5

Plane intersecting a $95^\circ$ cone at right angles:

enter image description here

I'm trying to build a homemade tracking platform for my telescope. I'm hoping that I can come up with an equation that I can use to cut a bearing for it on a milling machine. I think it is a hyperbola generated by the intersection of a plane to a cone. The plane intersects the cone $34.25$ inches from the tip and at a right angle. The cone has an angle of $95^\circ$ (twice my latitude).

Blue
  • 75,673
  • There isn’t “an equation” to describe this curve. In general, a curve in 3-d requires a system of two implicit equations. One can obtain a parameterization of the curve (which is really a set of three equations) so that varying the single parameter traces out the curve. Is that what you’re looking for? – amd Jan 01 '20 at 20:53
  • I was looking for a 2-d version projected on that intersecting plane. Something like $$1 = \frac{(y-k)^2}{a^2}-\frac{(x-h)^2}{b^2} $$ – Curtis Thompson Jan 01 '20 at 21:38
  • @Blue: fully right, sorry for the mistake – G Cab Jan 01 '20 at 21:55
  • @Curtis Thompson Wooden cone? Btw If there is a means to fix the cone ( a hole along axis ?) and tilt axis (universal milling machine table ) at the right inclination you could perhaps straight away part them on the machine using a slicing milling cutter, or at least mark the line with a sharp tool. – Narasimham Jan 02 '20 at 03:28
  • @Narasimham, That is exactly what I did. Wood sort of worked but rough. I'd like to make the bearing out of 3/8" aluminum but my jig to make it is not rigid enough and the cutter chatters a lot. – Curtis Thompson Jan 03 '20 at 16:02

3 Answers3

4

Here's a side view of the scenario. Point $O$ is the tip of your cone. The cone's axis is the vertical through $O$; the horizontal represents the plane perpendicular to the axis at $O$. Angle $\theta$ is half the cone angle; so, it's your latitude. The cutting plane is represented by $\overleftrightarrow{VV'}$, which cuts the cone at a point at distance $d$ from $O$.

enter image description here

This answer explains that the eccentricity of a conic —call it $e$— is given by the angles made by the cone and by cutting plane with that horizontal plane. Here, we have $$e = \frac{\sin\angle VPO}{\sin\angle POV} = \frac{\sin\theta}{\sin(90^\circ-\theta)}=\frac{\sin\theta}{\cos\theta}=\tan\theta \tag{1}$$

The hyperbola's transverse axis is given by $$|VV'| = d\tan\angle VOV' = d\tan(180^\circ-2\theta)=-d\tan 2\theta \tag{2}$$

(Note that $2\theta$ here is $95^\circ$, an obtuse angle with a negative tangent; thus, the negative sign in $(2)$ makes the value positive, as expected.)

Choosing convenient coordinates in the cutting plane ($x$-axis aligned with $\overline{VV'}$, origin at the segment's midpoint), the equation of the hyperbola in that plane is

$$\frac{x^2}{a^2}-\frac{y^2}{b^2} = 1 \tag{3}$$ where $$\begin{align} a &= \frac12|VV'| &&= -\frac12d\tan 2\theta \\[4pt] b &= a\sqrt{e^2-1} &&= -\frac12d\tan2\theta\sqrt{\tan^2\theta-1}=\frac{d}{\sqrt{1-\cot^2\theta}} \end{align} \tag{4}$$

In the particular case where $\theta = 95^\circ/2 = 47.5^\circ$ and $d = 34.25$, this gives $$a = 195.74 \qquad b = 85.5349$$ Since $e = 1.09131$, this is "close" to a parabola near the vertex, although the resemblance fails quickly, as the hyperbola approaches asymptotes and a parabola, of course, doesn't.

Blue
  • 75,673
  • 1
    Very nice! Exactly what I was looking for. I will carefully verify before committing to the milling machine. – Curtis Thompson Jan 02 '20 at 00:02
  • 1
    Blue, The data I get from your equation seems to match what I see from my feeble attempt at making the bearing free-hand. Thanks again. – Curtis Thompson Jan 02 '20 at 01:11
  • 1
    Your results agree with mine obtained via a different method. – amd Jan 02 '20 at 07:50
  • @CurtisThompson and @ amd: Good to know. I'm annoyed (and embarrassed) by how often a silly error has made me redo a project on my laser cutter, so I want to be extra-cautious when someone else's materials are on the line. :) – Blue Jan 02 '20 at 08:07
2

An equation of a right cone with apex at the origin, axis along the $z$-axis and aperture angle $2\phi$ is $x^2+y^2=z^2\tan^2\phi$, which can be represented by the matrix $Q=\operatorname{diag}(1,1,-\tan^2\phi,0)$. Setting $\mathbf x = (x,y,z,1)^T$, the equation can then be written in the matrix form $\mathbf x^TQ\mathbf x=0$.

We can similarly write the parameterization $\mathbf p_0+s\mathbf u+t\mathbf v$ of a plane as $$\mathbf X = \begin{bmatrix}\mathbf u&\mathbf v&\mathbf p_0\\0&0&1\end{bmatrix} \begin{bmatrix}s\\t\\1\end{bmatrix} = M\mathbf x.$$ This amounts to defining a coordinate system on the plane with $\mathbf p_0$ as the origin. In this coordinate system, the conic defined by the intersection of the cone and plane has the equation $$(M\mathbf x)^TQ(M\mathbf x) = \mathbf x^T(M^TQM)\mathbf x= 0,$$ i.e., this conic has the matrix $M^TQM$.

A simple and convenient choice for $M$ for this problem is $$M=\begin{bmatrix}0 & -\cos\phi & -d\sin\alpha \\ 1 & 0 & 0 \\ 0 & \sin\phi & -d\cos\alpha \\ 0&0&1 \end{bmatrix}.$$ This makes the $y$-axis on the plane the transverse axis of the hyperbola and places the origin at the vertex of the lower branch, which is a distance $d$ down the side of the cone. We then have $$M^TQM = \begin{bmatrix}1&0&0 \\ 0&\cos2\phi \sec^2\phi & d\tan\phi \\ 0&d\tan\alpha&0\end{bmatrix}.$$ This corresponds to the equation $$x^2+y^2\cos2\phi\sec^2\phi+2dx\tan\phi=0.$$ Using any of the usual methods we can find that the center of this conic is at $\left(0,-\frac d2\tan2\phi\right)$. Translating the origin to this point produces the equation $$x'^2 + y'^2\cos2\phi\sec^2\phi=d^2\sec2\phi\sin^2\phi.$$ For $\phi\gt\pi/4$, this is a hyperbola (although since $\phi$ is close to $\pi/4$ it is very close to a parabola). From this equation, we can find that the transverse half-axis length is $-\frac d2\tan2\phi$ and conjugate half-axis length is ${d\over\sqrt{1-\cot^2\phi}}$. In your case, $\phi=95°/2=47.5°$ and $d=34.25$, so the half-axis lengths work out to be approximately $195.74$ and $85.54$, respectively.

amd
  • 53,693
  • I’m not familiar with what input the milling machine needs, but it’s not too hard to come up with parameterizations for the arc that you need to trace. With known endpoints, for example, a rational Bézier curve might be convenient. – amd Jan 02 '20 at 19:06
  • thank you for this beautiful answer. I'm trying to figure out a few things (that will probably seem trivial to you). The p0+su+tv parameterization should be analogous to ax+by+c*z=d, but I don't understand how what you wrote captures d, etc. I understand that x,y,z shouldn't be used as they are in the coordinate system of the cone and not the plane? (also there are capital X matrix and small x vector... a bit confusing to me) – dpdp Mar 23 '21 at 23:31
1

enter image description here

In Cartesian coordinate, the surface equation of the cone is

$$z^2 = \tan^2\theta(x^2+y^2)\tag 1$$

where $\theta = \frac {85}2=42.5$ deg. The normal vector to the surface of the cut is $n =(0, \cos\theta, \sin\theta)$ and the vertex point of the cut is $V(0, -d\cos\theta, -d\sin\theta)$. Then, for any point $P(x,y,z)$ in the plane of the cut, $n\cdot(P-V)=0$ holds. Explicitly, the equation of the plane is

$$y\cos\theta+z\sin\theta+d=0\tag 2$$

Thus, the equation for the edge of the cut is given by the intersection of the cone (1) and the plane (2), which is a hyperbola. For instance, it is seen from its projection onto the $xy$-plane obtained by eliminating $z$,

$$x^2 - (\cot^4\theta - 1)y^2 +2d\csc\theta\cot^3\theta\> y - d^2\csc^2\theta \cot^2\theta= 0$$

which is a hyperbola because $\cot^4\theta - 1>0$.

Quanto
  • 97,352