accelerometer issues and questions (Parallax/Memsic dev...

Archiver

Thanks to both of you who responded to my questions. I realize the
limitations and nonlunearities above 60 degrees. I undwerstand that the
experiment and code provided in the Parallax/Memsic demo kit takes this
feature into account. The application is for 0-360 degrees range. This is
accomplished by using both axis outputs from the accelerometer. This way (I
assume) when one sensor is least sensitive the other axis unit is at maximum
sensitivity. I am not real clear on the reasons for the ArcTan function
(ATN).
I was hoping Parallax or Memsic or EMEsytems would respond to my post since
the part is made by Memsic, the kit is sold by Parallax, and EMEsytems wrote
the code.
Thanks Again !


[noparse][[/noparse]Non-text portions of this message have been removed]

Archiver

As I often remind list users ... you should download the data sheet for the
Memsic device. The calculation of angle using both axis of the 2125 is comes
directly from Memsic data. It works because the PBASIC ArcTan function
returns a vector angle for a line that extends from 0, 0 to X, Y -- the
Memsic 2125 returns X and Y in g-force (that's why both sensors are
required). Since the sensor returns positive and negative g force values,
all quadrants of a circle can be covered.

To be fair, ArcTan was not previously documented but is included in the new
PBASIC Help File that will be provided with the compiler update that will be
out soon. Since we're discussing it though, ArcTan returns Brads (0 - 255),
not degrees, so a conversion to degrees is required.

-- Jon Williams
-- Parallax


In a message dated 4/24/2003 2:09:54 PM Central Standard Time, azeasi@a...
writes:

> Thanks to both of you who responded to my questions. I realize the
> limitations and nonlunearities above 60 degrees. I undwerstand that the
> experiment and code provided in the Parallax/Memsic demo kit takes this
> feature into account. The application is for 0-360 degrees range. This is
> accomplished by using both axis outputs from the accelerometer. This way (I
>
> assume) when one sensor is least sensitive the other axis unit is at
> maximum
> sensitivity. I am not real clear on the reasons for the ArcTan function
> (ATN).
> I was hoping Parallax or Memsic or EMEsytems would respond to my post since
>
> the part is made by Memsic, the kit is sold by Parallax, and EMEsytems
> wrote
> the code.
> Thanks Again !



[noparse][[/noparse]Non-text portions of this message have been removed]

Archiver

Thanks for your response. I did download the Memsic documentation but am
still a little unclear on the outputs from the accelerometer. One last
question...why the difference in the divisor for the demo kit source code.
One has divide by 8 the other divide by 16. What are the implications from
using 8 or 16? Is the acrtan function available in the Version 1.33 compiler?
Is it different in the version 2 compiler? Thanks again!


[noparse][[/noparse]Non-text portions of this message have been removed]

Archiver

Your question on divisor is not clear to me. I wrote the demo source code
and didn't use 16 as a divisor anywhere. Please double-check to make sure
you're not missing something basic [noparse][[/noparse]no pun intended] and then re-ask your
question.

ATN is available in all versions of the BS2 compiler -- it just wasn't
documented. It takes two inputs: X and Y which are signed bytes (-127 to
+127). The output is the angle in Brad (0 - 255) to the point specified by
X, Y. You can convert Brads to Degrees by multiplying by multiplying Brads
by 1.406. In PBASIC:

degrees = brads */ $0168

or

degrees = brads + (brads ** $6800)


-- Jon Williams
-- Parallax

In a message dated 4/26/2003 10:47:44 AM Central Standard Time,
azeasi@a... writes:

> Thanks for your response. I did download the Memsic documentation but am
> still a little unclear on the outputs from the accelerometer. One last
> question...why the difference in the divisor for the demo kit source code.
> One has divide by 8 the other divide by 16. What are the implications from
> using 8 or 16? Is the acrtan function available in the Version 1.33
> compiler?
> Is it different in the version 2 compiler? Thanks again!



[noparse][[/noparse]Non-text portions of this message have been removed]

Archiver

Thanks again. The fact that ypu did not use 16 as a divisor is exactly my
question! In the documentation posted in the 2003 catalog the formula for
tilt is "tilt=(xmg?16) ATN (ymg/16)" hence a disisor of 16
BUT in the documentation listed with the demo kit that arrived with the
Memsic accelerometers the formula is listed as " tilt= (xmg/8) ATN (ymg/8)"
thus a divisor of 8!
So to post my question again. What is the reason for the different divisors?
What will be the differences in the result using the tilt formula with the
disor of 8 versus the tilt formula with the divisor of 16?
Thanks again!


[noparse][[/noparse]Non-text portions of this message have been removed]

Archiver

Here's the raw truth: The catalog formula was a quickie -- that works -- and
yet a quickie nonetheless. After thinking about it a while and checking the
numbers, we changed to eight to preserve some resolution while bringing the
tilt value into the range required by ATN (-127 to +127).

When in doubt always use the data that comes with the product; that will
likely be the best code we'll have at the moment for a given device.

-- Jon Williams
-- Parallax


In a message dated 4/26/2003 4:08:01 PM Central Standard Time, azeasi@a...
writes:

> Thanks again. The fact that ypu did not use 16 as a divisor is exactly my
> question! In the documentation posted in the 2003 catalog the formula for
> tilt is "tilt=(xmg?16) ATN (ymg/16)" hence a disisor of 16
> BUT in the documentation listed with the demo kit that arrived with the
> Memsic accelerometers the formula is listed as " tilt= (xmg/8) ATN (ymg/8)"
>
> thus a divisor of 8!
> So to post my question again. What is the reason for the different
> divisors?
> What will be the differences in the result using the tilt formula with the
> disor of 8 versus the tilt formula with the divisor of 16?
> Thanks again!



[noparse][[/noparse]Non-text portions of this message have been removed]

Archiver

Jon,

Can you shed more light in this ATN function? It appears that it is
supported by all the BS2 editors for all the BS2 versions, but the
fact that it has been undocumented until now makes me suspicious.

Were there bugs in early implementations? Is that why its existence
was hidden from us for so long? If this is the case, can you tell us
what versions of the Stamp we should refrain from using ATN on?

Or if ATN was not supported on some early BS2s, which ones can we not
use it on?

And if ATN is not supported on certain BS2s, you wouldn't do
something tricky like detecting the firmware version, and emulate the
function in EEPROM, would you?

--- In basicstamps@yahoogroups.com, jonwms@a... wrote:
> As I often remind list users ... you should download the data sheet
for the
> Memsic device. The calculation of angle using both axis of the
2125 is comes
> directly from Memsic data. It works because the PBASIC ArcTan
function
> returns a vector angle for a line that extends from 0, 0 to X, Y --
the
> Memsic 2125 returns X and Y in g-force (that's why both sensors are
> required). Since the sensor returns positive and negative g force
values,
> all quadrants of a circle can be covered.
>
> To be fair, ArcTan was not previously documented but is included in
the new
> PBASIC Help File that will be provided with the compiler update
that will be
> out soon. Since we're discussing it though, ArcTan returns Brads
(0 - 255),
> not degrees, so a conversion to degrees is required.
>
> -- Jon Williams
> -- Parallax
>