Compass chips

I'm looking at this chip: HMC5983-TR, along with its friends and relations.

Anyone have any mileage with these? Are any ones particularly better or worse? Are any suppliers particularly better or worse?

Mostly, I'm interested in knowing the direction of a rapidly moving platform with respect the Earth's magnetic field, as an aid to moment-to- moment navigation. So both accuracy and high bandwidths are big pluses to me.

--
Tim Wescott 
Wescott Design Services 
 Click to see the full signature
Reply to
Tim Wescott
Loading thread data ...

I recently researched sensors, for a system that would measure ambient fields and apply a correction to an instrument. There was a medium-big thread on that here, a couple months back.

I didn't like any of the Hall or AMR/GMR type chips for my application. They are usually slow, inaccurate, and often weird. I decided to use a fluxgate. Autonnic has some nice stuff, coils and boards. Ditto Speake & Co. Both British.

--
John Larkin         Highland Technology, Inc 

jlarkin att highlandtechnology dott com 
 Click to see the full signature
Reply to
John Larkin

The only one I've used was already in a can, along with MEMS gyros and accelerometers and a GPS receiver, to create an integrated inertial nav box. I don't know which specific magnetic-compass device it used.

From playing with it on the bench, they weren't fooling about not using steel bolts or otherwise having steel, iron, etc close to it. There are calibration procedures that can get around some of the effects of nearby iron, but you have to run them.

In the finished product, I'm pretty sure I ended up ignoring the magnetic heading data, and just using the inertial and GPS stuff. I couldn't get out of having steel, and motors drawing non-trivial current, relatively close to the can.

One other observation is that pretty much every one of these I've ever seen in a complete system, including ones where the user is expected to be clueless (automotive), have some kind of end-user-accessible calibration procedure. Usually you push the button and then rotate the compass sensor through 360 or 720 degrees and that's it.

Depending on where you are on the planet, the Earth's magnetic field can do strange things. For instance, the nominal magnetic declination in Oregon ranges from roughly 16 to 19 East, but there are some local spots where it can be 4 East or 24 East - see

formatting link
. Sometimes these places will also be noted on the VFR aeronautical chart, as in these ones
formatting link
north and south of Harney Lake.

Matt Roberds

Reply to
mroberds

Presuming there are no problems with nearby ferrous objects, you might still have to deal with problems like the one hinted at here...

formatting link

Sensor fusion and estimation is a seriously interesting topic. I don't know how much has been put into the public domain by various amateur organisations that create, for example, DIY drones.

Reply to
Tom Gardner

Ugh. I'd kind of thought about the whole magnetic disturbance thing, but not about magnitudes.

I think maybe I'll just leave the compass out, and let the board get smaller. It's another sensor to fuse in a sensor fusion application, but it's only worthwhile if the data is good.

--
Tim Wescott 
Control system and signal processing consulting 
 Click to see the full signature
Reply to
Tim Wescott

How slow is slow? It needs to be teeny -- a 4mm square chip and its associated traces and bypass caps takes a huge bite out of my board space budget.

I'm thinking maybe it's not worth the effort and board space.

--
Tim Wescott 
Control system and signal processing consulting 
 Click to see the full signature
Reply to
Tim Wescott

That's just a geometric problem due to the fact that left entirely to its own devices a compass needle would really rather point down into the ground in most places on earth. I already know how to deal with that one.

I'm not sure either. It's certainly a complicated one. I've done sensor fusion stuff before: it's not trivial.

--
Tim Wescott 
Control system and signal processing consulting 
 Click to see the full signature
Reply to
Tim Wescott

I'd like about a KHz of bandwidth, and most of the compass-type chips can't do that. But if that's all the space you have for a 3-axis sensor, a fluxgate is out.

--
John Larkin                  Highland Technology Inc 
www.highlandtechnology.com   jlarkin at highlandtechnology dot com    
 Click to see the full signature
Reply to
John Larkin

That's an interesting map. I didn't know there could so much local variation. I guess all that subterranean iron in motion can cause huge magnetic disturbances. Add to that the fact that the magnetic poles wander over time, it's even more complicated.

My personal experience with magnetometers is that we tried to prototype the use of one in a device that included a pair of speakers (think permanent magnet). I can say confidently that such an arrangement is doomed to fail. The market forces of miniaturization make it worst.

JJS

Reply to
John Speth

In the application I was working on, it didn't matter so much, because the device was always used in a defined area that was maybe 100 m on a side, there was plenty of time to do calibration passes, and we almost always had GPS as well. If we could have beat the problems with steel and motor currents, the compass would have been a useful additional reference.

The main things the compass seems to be good for is an absolute heading at zero or low speeds, before you switch to GPS, or an absolute heading if you don't have GPS for some reason.

Matt Roberds

Reply to
mroberds

GPS won't give you your heading - when you are moving it will give you your track. The earth's magnetic field gives you heading, albeit with the significant ambiguities noted by others.

To understand the difference, consider facing north, i.e. the heading is north. Now step sideways to the left while still facing north - the track is to the west.

The threat indicator called FLARM has, necessarily, a significant well understood non-ideality: it indicates the threat relative to your track - which causes pilots to start looking relative to the aircraft's heading, until they consciously correct.

formatting link

Reply to
Tom Gardner

How much is a heading off from a track in an airplane? The difference would only be due to wind?

I used to geocache a lot and tried teaching some friends how to use the GPS. Often they just couldn't get used to the idea that this wasn't a compass and if you turned your body (and the GPS with it) the GPS didn't indicate that. As you say, it only knows which way you have moved and even then only once your movement was significant, 10 or 20 feet to get a decent "track" as you call it. But when I would use the GPS every week I got very used to it.

--

Rick
Reply to
rickman

It can be huge, especially if you're flying a slow plane in a strong cross-wind.

While its possible to fly slightly sideways, the significant error will be the wind (except in helicopters which can fly sideways, backwards, etc).

Hope that helps! Best Regards, Dave

Reply to
Dave Nadler

To meet FLARM's cost target, it was not practical to do a wind calculation (though perhaps an extremely simplistic circling-drift-estimation might have fit, that wouldn't have helped where the problem is more serious in strong winds - typically wave flying).

In the ILEC SN10, I used true ASI (requires pitot-static sensor plus outside air temperature), and GPS track, to back-calculate wind. The cockpit display is nose-up via back-calculated heading, so avoiding the issue you mention with FLARM and making it easy to choose optimal cloud streets with the course.

Hope that helps! Best Regards, Dave (principal designer ILEC SN10, former FLARM developer ;-)

Reply to
Dave Nadler

When landing, you sometimes sideslip to lose height without gaining speed. See a 45degree track/heading difference in

formatting link
lines up at 1:00, then look at the red woolen thread on the cockpit canopy. The sideslip is removed immediately before touchdown.

Apart from that, ...As much as you want, if there is a strong crosswind. Consider a glider flying at 40kt at

5000ft with a 60kt crosswind component! It is most easily visible when landing with a crosswind or when ridge running - where you may have a 30kt perpendicular to a ridge and be travelling at 60kt along the ridge.

Basically yes, but if you really want to get into it you'll need to consider the difference between true north and magnetic north, yes.

All sounds right!

Reply to
Tom Gardner

I'm unconvinced that it would be possible to give a reliably accurate correction for wind, especially when thermalling or where wind direction is highly variable near ridges or mountains.

I'd always prefer something with predictable and trustworthy characteristics - even if those characteristics are in some ways not ideal. FLARM got it right!

Ah, the easy case :)

Good to hear from the horse's mouth! FLARM is a good piece of kit and I wonder if it will ever become mandatory in drones, e.g. to enable "sense and avoid" in Class G airspace

Reply to
Tom Gardner

I've flown "backwards" at -10kt in a fixed wing aircraft - specifically a K8 glider at 2500ft in a 40kt headwind with an airspeed of 35kt.

Reply to
Tom Gardner

Or you're sideslipping, see a 45degree track/heading difference in

formatting link

Ahem. Backwards at -5kt, doh.

Reply to
Tom Gardner

I've done it, though if the wind changes rapidly (when you cross into a different valley) it takes a while to converge.

I think so, but I'm biased ;-)

Well, no, the easy case is , , which is just a vector subtraction. Because of compass issues others mentioned above, I have no heading input. So, I have , , which is the set of solutions differencing a circle and vector.

Apply (poor man's) Kalman filter and presto, wind.

I joined the FLARM team for a while to get FLARM into USA (software for PowerFLARM and legacy FLARM v5). We're now over

800 units deployed in USA and growing well (I think its around 26,000 worldwide).

While FLARM is now mandatory for gliders in some European countries, the FLARM group would prefer to avoid anything mandatory, with the inevitable authorities showing up to help.

Now I gotta find some more interesting projects... See ya, Dave "YO electric"

Reply to
Dave Nadler

I guess the real issue is that I never considered that heading was anything other than the direction you were headed rather than the direction your window was facing. Now I know...

--

Rick
Reply to
rickman

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.