Hello,

Do you mean this one:
https://www.embedded.com/design/other/4216719/Performing-efficient-arctangent-approximation
Used in this post:
https://forum.allaboutcircuits.com/threads/gps-nmea-antenna-aiming-tracker.96520/page-6#post-728060

Bertus

 

Hi,
I found this. Any good?
C

 

Hello,

Do you mean this one:
https://www.embedded.com/design/other/4216719/Performing-efficient-arctangent-approximation
Used in this post:
https://forum.allaboutcircuits.com/threads/gps-nmea-antenna-aiming-tracker.96520/page-6#post-728060

Bertus

It is the page. Thanks
This is the better equation. It needs only to check the signs of x, y in four quadrants.

atan_approx(x) = (Pi/2)*(b*x + x*x)/(1 + 2*b*x + x*x) where b = 0.596227, with a maximum approximation error of 0.1620º

 

It is the page. Thanks
This is the better equation. It needs only to check the signs of x, y in four quadrants.

atan_approx(x) = (Pi/2)*(b*x + x*x)/(1 + 2*b*x + x*x) where b = 0.596227, with a maximum approximation error of 0.1620º

Well done B.
Hi J,
Is this the total equation?
Is it correct that Pi and b need to be set up as CONSTANTS?

E,
Here: https://forum.allaboutcircuits.com/threads/gps-nmea-antenna-aiming-tracker.96520/page-6 at #109
you say that you fully tested this equation, did you convert it to Oshonsoft? I've searched but can't find it.
C

 

hi C,
I am looking thru my archives for this, if I find it I will post.
BTW: the atn equation we were using has one worst case error of only 0.27 degrees, what accuracy are you working too.

Also remember the rounding 'funnies' we had with Oshonsoft, when using 5 or 6 decimal places in the maths.

E

 

hi C,
I am looking thru my archives for this, if I find it I will post.
BTW: the atn equation we were using has one worst case error of only 0.27 degrees, what accuracy are you working too.

Also remember the rounding 'funnies' we had with Oshonsoft, when using 5 or 6 decimal places in the maths.

E

Hi E,
1Deg is fine.
I've also been looking.
Is this any good?:
C

 

Well done B.
Hi J,
Is this the total equation?
Is it correct that Pi and b need to be set up as CONSTANTS?

E,
Here: https://forum.allaboutcircuits.com/threads/gps-nmea-antenna-aiming-tracker.96520/page-6 at #109
you say that you fully tested this equation, did you convert it to Oshonsoft? I've searched but can't find it.
C

It is not the total equation. It needs to check the four quadrants, x>0 y>0, x>0 y<0 , x<0 y<0 , x<0 y>0
I have tested it succesfully in another language and will write it in Oshonsoft soon.

 

It is not the total equation. It needs to check the four quadrants, x>0 y>0, x>0 y<0 , x<0 y<0 , x<0 y>0
I have tested it succesfully in another language and will write it in Oshonsoft soon.

Hi J,
Look in #362 It looks as though Octants were use. The first one works, but the < and > need introducing.
Did you recall if CASE was used?
There is also a Z component
C.

 

Hi J,
Look in #362 It looks as though Octants were use. The first one works, but the < and > need introducing.
Did you recall if CASE was used?
There is also a Z component
C.

Octants were needed in the other equation. This calculates in 90 degrees quadrants.
How the direction is calculated with x, y, z?

 

Octants were needed in the other equation. This calculates in 90 degrees quadrants.
How the direction is calculated with x, y, z?

Hi J,
Just found your calculation for Octants.

Thinking about it, as the main purpose for me is videos, and the Z component is for tilt compensation, X and Y will most likely be ok..

Here is a page of explanation:
C

 

Here is the Oshonsoft untested version of arctangent.
Might have syntax and other errors.

' Result in degrees clockwise from North = 0
'
function atn( x as single, y as single) as single
const b=0.596227
dim w as single

if y=0 then
   y=y+0.000001 ' to prevent divide by zero
endif

w=x/y
at1 = 90*w*(b+w)/(1+2*b*w+w*w)
at2 =-90*w*(b-w)/(1-2*b*w+w*w)

if x>=0 then 
if y>0 then
   atn=at1
endif
endif 

if x>=0 then
if y<0 then
   atn=180-at2
endif
endif
 
if x<0 then
if y<0 then
  atn=180+at1
endif
endif

if y>0 then
if x<=0 then
  atn=360-atn2
endif
endif

end function

 

Here is the Oshonsoft untested version of arctangent.
Might have syntax and other errors.

Hi J,
This compiles, does it look somewhere near?

What is 'W' in the equation?
C.

 

hi C,
That file is zero bytes long!
E

 

hi C,
That file is zero bytes long!
E

Hi E,
Try again.
C.

 

Hi J,
This compiles, does it look somewhere near?

What is 'W' in the equation?
C.

You can see, that it is x/y or the tangent from which we calculate the corresponding angle
Why do you rename it "what" ?

edit: found that w is a stupid reserved undocumented internal variable!

 

Here is the Oshonsoft untested version of arctangent.
Might have syntax and other errors.

' Result in degrees clockwise from North = 0
'
function atn( x as single, y as single) as single
const b=0.596227
dim w as single

if y=0 then
   y=y+0.000001 ' to prevent divide by zero
endif

w=x/y
at1 = 90*w*(b+w)/(1+2*b*w+w*w)
at2 =-90*w*(b-w)/(1-2*b*w+w*w)

if x>=0 then
if y>0 then
   atn=at1
endif
endif

if x>=0 then
if y<0 then
   atn=180-at2
endif
endif

if x<0 then
if y<0 then
  atn=180+at1
endif
endif

if y>0 then
if x<=0 then
  atn=360-atn2
endif
endif

end function

atn2 should be at2 in the function.

 

I tested it with x=1, y=1 and got 44.999, which is correct. I will test the whole circle.

 

atn2 should be at2 in the function.

Hi J,
I spotted that, see #372.
C

 

You can see, that it is x/y or the tangent from which we calculate the corresponding angle
Why do you rename it "what" ?

edit: found that w is a stupid reserved undocumented internal variable!

Hi J,
Is it correct to rename 'W' as tang?
C.

 

Hi J,
Is it correct to rename 'W' as tang?
C.

Yes, it is ok.

I tested x, y = 0,1... 1,0....0,-1 and - 1, - 1 all correct.
0,1 gives 360. If needed could add:
if atn >= 360 then
atn = atn-360