Magnetometer vs. Magnetic Rotary Sensor

[r_main]

I need a magnetic rotary sensor, similar to the Melexis MLX90316.

However, I'd like a low power 3.3V sensor with SPI inteface, and minimum 0.1degree accuracy.

Would the Freescale MAG3110 work in such a configuration?

I suspect that the gain in the analog portion of the MAG3110 would be so high that it would saturate with any significant magnetic field (i.e., adjacent to the magnet on the shaft to be sensed).

Is there some other way to sense angular location without contact? It must also be mechanically robust.

Thanks,

Robb.

[reyes]

Hi Robb,

The Freescale Magnetometer MAG3110 is not a dedicated Rotary Position Sensor has the Melexis MLX90316, but it could do the work, and even with more precision using an accelerometer in addition to the magnetometer, however, the MAG3110 doesn’t have SPI communications interface (only use I2C).

Another solution that we can offer that can suit with your application is the Tilt measuring using a 3 axis accelerometer, you can see how to do this in the following application note: http://cache.freescale.com/files/sensors/doc/app_note/AN3461.pdf

However, we currently don’t have any accelerometer with SPI interface recommended for new designs.

Regards,

Jose

[r_main]

Thanks for the reply, Jose.

Actually, we will be using a Freescale MMA8451Q for tilt sensing in this design.

The magnetic rotary sensor is to sense rotation of a shaft (so tilt sensing is not an appropriate solution).

My concern with attempting to use the MAG3110 is that the amplifiers downstream of the hall sensors would simply saturate in the strong magnetic field of this application. Are you suggesting this will not happen? Do you have experience with the MAG3110 in this kind of application to know this is not the case?

I'm not really worried about using I2C in the design - I prefer SPI for the speed (I2C can be soooo slow, and this is a battery operated device).

Thanks again.

Robb.

[michaelestanley · Accepted solution]

Gentlemen,

FYI, the FXOS8700CQ includes both mag AND accel and I believe it is pin compatible with the MMA8451Q.  But you DO need to pay attention to your max magnetic field as mentioned in the earlier post.  The sensor has a max value of +/- 1200 microTeslas.

Mike

[r_main]

Thanks for your input, Mike.

As I stated to Jose, this design has evolved away from needing a magnetic angular sensor.

However, I noticed an earlier post (I believe it was you) indicated that the MAG3110 also has some nasty temperature drift that should be periodically compensated for. I had not noticed this in the MLX part - is this issue something common to this technology? If we encounter this type of need again, should I be concerned with stability over temperature regardless of whose part I'm looking at?

Cheers,

Robb.

[michaelestanley]

Robb,

It's certainly something you should check out.  I am usually not too concerned about temperature effects whenever our hard/soft iron compensation routines are used.  Those calculations happen multiple times a second, and any  temperature scale/offset issues should get corrected as a side benefit of those calculations.

Mike

[reyes]

Hi Robb,

Firstly, I believe that it’s important to remark that the MAG3110 is not a hall sensor. 

And also I’m thinking that you should be concerned about saturating the full scale range (FSR) of the device because it is not a very big scale (+/-1000uT).

It is hard to answer your question regarding this because we have no idea how strong the fields get in your application. I am seeing that typical magnetic field strength for Melexis part is 50mT which is 50x the FSR of MAG3110. So, according to my point of view according to the information you have sent, I believe that saturation is possible, but you may be able to get around this by using a weaker magnet or moving it away from the sensor.

Once the saturation problem is solved, in theory, it is possible to calculate the magnetic vector that is coplanar to the sensor top surface using only the x and y axes of MAG3110. If you want less 1 deg of accuracy you will probably need to do perform the hard/soft iron calibration. And even using this calibration I cannot warranty that you can achieve this since this will depend on many other different factors.  

You can probably want to take a look at our FXOS8700 instead. It has slightly higher FSR and better temperature performance: http://cache.freescale.com/files/sensors/doc/data_sheet/FXOS8700CQ.pdf

[r_main]

Thanks for the response, Jose.

As we've worked through this design, we've evolved away from using a magnetic sensor for angular position. I understand your suggestion regarding magnetic field strength management. In this application, the slightly higher range of the FXOS8700 would likely not be sufficient to warrant the extra cost. We do need an accelerometer, but for various reasons it would inappropriate to locate the accelerometer where the magnetometer must be located.

Thanks for your support & interest.

Cheers,

Robb.