Last winter there was an interesting thread (and related threads, e.g., http://groups.yahoo.com/group/microlights/message/98346) on the Yahoo Microlight mailing list, discussing whether some relatively inexpensive ‘backup’ device could be fabricated that might give some useful indication of aircraft attitude.

Since then I’ve done a fair amount of research and then built an appropriate piece of hardware (from off-the-shelf stock – I just needed to make four solder joints) – and then wrote some software that runs on that hardware to experiment with the ideas. (And added some others, too.)

It’s very much an ongoing and still-improving project, but having tested it in the air, I think it’s worth sharing with others. A very much first-pass writeup is now on the web, with a few in-flight pictures even, at: http://speleotrove.com/avionics/

All comments, questions, suggestions welcome!

Mike

(Edited to add pic. of gadget in the air.)

Image Attachment : InFlight-2804-s.jpg   183KB (image/pjpeg) This image has been viewed 251 time(s).

Nice one Mike; the bod that invents this deserves a BMAA prize for safety; hope you carry on with this to prevent another tragedy like the one over the channel in murky weather.

I was rather poo-pooed when first raising the possibility of this sort of gadget, based upon a rate gyro, by Jeremy Harris, and was duly daunted as he is an extremely capable all - rounder on this sort of thing.

The principle I was trying to tempt someone smart on electronics to build upon, was the simple fact that when a flexwing is banked, it is turning. Any Rogallo is the same; it is more or less impossible to bank one without initiating a turn.

The follow-on logic was that a rate gyro, normally useless if just bolted on to measure bank as drift would turn the aircraft upside down in a minute or two, was perfectly adequate when measuring rate of turn. Correlate the two and you have a crude but cheap and simple bank angle indicator. You could also utilise it to control the rate of turn to exit a cloud you had just flown into accidentally, which is too damn' easy.

The only contribution to the debate was to ask the question, I know little about the electronics.

Someone has built and sold one for gliders.

Cheers

Kev

Yes, gyros can certainly add to the equation.   One quite nice approach is the Ardinuo collection (I nearly went that route) -- for example the pilot cards for that.  For some details, go to http://www.coolcomponents.co.uk/ and select UAV/Flight.

Mike

Hello Mike

You've achieved this using GPS; reliable enough to use for a get you out of jail, which is perfect; safety Nazis might argue that it is dependent on the GPS signal, but as I've only ever lost this for a few minutes in hundreds of flying hours, this would not be a valid objection to your idea.

As a simple Mech Eng type compared to your high plane of existence, lacking your level of electronic nous, I could only argue for the effort to create such a device against folk who were understandably worried some berk would treat it as a licence to fly in murk, and outside their licence privileges.

Hope Paul Leigh is following this thread, he's also built practical electronics at this level, so we might see something useful for the awful weather conditions we so often see in the UK.

As Chuck Yeager once said, Britain has the worst flying conditions in Europe...

Cheers and admiration

Kev

Post Edited (Andy Fell) : 13/09/2010 14:35:01 (GMT+1)

Post Edited (Andy Fell) : 13/09/2010 17:39:24 (GMT+1)

I'm very skeptical about the idea that one would correct for turbulence by staring at an AI device .. the human brain is in the loop, and just isn't fast enough (about 1/3 second from eye to reaction, at best).  But yes, the 10Hz device would at least let one test the hypothesis (and yes, the numbers change from one message to the next).   Lots of tests and experiments possible.   I'd like to determine, for example, how accurate the relative positions reported are (we know the absolute positions could be wrong by 5 or more metres .. but it could be the change of position reported is much better).

On the smooth-moving ... yes in that sense I do use the 10Hz capability; at each redisplay I use the difference from last reading and the one a second ago.  So the display updates 10 times a second, in effect.  But there's still a lag of about half a second.  (And one could get the smoothing effect by interpolating 1 Hz readings.)

The step-up inefficiency is a mystery -- it should be possible to get at least 75% efficiency, but this one seems to be worse than 50%.  Needs more investigation .. low on the priority list, however, as 7 hours is a lot longer than I ever expect to be in the air, and carrying a spare AA battery is not a huge weight burden ... :-) 

Mike

Post Edited (Andy Fell) : 13/09/2010 20:53:46 (GMT+1)

The load factor (total acceleration), lf = sqrt(g² + ca²) = 1 / cos(b)
(so the bank angle could in theory be calculated from accelerometer data, though not its
direction – except perhaps by dead reckoning from initial uncoordinated movements)

I have added that back in.

I did test exactly this -- unfortunately the absolute value of the load factor vector seems to be all over the place when calculated from the accelerometer values. I set up a 'hold' button to test this -- so I could freeze the accelerometer data in the middle of a steady turn, then record and study it later. The calculated load factor was never as much as it was expected to be, and in several cases was less than 1g, even in a fairly steep turn (45').

This might be related to the effect I noted on the website (aeroplane seems to turn faster than the bank angle would suggest) -- but in any case, I don't think using accelerometer data alone is a good idea. GPS has the huge advantage of being an externally referenced system, so no drift problems. [And yes the GPS system could fail or be jammed ...]

Also relevant is that there is almost no documentation on the algorithms and filtering applied within the accelerometer MEMS chip. It is entirely possible (and I suspect it might be so) that in some circumstances it carefully normalises the 3 axes to give a total acceleration of 1g. (But then why does it offer 2g and 6g scales?)

I'd be very interested if other people could measure their radius of turn (e.g., from GPS ground track) for a given bank angle in the turn (carry a protractor -- estimation is hard).

Mike  (edited for formatting)

Post Edited (Mike Cowlishaw) : 14/09/2010 15:28:16 (GMT+1)