How easily can an IMU be damaged by soldering?

Having a hard time getting magnetometers calibrated on 2 of 3 identical 9DOF IMUs.

Both have an X reading that never or barely rises above 0 (almost all X readings are negative). I've no idea if this is common but even after calibration and tilt correction neither yields an evenly distributed azimuth.

It dawned on me that in both cases I'd taken to the part with a soldering iron to change the I2C addresses (bridging 2 pads) and I'm wondering if I might have lingered too long reassuring myself that it looked competent? I've never been good with a soldering iron.

The first item, still at the default address (and never touched by an iron) seems to calibrate pretty well (decent range both sides of 0 on all axes and yields a useful bearing through 360).

Not sure if I should order more of the same (and take more care to be quick with the iron) or just look for a different part. The current part is desirable because it comes with Qwiic connectors (and, I think I said, I'm rubbish at soldering).

Have you run the IC2 address finder sketch to if they respond ?

Hi, not having any problem talking to them, just wondering why on both of the ones I have changed from the default address (with a soldering iron) the X axis reading seems so out of whack (and if this is why the bearing I calculate from them is poorly distributed).

In terms of min/max values (being the basis of calibration), the one that is still at the default address yields;

-03864 [<] +03028
[Y] -04346 [<] +02476
[Z] -03479 [<] +03369

But the two with the alternate address (via soldering) have an almost entirely negative X reading;

[X] -06484 [<] +00269
[Y] -02355 [<] +04489
[Z] -03434 [<] +03257

[X] -07941 [<] -01054
[Y] -01670 [<] +05189
[Z] -03073 [<] +03869

Small sample size, so it might be coincidence but it might also be a pattern that suggests heat from soldering has compromised the sensor. I was hoping somebody could tell me if all-negative values is nothing unusual or if it suggests I've cooked them.

Hi, @ionman
Can you please post a picture of your modules please?

Thanks.. Tom... :smiley: :+1: :coffee: :australia:

More likely to be some component leads are made from plated steel and your soldering iron AC power magnetized them. Your board would have been soldered with heated air.
Paul

1 Like

How would you correct that?

Thanks. My other thought had been static charge, but that sounds very plausible. Is there any way to guard against that in future without having some sort of hot-air station?

EDIT: Nobody seems to speak well of them but maybe a solder pen (just for this particular case)? I guess there's butane or a 5V (USB) DC iron.

I would use a sensitive magnetic compass to see if magnetism can be detected,
Paul

If it is magnetized, what could you do about it? Is there a method of selectively degaussing a tiny pin?
Last time I degaussed anything was to color correct a CRT that my friend's kit touched with a speaker magnet, about 25 years ago.

1 Like

I suppose I could just unplug the iron immediately before applying it to the job, would that work?

You evidently have a very common "hard iron" offset, but have not explained how you calibrated the magnetometer and why calibration can't correct the offset.

Please post calibration plots.

An example of severe hard iron distortion, and how it was dealt with properly and effectively, is explained in this forum post: https://forum.pololu.com/t/correcting-the-balboa-magnetometer/14315

1 Like

Wow, great link, only skimmed but I think that might prove very useful when I get to dig into it later.

As yet I've just been doing the "finding a set of offsets and scale factors, using a simple axial min/max approach" as described in the article. So far I hadn't found an example of ellipse-fitting that would be practical to distribute within my project (end users need to be able do this) - but this might be just the basis.

At the same time, should I take from your response that there's nothing unusual or severe about that one axis being so far 'off center' (and only apparently on the 2 of 3 I've taken an iron to)?

EDIT: Just noticed, you're the author, thanks for sharing, it looks like gold!

Yes. Magnetometers are not very useful "out of the box". At least, they are not useful as electronic compasses without calibration.

Hi,
Which 9dofIMU do you have?
What does the actual module look like?
What IMU chip is it using?
Can you please post a link to where you bought them, or a picture of your module?

Thanks.. Tom... :smiley: :+1: :coffee: :australia:

Adafruit LSM6DS33+LIS3MDL Adafruit LSM6DS33 + LIS3MDL - 9 DoF IMU with Accel / Gyro / Mag [STEMMA QT Qwiic] : ID 4485 : $9.95 : Adafruit Industries, Unique & fun DIY electronics and kits

Hi,
Front and back of device.

How can you damage the IMU?
Did you clean the pads before applying solder?

Tom... :smiley: :+1: :coffee: :australia:

Soldering on an IMU can change the output characteristics of the IMU.

The IMU is sensitive to the surface tension of the soldered connection. Adding/removing connections, more/less solder, changes the surface tension that the gyro and the accelerometers 'feel'.

So the lesson of the day is use pin headers?

Hi,

But @ionman only soldered the address pads on the other side of the PCB.
What happens when they are IR soldered in the first place?

Tom... :smiley: :+1: :coffee: :australia:

Soldering of/on/over/around/between and so on and so forth of an IMU is well documented, do that internet search thingy.