TMAG5170-Q1: Offset drift following exposure to neodymium magnets

Clement,

Thank you for reaching out on E2E.  I have a few follow up questions for you.

1. Can you provide any more details regarding the strength of magnet you were testing with? Specifically, what magnitude field was the sensor exposed to?  What was the physical geometry of the magnet you used?
2. Are you able to quantify the magnitude of the offset change you observed?
3. Also, can you please confirm that there are no nearby ferromagnetic materials which may have been inadvertently magnetized during the test?

Thanks,

Scott

Hi Scott,

I was testing the TMAG5170 with two stacked magnets reference 9161 from Radial Magnets. These are flat cylindrical magnets that I approached at 2 cm of the sensor. Through simulation, I found that the maximum field of the magnet at this distance is 0.5 Tesla.

The offset change that I observed on Y axis is of -15 uT and on Z axis is of -24 uT.

The magnetometer was placed in an empty room on the floor. The sensor was taped on a plastic piece and connected to an Arduino Nano 33 IoT placed at a few centimeters. The Arduino board may contain ferromagnetic materials but it did not aimed the Arduino nano when I exposed the magnetometer.

I hope I have answered you accurately enough.

Thank You,

Clement

Clement,

I understand.  For your result I assume you took the average of several samples?  The change you have reported is in line with the RMS noise for this device when set to the highest rate of oversampling.

I attempted to run a test on my EVM using a 1"x1/2" diametric N42 magnet, and placed the magnet as close to the PCB as I could by hand without touching (to avoid changing alignment to Earth's magnetic field : 25- 65 uT) .  Approaching the sensor from all sides resulted with variation < +/-5 uT.  This is visible when looking at a running average of 500 samples.  The best points for comparison using this method are near the mid-points of each sample area (to avoid including data from when the magnet was closeby)

Z- Axis

Y-Axis

Can you confirm also that the PCB was not disturbed during the test?

Hi Scott,

Yes, to see if an offset drift appears after an exposure to my magnets, I took the average of 9999 data for each record. I compared the average of raw values and I converted it into uT. The magnetometer is set to 32x - 0.4Ksps (3-axes) sampling mode. I set the range to ±25 mT.

I observed a drift on the two magnetometers that I tested. Both were stable over the average of the pre-exposure records (during a week). 

I can also confirm that the PCB was not disturbed during the test.

To give you more information, I put on the following link the CSV files of my tests. I also took a picture of the placement of magnetometers.

https://fromsmash.com/details-TMAG5170-test-w-N52-magnets

If you would like me to send you all the records, please do not hesitate to ask me.

Thank you,

Clément

Clement,

It looks like I'm not able to pull the data off this file sharing service.  In your tests did you apply the field only once or multiple times?  If done more than once, did the result change a second time?

Thanks,

Scott

Hi Scott,

I put the files on another file sharing service : https://filesender.renater.fr/?s=download&token=f6d56bf1-2e34-4794-b0cc-ef2a8275e061.

In my tests, I have applied several times the field. It is at the time of the first exposure that I observe an important drift. On the following exposures, the averages are less stable from a record to another.

This is a plot of Z axis averages. In phase 1 of my experience, the magnetometer is not exposed to magnets. The record following the one with the red dot is the first one after an exposure. In phase 2, there is a succession of before and after exposure records.

And this is the same plot for Y axis :

If you want I can send you all my datas and vizualisations.

Thanks,

Clément

Celement,

Thank you for sharing these plots.  I am not able to use external file sharing services for security reasons.  I see the shift you have indicated, and am reaching out to our validation team to find out if they have seen anything similar during their testing. 

To be sure I understand, at the first exposure to a strong magnetic field (>500 mT) this shift occurs, but upon subsequent exposures there is not as dramatic of an effect? On these devices are you able to apply a similar field with opposite polarity and confirm whether there is any hysteresis in this response?

Thanks,

Scott

Hi Scott,

I applied the same field with opposite polarity and this has almost cancelled the offset drift.

The cause may be the ferromagnetic materials contained in the Arduino board. I will try next week to move the magnetometer away from this board.

Did your validation team find anything similar ?

Thanks,

Clément

Clement,

The validation team did not locate any hard data indicating the result of the early tests they did similar to yours.  Please keep us updated on the result of your test moving the magnetometer.  In the mean time, I will also attempt to capture a larger data set of my own with the evaluation module.

Thanks,

Scott

Hi Scott,

Thank you for your feedback from the validation team. I have remade my test setup as I planned before. The arduino board is far away from the magnetometer and I was careful to make sure that there were no materials near the TMAG5170.

I will take measurements over several days and I will keep you informed of my results.

Thanks,

Clément