First step towards cheapo IMU.

[billyzelsnack]

Let us know how well it works.

[billyzelsnack]

So I was thinking.. Maybe putting a magnetometer in a non 100% plastic robot is a bad plan? Seems like something so sensitive really is not going to like all this bits of metal constantly moving about in such close proximity. Though maybe it's insignificant and the real issue is a large masses of of metal ( desks and such ) near by bending the field lines. Anyone with experience here?

[JonHylands]

Well, I got my HMC6343 yesterday, and it took about an hour to write the code to interface to it, which worked the first time. Of course, the I2C code I ended up using had a couple nice examples, so it was easy to adapt them.

So it works really well - its very responsive, and will update its internal sensor values at either 1 Hz, 5 Hz (default), or 10 Hz.

I'm impressed - the hardest part of the whole thing was building the cable, with some built-in pullup resistors.

I used one of my General I/O boards, which has pin sockets for the two pins (C4 and C5) that are used for I2C, with a 3.3 volt regulator and a hacked in serial port. I grabbed some I2C interface code from here:

http://www.uoxray.uoregon.edu/orangutan/i2c.zip

which worked perfectly.

My code looked like this:

Code: Select all

      i2c_start_wait (COMPASS_ADDRESS_WRITE);

      i2c_write (COMPASS_HEADING_COMMAND);
      i2c_rep_start (COMPASS_ADDRESS_READ);
      heading = (i2c_readAck () * 256) + i2c_readAck ();
      pitch = (i2c_readAck () * 256) + i2c_readAck ();
      roll = (i2c_readAck () * 256) + i2c_readNak ();
      i2c_stop ();

      printf ("Heading: %4d   Pitch: %4d   Roll: %4d\n", heading, pitch, roll);


- Jon

[billyzelsnack]

Is the data consistently nice and clean? What about when near metal objects? What about near an operating Bioloid?

Looks like it gives you a heading/pitch/roll. So that's an actual absolute 3-axis orientation ( minus gimble lock? )? I don't fully understand the implication of adding in the magnetometer. What's your understanding of how it pairs with an accelerometer?

[billyzelsnack]

So I went and did a little more research.

Looks like you don't get an 3 axis absolute orientation. It gives you basically the same data as an accelerometer except its data is not going to be corrupted by translations. However its data can be corrupted by nearby metal objects.

[JonHylands]

This is an example data set, with the compass board held down in one place:

Code: Select all

Heading: 286.9   Pitch:   2.1   Roll:    0.3
Heading: 287.6   Pitch:   2.2   Roll:    0.2
Heading: 287.4   Pitch:   2.2   Roll:    0.5
Heading: 286.8   Pitch:   2.1   Roll:    0.3
Heading: 285.9   Pitch:   1.8   Roll:    0.4
Heading: 287.0   Pitch:   2.1   Roll:    0.3
Heading: 287.6   Pitch:   2.3   Roll:    0.3
Heading: 286.7   Pitch:   2.1   Roll:    0.4
Heading: 287.1   Pitch:   2.1   Roll:    0.4
Heading: 287.3   Pitch:   2.2   Roll:    0.4
Heading: 287.3   Pitch:   2.2   Roll:    0.2
Heading: 286.0   Pitch:   1.8   Roll:    0.3
Heading: 286.6   Pitch:   1.9   Roll:    0.3
Heading: 286.9   Pitch:   2.0   Roll:    0.3
Heading: 287.3   Pitch:   2.2   Roll:    0.2
Heading: 286.4   Pitch:   1.8   Roll:    0.5
Heading: 287.4   Pitch:   2.3   Roll:    0.3
Heading: 286.1   Pitch:   2.0   Roll:    0.3
Heading: 286.0   Pitch:   2.0   Roll:    0.0
Heading: 286.6   Pitch:   2.0   Roll:    0.2
Heading: 286.4   Pitch:   2.0   Roll:    0.3
Heading: 286.1   Pitch:   1.9   Roll:    0.3
Heading: 287.0   Pitch:   2.0   Roll:    0.3


The only point of having the accelerometer is to allow the actual level orientation to be extracted from the 3-axis magnetometer. Of course, if the compass is being bumped around a lot, you won't get a stable heading.

I'll try and do a test with it mounted on MicroRaptor soon...

- Jon

[billyzelsnack]

I was messing around with getting data out of heli gyros again tonight. I came across this post..

http://www.arduino.cc/cgi-bin/yabb2/YaB ... 926139/1#1

Grabbing data this way would be nice because the load on the uc is not much so it scales better for sampling multiple gyros.

I gave it a go and it didn't work out so well. I would get many different values and they may or may not of been changing in response to mucking the gyros. My electronics skill is zero so this probably is way off, but would I need to be sampling this at a specific rate? I guess I don't understand how it magically turns the digital PWM into a single analog value.

[billyzelsnack]

haha..

Look at the author..

http://www.nabble.com/Re%3A-How-to-read ... 61252.html

[billyzelsnack]

More infos on converting PWM to analog.

http://www.ontrak.net/pwm.htm

[billyzelsnack]

More much infos and a much more elaborate method..

http://homepages.which.net/~paul.hills/ ... coder.html

[JonHylands]

haha..

Look at the author..

http://www.nabble.com/Re%3A-How-to-read ... 61252.html

Dave is my brother, in case you were wondering...

- Jon

[billyzelsnack]

Yeah I know. I just thought it was funny that I go searching for some random thing to post in a thread that another Hylands would most likely read.

[billyzelsnack]

PS3 gamepads have a 3 axis accelerometer ( I knew that ) AND a gyro! Combined with bluetooth this might be a pretty easy ( almost ) solution.

IF YOU CAN GET IT TO WORK. haha.

I tried for a couple of hours and could not get any of the various Windows drivers found via google searching to work.

The SIXAXIS controller is being discontinued so pretty soon it should be possible to find these suckers for cheap.



http://www.kako.com/neta/2006-018/2006-018.html
Could the gyro be the one with the connector!!!! Man. If so I am going to start buying broken controllers on ebay.

[billyzelsnack]

Ok this is funny..

http://www.murata.com/catalog/s42e.pdf

Apparently that's the gyro in the SIXAXIS. Which I think is the same gyro inside the RC heli gyro I've been messing with. Which also is the same gyro that's in lower end heli 4-in-1 controllers. So I have literally 8 of these things laying around in my workshop. haha.

So after looking at that spec and then looking at the actual gyro on the RC gyro board.. 4 pins!!! How did I not notice this before. I can easily just unsolder it and not mess around with reading the PWM! I guess it is nice to be able to read it because that makes it plug and play with my high-end heli gyros when I end up daring to test them.

One other thing.. I also saw in my readings ( and mentioned in that pdf above ) that you can't put multiple gyros in the same circuit. I guess they mess with each other. Hopefully that's just power and they can share the same ground.

[billyzelsnack]

I'll try and do a test with it mounted on MicroRaptor soon...

- Jon

Did you do the test yet? I'm really interested in how much interference a bioloid based robot will create. I'm thinking of ordering one of these..

http://www.pnicorp.com/products/all/sen-xy

and want to know if I should even bother if the interfence is going to make the data useless.