# Cosine Error: 5mm Ball Tip Stylus at 7.5° Angle

• Ranger Mike
In summary, the cosine error of a measured point on the surface is not a simple value, but a vector that gets larger when the angle between the probe and the normal vector gets larger. The cosine error can be quite large and can affect the accuracy of measurements.
Ranger Mike
Gold Member
Cosine error of a measured point on the surface is not a simple value, but a vector. If the ball contacts the part surface at a point located a distance from the theoretical or nominal point then the angle between the probe and the normal vector gets larger, P1 P2 will increase. We have cosine error. This error occurs when the part surface varies compared to the CAD nominal. If the angle between the probe actual touch point P and the normal vector P2 gets larger, P1 P2 will increase.

Help..If I have a 5 mm ball tip stylus and can only vector in to the surface at 7.5° how much cosine error will I have?

Ranger Mike said:
Cosine error of a measured point on the surface is not a simple value, but a vector. If the ball contacts the part surface at a point located a distance from the theoretical or nominal point then the angle between the probe and the normal vector gets larger, P1 P2 will increase. We have cosine error. This error occurs when the part surface varies compared to the CAD nominal. If the angle between the probe actual touch point P and the normal vector P2 gets larger, P1 P2 will increase.

Help..If I have a 5 mm ball tip stylus and can only vector in to the surface at 7.5° how much cosine error will I have?

View attachment 323978
For what you have labeled as the cosine error ( call it ##\epsilon##) in the diagram I'm getting:

$$\epsilon = r \left[ 1-\cos \alpha + \sin \alpha \cos \alpha \tan \frac{\alpha}{2} \right]$$

Last edited:
ero, thank you for taking the time to look at this.
error = 5 mm [ 1 - cos 7.5° + sin 7.5° times cos 7.5° times (tan 7.5° /2) ]

erobz
Ranger Mike said:
ero, thank you for taking the time to look at this.
error = 5 mm [ 1 - cos 7.5° + sin 7.5° times cos 7.5° times (tan 7.5° /2) ]

Second go, I think a cleaner version is:

$$\epsilon = r \left( 1- \cos \alpha \right) \left( 1 + \cos \alpha \right)$$

But they should give the same result.

Or even cleaner!

$$\epsilon = r \sin^2 \alpha$$

Sorry for all the changes...but as I keep looking I keep seeing more simplifications...

I checked the first against the last and they are equivalent. Computationally, better to use the last one!

I get ##\epsilon = 5 [\text{mm}] \sin^2 (7.5°) \approx 0.085 [\rm{mm}]##

berkeman
when checking high precision parts, the error of 0.0033" is HUGE.

thank you for the great work.

berkeman and erobz

• Optics
Replies
1
Views
1K
• Mechanical Engineering
Replies
1
Views
1K
• General Math
Replies
1
Views
1K
• Calculus and Beyond Homework Help
Replies
3
Views
4K
• Engineering and Comp Sci Homework Help
Replies
7
Views
822
• Mechanical Engineering
Replies
17
Views
7K
• Introductory Physics Homework Help
Replies
1
Views
4K
• Classical Physics
Replies
16
Views
2K
• Calculus and Beyond Homework Help
Replies
5
Views
2K
• Introductory Physics Homework Help
Replies
6
Views
5K