13 replies to this topic

[Mr.RC-Cam]

Came across a new tilt compensated compass today. $249. Details: http://www.oceanserv...om/compass.html

When something like it creeps below $150 I believe there will be plenty of us ready to jump on it!

[Blue Sky]

Hi!

THese use mems type accelerometers for gravitational tilt detection.
They may still have a problem when subjected to a 3d accelerated frame of refrence.

-Dave

[Mr.RC-Cam]

Actually, I was thinking of VR head mounted pan & tilt rather than model mounted sensor apps.

[twinturbostang]

Perhaps he means if you whip your head around (or actually in tilt), the sensor may overshoot some due to the added g forces. In theory though, it could still work. As long as you make nice smooth motions, it should be fine. And even if you move rapidly, the sensor should sense the correct angle after the acceleration has stopped. Which means it might overshoot some, but would then settle to the correct angle.

[Blue Sky]

Actually, I was thinking of VR head mounted pan & tilt rather than model mounted sensor apps.

Ah so.

These seem to be intended for marine and automotive use.
They might still work O.K. on planes that are stable.

To my mind people have been looking at things backward,
or should I say at right angles, as far using geomagnetic fields
for stablizing a plane are concerned. It seems to me we should be using the geomagnetic field to determine inclination or tilt so as to provide a fixed frame of refrence for everything else.
Where I live the strongest component of the earth' magnetic field is vertical. Inclination is actualy easier to sense than compass heading.
I realize that the inclination should still be combined with other sensor readings, I just think that inclination should be first order rather than compass heading.
I'm a hobbiest and not an engineer, I'm sure this must have been thought through before by people with a better grasp of things than I.
I think I'll play around with it anyway.
I'd really like a geomagnetic replacement for the Copilot.

-Dave

[Pepelatz]

Came across a new tilt compensated compass today. $249. Details: http://www.oceanserv...om/compass.html

When something like it creeps below $150 I believe there will be plenty of us ready to jump on it!

CMPS03 - Robot Compass Module $50

[Mr.RC-Cam]

If only the $50 compass was tilt compensated. Then life would be good.

[twinturbostang]

If only the $50 compass was tilt compensated. Then life would be good.

What does that mean for us? If it's not "tilt compensated", does that mean there are errors in the measured heading as the unit is tilted from level?

[Mr.RC-Cam]

Yes, that is correct. It does not take much tilt to cause substantial errors in compass data; Tilt compensation prevents that. Some compass modules do the correction mechanically (with limited range), but the modern solution is with complex software and additional sensors. It is expensive to do, but the prices slowly are coming down thanks to MEMs technology.

[Mr.RC-Cam]

Old discussion, but still active in my mind.

I've been patiently waiting for a big price drop on these things, but so far that hasn't happened. There is hope since components like the Honeywell HMC6343 are making little steps towards lower cost. Sparkfun and other vendors sell it for $150/qty-1, so perhaps we will see something under $100 one day.

[pseddon]

65Euros http://www.mikrokopt...cwiki/en/MK3Mag

regards Peter

Old discussion, but still active in my mind.

I've been patiently waiting for a big price drop on these things, but so far that hasn't happened. There is hope since components like the Honeywell HMC6343 are making little steps towards lower cost. Sparkfun and other vendors sell it for $150/qty-1, so perhaps we will see something under $100 one day.

[Mr.RC-Cam]

Thanks for the link. I'm not sure how tilt compensated that one is since the specs are a bit sparse.

The Honeywell is rated for ±60 degrees, which is fine. There's a comment in the MK3Mag wiki text that has me concerned: "Special attention should be taken to ensure that the MK3Mag is Level (i.e.: parallel towards the FC). An angle of 5° for example would have the effect of an error of >5°."

The idea behind tilt compensation is to decouple the sensor's angular tilt from spoiling the compass heading data. I've tried a couple tilt compensated compass solutions and found them to be poor performers. So, apparently this is a hard pony trick to do if best performance is needed.

[Kilrah]

There's a comment in the MK3Mag wiki text that has me concerned: "Special attention should be taken to ensure that the MK3Mag is Level (i.e.: parallel towards the FC). An angle of 5° for example would have the effect of an error of >5°."

This comment is related to the fact that on a common MK setup the MK3Mag does NOT have its own accelerometer, but uses the one on the flight controller as it's already available and thus saves the cost of a chip. So it's normal to require proper alignment between the board holding the accelerometer and the separate one holding the magnetometers, so that the sensors are physically aligned.

The MK3Mag however does have a footprint to hold the accelerometer onboard should one want to use it as an independent unit. I don't think code has been made to use it yet though, so it would require some fiddling with the open source code to integrate the flight controller's accelerometer handling code into the MK3Mag's firmware. Everything is already there, just adaptation is needed.
The results and precision of the standard setup are amazing. I did a few tests on the ground turning the thing around while looking at the heading value, it's rock solid whatever the tilt is, and totally repeatable. Now what I haven't done is look at the code to see if the info from the gyros was mixed in to calculate the heading, in which case performance with the stand-alone unit would be a bit lower. But at least in the MK it's very good.

For the record as I see the beginning of the thread and the Ocenaserver compass... I actually bought one of those last year from Sparkfun, and couldn't get any data out of it. I sent it for exchange, received a replacement (about 1 month round trip), and it was just the same. It's still somewhare in a box around here, I spent hours on it but never even got a single byte of "I'm alive" data on its output...

Edited by Kilrah, 23 January 2009 - 03:41 AM.

[Mr.RC-Cam]

The MK3Mag however does have a footprint to hold the accelerometer on-board should one want to use it as an independent unit. I don't think code has been made to use it yet though...

What I want is a fully independent compass module that is ready to integrate. Something with a I2C port and $50 price tag would be nice.

Now what I haven't done is look at the code to see if the info from the gyros was mixed in to calculate the heading

I would expect that they are including the gyro in the compass function. It would make a lot of sense to do that.

For the record as I see the beginning of the thread and the Ocenaserver compass... I actually bought one of those last year from Sparkfun, and couldn't get any data out of it.

That is too bad. It would have been nice to hear what you thought of its raw compass data. I hear the Honeywell HMC6343 needs a lot of data averaging to settle it down, but apparently the tilt compensation works well.