HMC6343 - "high end" electronic compass module. Any good ?

[Jon Russell]

Hi,

SparkFun are selling the Honeywell HMC6343 magnetometer.

It's expensive (about $60 on DigiKey). But the SparkFun price is even worse ... $150 on a breakout board.

According to SparkFun's blurb its meant to be high accuracy, and more immune to interference (both mechanical and magnetic)

It has a built in uP that does the data fusion for the tilt compensation and calibration, so it just outputs the heading, and does all the maths for you.

https://www.sparkfun.com/products/12916

https://cdn.sparkfun.com/datasheets/Sensors/Magneto/HMC6343.pdf

Has anyone used one of these ?

Are they any better than a bog standard magnetometer ?

Worth the extra cash ?

Thanks.

Regards,

Jon.

 

[sk...@fll-freak.com]

So I have not used this specific high end compass, but I do have experience with other ones. In fact at $150, this would be  dirt cheap compared to what I am using at work. My units are rated for +- 0.1 degree accuracy. Sounds great till you read the fine print, ... they still only work in a benign magnetic environment.

The data I posted a while ago of magnetometer data over a man hole cover is just as much an issue for a $10, $150, or $2K unit. The only advantage I see is that the math is done for you. But that in itself may be a problem in that you are now removed from the inner workings of the device. If you don't like their math, your toast.

A few interesting issues:
1) Unit using magnetometer started to report headings with over one degree error. Problem traced back to the metal closures of the user's undergarment.
2) Unit would not calibrate to better than 1.5 degrees. Problem traced to an inadvertent replacement of a #2 screw 2 inches away with a low grade stainless version.
3) Little help from manufacturer when having problems. Always comes down to their secret sauce in the unit's firmware.

Here is one problem I do not think throwing money at will solve. Your $20 magnetometer will be just as useful as the $150 unit (if you can do the math) in the harsh magnetic environment of a traditional rover.

[Ted Meyers]

Well, I do not have Skye's level of experience, but I have spent a lot of time fiddling with magnetometers on rovers.  I have to say that I totally agree; there is no way that a more expensive, more accurate, and generally better sensor will help.  There are just too many variables.  I find that even a small change on the rover can throw off the calibration, so I have to figure that any iron in the local environment will mess with you no matter what sensor you use.

--
You received this message because you are subscribed to the Google Groups "diyrovers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to diyrovers+...@googlegroups.com.
To post to this group, send email to diyr...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/diyrovers/7111729b-c427-436a-a8e4-f6ff1598cdf5%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Thomas Roell]

A naive question on my side. There are magnetometers with 12 bit ADCs (HMC5883L), and newer ones with 14 bit ADCs (the in in the MPU9250). Given the massive inaccuracies, and the measurement noise, does this difference of 2 bits actually make a difference in real life ?

- Thomas

[Skye Sweeney]

If you are assuming a rover platform in a harsh magnetic enviroment,
then the extra bits will not likely buy you anything. You would end up
with just more precise garbage answers.

Now if your sensor is on top of a mountain in Greenland (do they have
mountains?) away from anything that would corrupt the magnetic field,
those extra bits might be very nice to have.

I subscribe to the space weather emails. I get a warning for things
like solar flares and other phenomena that would affect objects in
space and communications on the surface. One thing they warn about are
sudden magnetic anomalies where the magnetic field takes a short jog
away from baseline then back again. The variations are small enough I
have never seen them with a top of the line sensor in an urban
environment. But I suspect you might with a 14 bit sensor in a rural
area.