Help with 5883L crazyness.

Hi Joey,

This may be off base but if I treat your X and Y data as vectors and calculate Result = Sqrt( X^2 + Y^2) I get a Result that is at least in the ball park for the headings that you listed.  Like I said I have no experience with these devices but it would make sense that they would give output in a vector format.

John

Ok now you can see I really don't grasp it quickly as that is exactly what you were doing. No wonder my calculations matched yours. I will put my thinking cap back on and let you know if something better pops out.

John

Is there a correction factor necessary for the angle of the Earth's magnetic field in your area?

The Data Sheet mentions a calibration factor that can be calculated using the self test.

HMC5883L 18 www.honeywell.com SCALE FACTOR CALIBRATION

Using the self test method described above, the user can scale sensors’ sensitivity to match each other. Since placing device in positive bias mode (or alternatively negative bias mode) applies a known artificial field on all three axes, the resulting ADC measurements in data output registers can be used to scale the sensors. For example, if the expected self test value for X-axis is 766 and the actual value is 750 then a scale factor of (766/750) should be multiplied to all future readings of X-axis. Doing so for all three axes will ensure their sensitivity are well matched,

The built-in self test can also be used to periodically compensate the scaling errors due to temperature variations. A compensation factor can be found by comparing the self test outputs with the ones obtained at a known temperature. For example, if the self test output is 750 at room temperature and 700 at the current temperature then a compensation factor of (750/700) should be applied to all current magnetic readings. A temperature sensor is not required using this method.

John

Hi John, the magnetic declination in my area is HERE:

http://www.magnetic-declination.com/

2 degrees 46 minutes

I am basically using the code form here:

https://learn.adafruit.com/adafruit-hmc5883l-breakout-triple-axis-magnetometer-compass-sensor/wiring-and-test

The only difference is I am not using Arduino, but I confirmed the X,Y and Z values do match.  I am going to check to trig function though to ensure they match.  I am using ATAN2.

Basically its:

heading = atan2(fy,fx)

float declinationAngle = 0.22;  //I need to put 2.5 here but it wont make a difference.
        heading += declinationAngle;
        if(heading < 0)
        heading += 2*PI;
        if(heading > 2*PI)
        heading -= 2*PI;

currentHeadingDegrees = heading * 180/M_PI;

M_PI Is built in and defined as 3.14159265358979323846

I'll check out the calibration, that may explain it.  I ordered another one of these since I already have 2 and one is completely junk.  So maybe 3rd time a charm?  Different brand though, same chip.

The good news is that the chip is always the same, gives the same readings.  It just doesn't scale linear to the actual heading for an adjustment.

I am tempted to erase my first two replies but I have learned from past experience that even bad advice can sometimes trigger an insight. If I take the dot product of the values of your test I get the following numbers:

North:   552.20

East      564.02

South   612.94

West    612.37

If the magnetometer is measuring the strength of the field it is expected that we would have the same number with the dot product of the vectors regardless of the orientation of the magnetometer.  These numbers are close particularly the South and West. I assumed that your angles were precisely 90 degrees as the numbers were crunched using the easy Pythagorean Equation. This would seem to indicate that the magnetometer is working correctly and it is a mathematical translation problem. I am guessing that the N and E were not precisely at right angles and so my calculations would have been off for them.

Sorry for the inundation of ideas but I use this as a problem solving method.

John

Here is the last thought before I quit for the night. Does it make a difference for your equations which axis the chip is rotated around. From your readings it seems to me that the field is practically perpendicular to the Z sensor and that you rotated around the axis of the Z sensor. Does that make any sense?

John

Hi Joey,

This is how I'd try to troubleshoot this - unfortunately I've not used this sensor before, but there is a typical process I would use for sensors:

Basically, try to find known values (e.g. a worked example in the datasheet), and then plug those values into your code so that the data from the sensor

is overwritten with your known values, so you can see if the rest of the coded algorithm behaves as expected. I do it something like this:

#define ALG_TEST 0
... code...
i2c_read(buf); // read into buffer the actual sensor data
if (ALG_TEST)
  {
   buf[0]=0x11;
    buf[2]=0x22;
    buf[3]=0x33;
  }
... data processing ..
... remainder of your code ..

Basically at any point during your code development you can set ALG_TEST to 1 to quickly confirm the algorithm is not broken, and set it back to 0 to go back to normal operation.

The problem is finding some known values and the correct computed value. To do this, perhaps use this arduino compass web page which (about half-way down)

has a screenshot of captured raw values, and the computed results. You could plug the raw values into your code, and check if the result matches.

Another way I frequently use is to come up with some raw values and the expected result by checking if the datasheet has a worked example (in your case unfortunately

the datasheet doesn't have this) or by inventing some sensor values, and writing some formula to generate the output, using (say) Excel.

Then, at any point during testing, I can plug the raw sensor values into the spreadsheet and see what value I expect. This really helps to confirm the coded algorithm is

working well. I do this a lot with virtually any IC. For example, here is my 555 spreadsheet:

I can plug in values into the green cells, and the red cells show the expected output for the 555. I'd suggest a similar thing

with the IC you're using. It takes a while to create, but worth it in the long run.

Also, with the two modules you have, do they both give similar values that match each other?

How are they connected to the RPI - short(ish) wires? What resistors are you using for pull-ups for the I2C -

are these already on the board, or external? Could you attach a photo of the set-up if possible?

Also, if you can post the code then I can give it a visual inspection, in case anything seems odd.

Go to https://github.com/screamingtiger/QuadCOP

And look at head.h and heading.cpp

I think I have missed a stip.  there is some code that is setting up gains

I'm not using these but I did think I confirmed I am getting same values as the arduino I was using.

Anyways have a look at the code if you will.  Its pretty simple actually.

The god news is that I can still have a test auto flight today.  Because my magnometer is consistent, I can use it to fly in a square pattern.  I think that will be a good test.  But would love to have this working.