Shop OBEX P1 Docs P2 Docs Learn Events
Problems with Compass Parallax #29323 (HMC6352) — Parallax Forums

Problems with Compass Parallax #29323 (HMC6352)

RAOORAOO Posts: 4
edited 2011-01-22 20:14 in Accessories
I have run several tests on the compass and find that the Northern readings are
not reasonable. I have calibrated 3 times at 3 different locations, and have run
the tests at 5 different locations. As an example of the results, I have placed a
grid with the 120 degree line aligned with the compass reading of 120 degrees.
I have then taken readings at 0, 30, 60, 90, etc. around to 330 degrees. The
results are:
11, 39, 64, 91, 120, 151, 183, 212, 238, 269, 299, 338.
As you can see, the readings around the Northern axis are pretty bad.

So my question is, has anyone used this compass, and if so, what has been your
experiences with it?

Or perhaps I just have a bad module.

Richard

Comments

  • Beau SchwabeBeau Schwabe Posts: 6,566
    edited 2010-05-22 20:17
    RAOO,

    Your numbers seem to be right on target. Are there any differences between when you are calibrating versus running in the field? i.e. unshielded motors running, nearby metallic objects, etc.?

    Your always going to have problems near the North, only because this is where 0 Deg transitions to 360 Deg, unless you figure it out in software. One software method is to offset your values by adding 180 Deg, and then checking to see if THAT result is greater than 360 ... if it is, then you subtract 360. This has an effect of flipping the 0/360 Deg transition to the South instead of the North.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Beau Schwabe

    IC Layout Engineer
    Parallax, Inc.
  • RAOORAOO Posts: 4
    edited 2010-05-22 21:59
    Thanks for the quick reply, Beau.

    Two of the tests were run at different locations inside my home, with one calibration done there.
    Obviously, there are motors there, but none of them were running. One test was run outside
    the home with one calibration done there. Two more tests were run at locations further away
    from the home (like 50 feet), with one calibration done at one of the locations. With each of
    these tests, the results are similar (probably within spec of the chip).

    The transition around the 360/0 area won't be a problem for me since I have to deal with the
    magnetic declination anyway (I need values relative to true North). I am just really bothered
    by the large errors in the Northern region (the readings at 0, 30 and 330 are really out of spec
    and the value at 60 degrees is suspect). It seems to me that Honeywell wouldn't market
    something with those kinds of errors, which makes me wonder about a bad chip. That's why
    I'm interested in feed-back from others who have used this component.

    Richard
  • John AbshierJohn Abshier Posts: 1,116
    edited 2010-05-22 22:10
    Are you sure that the compass is level for all angles?

    John Abshier
  • RAOORAOO Posts: 4
    edited 2010-05-22 22:28
    Good point to be made, John. The compass module is plugged into a
    bread-board, which may not be exactly of constant thickness, and
    the module itself may not be exactly square to the board. The tests
    were run on a table that was leveled (although it's a bubble level) before
    the tests were run. But, taken all together, I think the setup was
    pretty close to level.

    Thanks for the input, I do appreciate your input and your time.

    Richard
  • dreamer123dreamer123 Posts: 1
    edited 2010-05-23 06:19
    I have problems using The same compass..I am unable to recieve any data from it..
    Cud anyone plz tell me the exact instructions to be sent for i2c communication..
    or plz share your program if anyone have used picbasic pro...
    Thanks in advance
  • RAOORAOO Posts: 4
    edited 2010-06-11 21:03
    Parallax was kind enough to provide me with a replacement board. Using that board,
    I ran the same tests as I stated in the original post. To summarize, I expected values of
    0, 30, 60, 90, 120, 150, 180, 210, 240, 270, 300, 330
    with board number 1 I got
    11, 39, 64, 91, 120, 151, 183, 212, 238, 269, 299, 338
    with board number 2 I got
    3, 34, 62, 90, 120, 154, 186, 214, 240, 267, 297, 330

    As you can see, the Northern numbers with the new board are much more reasonable.
    However, the Southern numbers are a little further out than one would hope.

    I should note that the readings received when the compass was aimed at true North
    and various angles off true North, were in keeping with the magnetic declination at
    my location.

    For my application, I intend to take readings at several more compass points and create
    an error correction capability.
  • Beau SchwabeBeau Schwabe Posts: 6,566
    edited 2010-06-12 03:15
    RAOO,

    Well, I was looking at the datasheet from Honeywell and looking at our datasheet, and was really hoping to find a typo. (no such luck) ...and it still could be a typo I guess, but unlikely.

    Looking at all of your numbers with the exception of one of them at 3% they are all within an error of 2.5% (<percent) ... Honeywell's datasheet and our own states that it should be within 2.5 Deg (<Deg) not percent :-(

    So, I'm still not completely convinced. Because a value of 2.5% seems acceptable, while a value of 2.5 Deg is even better ... 2.5 Deg equating to an accuracy of 0.7% which is way better that I would expect.


    Note: The attached file plots the percentage accuracy of your earlier numbers with your latest numbers.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Beau Schwabe

    IC Layout Engineer
    Parallax, Inc.

    Post Edited (Beau Schwabe (Parallax)) : 6/12/2010 3:32:36 AM GMT
    1267 x 683 - 161K
  • mohamedelnonomohamedelnono Posts: 6
    edited 2011-01-18 15:25
    i want to program the compass #29323
    if any one know ho to program it using proton basic please i need help ASAP
  • Mike GreenMike Green Posts: 23,101
    edited 2011-01-22 20:14
    mohamedelnono,
    Please do not post essentially the same message multiple times. It's against forum rules.
Sign In or Register to comment.