# 3D refraction of ray / vector operations

risingedge used Ask the Experts™
on
Dear experts,

I've been working at a problem for a while.  If anyone can help, I'd really appreciate it.  I'm working on refraction of an ultrasound beam as it crosses a planar interface between a Rexolite wedge and a steel sample block.  I've got it mostly worked out, but am struggling with a (the) key parameter:  How to define the (3D) vector after refraction across the plane.  I can get the angle with relation to the plane-normal by Snell's Law, and I'm assuming the vector will be in the plane defined by the normal and the incident wave, and the line will contain the point at which it intersects the plane.  This seems like it should be easy given the information I've already calculated, but I just can't seem to get the keystone in the top of the arch.  20140227-BeamPathWork-CloseUp.pdf
Can anyone help?

Thank You,
-Matt
Comment
Watch Question

Do more with

EXPERT OFFICE® is a registered trademark of EXPERTS EXCHANGE®
Most Valuable Expert 2014
Top Expert 2015
Commented:
If your unit incident ray is i and your unit normal is n,
and the ratio of wavelength between the two media is r,
then your refracted ray will be
ri+n(rn·i-sqrt(1-r^2+(rn·i)^2)))

Commented:
You are good.20140227-BeamPathWork-CloseUp-2.pdf

Thank You, now I can go to sleep.

Commented:
Sorry, I have a follow-up:  Does this equation have a name?  I wanted to look more into it to see if there were any assumptions (small-angle, etc.).  I'm getting angles that differ slightly from direct Snell results.
Most Valuable Expert 2014
Top Expert 2015

Commented:
If the number in the square root is negative, you'd have total internal reflection.
What are the numbers you are using when you get slightly different angles?

Commented:
If I try the simple case:
i=<0,0,1>
n=<1/Sqrt[2],0,1/Sqrt[2]>
r=0.73

I get <-0.28445, 0., 1.01465> which is 29.3 degrees from -n

If I use, Snell's Law:
Theta2 = 45 Degree;
Theta1 = ArcSin[Sin[Theta2]*0.73] = 31.1

I'm sure it's user error, but I'm searching for it.
Thanks again!
-Matt
Most Valuable Expert 2014
Top Expert 2015

Commented:
sorry, I used r and n the wrong way, try
ri-n(rn·i-sqrt(1-r^2+(rn·i)^2)))

Commented:
Hmm... That gives the right angle, but looks to be the mirror image across the refraction plane (i.e. reflected).  BeamPathWork-3-copy.pdf

Do you have a source, or are you deriving?

Thanks for the help!
Most Valuable Expert 2014
Top Expert 2015

Commented:
Looks like I reversed the sign of i too.

Commented:
I'm sorry, can you spell it out for me?  I tried changing all of the i's to (-i)'s, and a few other permutations, but I failed.
Most Valuable Expert 2014
Top Expert 2015

Commented:
i=<0,0,1>
n=<0.707106781186548,0,0.707106781186547>
r=0.73
ir-n(ir·n-sqrt(1-r^2+(ir·n)^2)))=<0.240619517519044,0,0.970619517519044>
-ir-n(-ir·n-sqrt(1-r^2+(-i·rn)^2)))=<0.970619517519044,0,0.240619517519044>
Is this what you got, and does it make sense with your conventions?

Commented:

I really do appreciate your considerations.
Most Valuable Expert 2014
Top Expert 2015

Commented:
I'm not sure what convention to use, in http:#a39894769, the norm was pointing with the incident ray, but in http:#a39896563, the norm is pointing against the incident ray, so the sign convention changes.

With
i={0,0,1}
n={0.70710678118655,0,-0.70710678118655}
r=0.73
the ray that makes sense to me when I draw it is
ri - n(ri·n + sqrt(1 - r^2 + (ri·n)^2)) = {-0.240619517519044,0,0.970619517519044}

Do more with

Submit tech questions to Ask the Experts™ at any time to receive solutions, advice, and new ideas from leading industry professionals.