Hi Folks. I just posted this to the [Microtrasat] list, and someone here may know better how to do it.....
Many Thanks Folks - I've changed the title to "[Microtransat] low power compass"
since it has little to do with raspberry pi which consumes too much power for me.
Here is a specification for a low power compass subsystem, and maybe someone out there
could implement it, including the handful of source code lines needed to
read any I2C non-"rs232" serial data from a compass chip, and squirt out
rs232 data (e.g. "H090" at 4800) to the autopilot picaxe.
Maybe, if needed, smoothing logic. Kalman filter based if they really must
Kalman filters ? Ahh - those were the days, in 1970, when implementing
the military radar tracking software to go on board UK Royal Navy Type 21 Frigates.
Our prototype used a kalman filter, but we ended up with simple alpha-beta smoothing.
KISS. Sorry I digress
I'll paste this Microtransat posting onto my picaxe forum thread, since that is probably
where I'm more likely to find someone who actually does it - including
what might be weeks or months of building and testing, rather than
a few minutes talking about it, like I am doing here.
The relevant links are on my
www.gpss.co.uk/rbdesign.htm page.
If someone succeeds, and shares the "shopping list" (< 100 USD total ?)
and picaxe program ( < 20 lines of code ?) then maybe I can use it
in my boat for 2014. It seems unlikely it would be before then.
Here's an outline spec for the compass subsystem:
Preferred compass chip: whatever you like that's suitable. Ideally it does't need a processor.
Preferred processor if needeed: a picaxe - familiar to me, including program loading tools.
Power consumption: < 3mA from 5v
(including chip, picaxe, and any power conditioning if not able to run off 5v)
Output: rs232 text (0v-5v swing) at 4800: magnetic direction in degrees clockwise. e.g. H090 <CR>
Rate: output at 1 Hz (but you may choose to take data faster from chip and smooth it)
Accuracy: better than 5 degrees; target accuracy 1 degree.
Time constant for smoothed output: less than 3 seconds to reach a stable direction.
Any student wanting to take this up, could obviously start with a first issue of software
that would immediately be useful. If they wanted to make it a more ambitious project,
they could easily extend it to provide "true heading" instead of "compass heading"
although that might be difficult without a low power GPS to provide date and location.
Probably better to KISS
Meanwhile, today I expect to receive the sails from Mr Postman,
for Snoopy's new boat that should be starting tests on Bray Lake next week,
prior to his next attempt from Barton-on-Sea - maybe on 1st April !
Robin
www.gpss.co.uk/autop.htm