P2 SPIN python to spin

pic18f2550

Hello,

does anyone know python and how to implement this in Spin?

Thanks.

    a = math.atan2(y, x)
    if ( a < 0):
        a += 2 * math.pi
    a = math.degrees(a) + 0.78

ersmith

Spin does not support floating point operations natively, so you'll have to download a floating point object. For Spin2 there's a BinFloat object.

Cluso99

I think CORDIC may be useful here but no idea how it works.

pic18f2550

I found out what the code section is supposed to be for.
It is a kind of sin-cos encoder with an offset value.
I am actually only looking for the CORTIC part, where an angle value is output from the x and y values.

Ariba

@pic18f2550 said:
...
I am actually only looking for the CORTIC part, where an angle value is output from the x and y values.

From the Spin2 documentation:

XYPOL(x, y) : length, angle32bit               Convert (x,y) to (length,angle32bit)

pic18f2550

pub atan2(x, y) : a
  a := xypol(x, y) sca ( 360 * 1_000)

This should fit.

evanh

You've got me trying to nut out how to do this ... Looking at what Eric has done for C support of same functions ... I've figured out for each case of arcsine, arccosine and arctangent each requires a single square-root calculation, amongst a few multiplies and divides, preceding the Cordic's QVECTOR command.

Here's three routines that I've tested/compared, correct for unsigned numbers first qradrant, against Eric's built-ins:

float  c_asin( float sine )  // sohcahtoa
{
    uint32_t  opposite, adjacent, angle;
    float  scale = (float)0x4000_0000;

    adjacent = sqrtf( 1.0 - sine * sine ) * scale;
    opposite = sine * scale;
    __asm {
        qvector adjacent, opposite
        getqy   angle
    }

    return  angle / (float)0x8000_0000 * pi;
}



float  c_acos( float cosine )  // sohcahtoa
{
    uint32_t  opposite, adjacent, angle;
    float  scale = (float)0x4000_0000;

    adjacent = cosine * scale;
    opposite = sqrtf( 1.0 - cosine * cosine ) * scale;
    __asm {
        qvector adjacent, opposite
        getqy   angle
    }

    return  angle / (float)0x8000_0000 * pi;
}



float  c_atan( float tangent )  // sohcahtoa
{
    uint32_t  opposite, adjacent, angle;
    float  scale = (float)0x4000_0000;

    adjacent = scale / sqrtf( 1.0 + tangent * tangent );
    opposite = tangent * adjacent;
    __asm {
        qvector adjacent, opposite
        getqy   angle
    }

    return  angle / (float)0x8000_0000 * pi;
}

pic18f2550

XYPOL(x, y) : length, angle32bit Convert (x,y) to (length,angle32bit)
Pretty thin description

What is the value range of X & Y ? 32bit with sign?
My values are in the range $F800..$07FF(-2048..2047 ) so no scaling should be necessary.

Two return values ? length, angle32bit

length should be sqr( X² + Y²)

angle32bit should contain the angle but what are the values e.g. at 45°, 90°, 180 and 360° without scaling?

In PASM:
QVECTOR x, y
GETX länge
GETY Winkel

evanh

SCA will require a single input. It can't be tacked in that way. I imagine most operators will have that issue.

pic18f2550

even without SCA the problem is present.

xy := xypol(x, y)'' sca ( 360 * 1_000)

evanh

I'm not sure how spin syntax goes. The xy will be no good as well, it'll need to be a pair of variables, maybe x,y := xypol(x,y)

pic18f2550

that's new to me, I didn't know that was possible.
Thanks

JonnyMac

If you don't need all of the values from a method that returns multiple values, you can use an underscore to ignore one (or more) of them. For example:

    x, _ := xypol(x, y)