topic Re: Simple Angle Conversion in LabVIEW

Simple Angle Conversion

FireFist-Redhawk — Thu, 16 Jul 2020 15:17:58 GMT

I cannot think this morning... what I need is super simple I feel like, but I can't figure out an easy way to implement it. Please halp.

Inputs: angles c and d:
current angle = c
desired angle = d

Output: angle z that is closest angle to c but is also in the set {d, d+-360, d+-720, ...}

examples:

c=0
d=270
z=-90

c=765
d=0
z=720

c=-361

d=-1

z=-361

c=0

d=3,602

z=2

Re: Simple Angle Conversion

GerdW — Thu, 16 Jul 2020 14:56:29 GMT

Hi Andrew,

@FireFist-Redhawk wrote:

inputs: angles c and d:
current angle=c
desired angle = d

output: angle z that is closest angle
examples:

c=765
d=0
z=720

Please explain this example!

Why should z be 720° when c=765° (=720°+45°) and d=0°???

Re: Simple Angle Conversion

FireFist-Redhawk — Thu, 16 Jul 2020 15:13:08 GMT

Hi GerdW,

I'm not sure I understand you're query. If my current angle shows 765, and I want to go to 0 (same place for me as +-360,+-720), then z is 720 because its the member of that set whose value is closest to my current angle, 765.

Re: Simple Angle Conversion

FireFist-Redhawk — Thu, 16 Jul 2020 15:19:23 GMT

That description was initially a comment label in LV, I went back and cleaned it up to possibly get rid of the confusion, and added a couple more examples.

Re: Simple Angle Conversion

crossrulz — Thu, 16 Jul 2020 15:26:17 GMT

I think this does everything you need.

The idea here is to find the differences and keep that difference in a shift register. Add (or subtract) 360 until the absolute value of the difference is greater. The previous value is therefore the closest value.

Re: Simple Angle Conversion

altenbach — Thu, 16 Jul 2020 15:25:29 GMT

Can you explain it graphically instead?

Re: Simple Angle Conversion

GerdW — Thu, 16 Jul 2020 15:32:28 GMT

Hi Andrew,

please explain this one:

c=0 d=3,602 z=2

Edit:

I think I get this one - decimal separator confusion. I'm used to commas instead of points…

Re: Simple Angle Conversion

Sebastian.Weber — Thu, 16 Jul 2020 15:54:06 GMT

Let's say

e=(c-d)

and z'=(z-d)

then, you are looking for a z' in {..., -720, 360, 0, 360, 720,...} for which the difference to e is minimal.

Dividing e by 360 using the quotient & remainder VI gives a quotient which, multiplied with 360 again, is the next lower multible of 360 to e. If the remainder is >180, add 360 to get the nearest value. This is z'. Add d to get z.

In LabVIEW:

EDIT:

c=0
d=3,602

z=2

z must be in {..., d-360, d, d+360,...}, so it must be d=3,602 to be next to c....

Re: Simple Angle Conversion

altenbach — Thu, 16 Jul 2020 15:57:56 GMT

@crossrulz wrote:

I think this does everything you need.

That's very dangerous code because it will run forever if any of the inputs is NaN. 😮 Maybe you want to reverse some logic so it stops on FALSE instead.

(I am pretty sure there is a loop-free solution ;))

Re: Simple Angle Conversion

FireFist-Redhawk — Thu, 16 Jul 2020 17:08:56 GMT

@altenbach wrote:

Can you explain it graphically instead?

I think we all have the idea at this point but it took me a few mins to make, so may as well post.

Re: Simple Angle Conversion

FireFist-Redhawk — Thu, 16 Jul 2020 17:12:02 GMT

@altenbach wrote:
(I am pretty sure there is a loop-free solution ;))

That is what I was looking for, a loop-free way of doing it.

Sebastian, thanks. Your solution works nicely. Kudos 😁

Re: Simple Angle Conversion

altenbach — Thu, 16 Jul 2020 17:48:47 GMT

Here's Sebastian's code slightly simplified... (Please validate!)

(Note that I swapped the comparison so an exact translation would use "less or equal". Not sure what's correct. If the remainder is exactly 180, both directions are equally close).

Re: Simple Angle Conversion

FireFist-Redhawk — Thu, 16 Jul 2020 18:03:08 GMT

@altenbach wrote:

Here's Sebastian's code slightly simplified... (Please validate!)

(Note that I swapped the comparison so an exact translation would use "less or equal". Not sure what's correct. If the remainder is exactly 180, both directions are equally close).

Looks like that does work, thanks altenbach!

If anyone is wondering what I'm doing, I'm building a routine for operators to perform an encoder verification test on a unit (the unit is a slip ring that has a built-in encoder). And I needed this as part of the algorithm telling the SmartMotor (another thing we make at a different facility) where to travel to next.