I've found a good ADC to save the "audio" that will deal with the laptop problem. I'll just need to develop a system on a different frequency so it won't make my plane crash :D
Interested in SAR? Think I'm mental? Want to see why I think its possible? Follow the links HERE, HERE, HERE and HERE.
Here is an edited version of the MITLL IAP course lecture powerpoint (it now includes the bill of materials).
Comments
@Tom No where its my own design. Once I've finished building it (and checked everything fits as I designed it) I will do a full post with everything on it, design plans, critical list of function and building.
Basically its what you could call "skeletal Arducopter" as it has just enough space for everything and the motor arms are 10mm square so any length could be used (I am using 20cm CF square tube although I could half that length)
Ah so yet another connection type in use on APM. The more I delve the more I'm impressed with the Dev Team. Well I hope my OF boards will provide a simple method (code and connections).
DorkBot is looking very good at the moment which is good (OMG how nice is it to have a DRU specially for it).
@ Thanks Mark. We can always dream.
@Randy Very interesting but the 3080 datasheet says it uses 4 wire serial communication which is not I2C which is what I want to get it to function so it works with APM easily.
I have not even read about the 3050 as only 3 sensors were available to me and two of them were in batches of 2000 so I chose the 5050 which means I can just buy 1 (or 4) :D
Design specs:
3080 - 40 inches per second, 6400Hz and 1600cpi. 1.4" x 1.6" PCB $2.6/chip
5050 - 30 inches per second, 4500Hz and 1375cpi. 1" x 1" PCB $1.6/chip
My only real annoyance is digikey being more expensive in the UK in fact given exchange rates its about 1.5x more expensive for me to do it :(
Re the optical flow boards you discuss, the thing that's better about the 3080 is it's resolution. It's 30x30 pixels instead of 19x19 for the 3050.
The nice thing about the 3050 though is that it's 5V instead of 3.3 so you could make the board much simpler and smaller.
By the way - you don't need an atTiny85 for these, you can just use the SPI interface.
Ritchie
You are a busy boy, I will follow your SAR project and Optical Flow too. good stuff...
@Dave Cheers :)
@MarcS For more details you can find copious details on the original creators site as well as build plans, how-tos and MATLAB files. Dr Gregory L Charvat has now hosted two learning courses to do synthetic aperture radar at MIT's Lincolm Laboratory and I have been emailing him over around 6 months getting my design correct. Here are the sites you get the information from MIT OpenCourseWare, MIT Lincoln Laboratory, Dr Charvat's weblog.
Processed image is done through MATLAB and to be honest resolution is not camera resolution :D time is dependent upon size of data but the Go State MATLAB only take a few seconds with my quad core.
As I say my first attempt will do MATLAB with my second reading from the SD card but this requires a serial interfaced ADC of high bandwidth (although audio line are used in the original setup 16-bit 44.1kHz I'm reaching for 24-bit 256kHz as a minimum). Once I have got that working correctly I will need to shift frequency up to be able to shrink the focusing antennas (currently 85mm diameter with 175mm length).
-
1
-
2
of 2 Next