Why the eyepiece faces sideways from the star
Someone has aimed a tube at a star but is looking through a small eyepiece on its side. There is no large lens in front of that eyepiece. How does light find this turn—and why put something inside the tube that blocks part of the opening?
In a Newtonian reflector, the large concave primary collects light and forms an image; the small flat secondary turns the already converging bundle towards a side focuser. The secondary makes the image accessible; it does not take over the primary’s focusing action. To understand the difference, first imagine the telescope without the little mirror. University of Arizona: a mirror with a fold flat, slide 13-39.
We will follow the light, examine how the parts are made and distinguish a misplaced focus from a faulty surface. The numerical telescope below is a hypothetical computer model, not an instrument we have physically assembled and tested.
The mirror was metal first
In the seventeenth century, a simple lens brought different colours to different distances. Newton explained how this led him to reflection: different colours obey the same geometrical reflection law. He also stressed the difficulty of polishing metal and giving it the required shape. Newton’s own text, ff. 461v–462r.
His first reflector dates to 1668; the instrument preserved at the Royal Society is his second model, from 1671. It has nested cardboard tubes and a wooden ball support. A photograph of a surviving telescope must not automatically inherit the date of the first experiment. The owner’s catalogue, MOB/005.
The historical surface also differs from the modern one. In Opticks, Newton describes concave metal ground to a sphere, and tarnishing that required renewed polishing. A modern parabolic glass mirror with a thin front coating is a different technological solution. The old account is not a recipe for safe home metallurgy. Newton, Opticks, Book I, Part I, pp. 75–77.
Mirrors removed the primary reflector’s chromatic focusing error, rather than every optical error. Lens eyepieces and correctors may have their own chromatic errors; lens objectives later gained substantial correction too, as explained in the article on achromats.
The primary already determines where the image forms
A star is so far away that its bundle is almost parallel near the telescope. Rays reach the rear of the tube and reflect from the primary’s concave front surface. The surface slope changes from centre to edge: each region sends its ray towards the image.
For an ideal paraboloid, a parallel bundle along the axis converges exactly to one point. This three-dimensional surface is generated by revolving a parabola. A sphere is simpler in shape, but its marginal and near-axis rays from an infinitely distant star do not share a focus: spherical aberration results. The size of that defect depends on aperture and focal ratio; the name “Newtonian” alone does not guarantee a parabolic primary. Arizona: the parabola, slide 46.
Without the secondary, the real image would lie in front of the primary, inside the open part of the tube. A paper screen could mark that plane for an ordinary safe target, but a head or camera would obstruct incoming light. We need a way to take the bundle sideways. The concept of a real image is explained in the article on focus and eyepieces.
Branch · Surface geometryOptional · you can skip this
Checking the exact focus, rather than merely drawing one
Choose x along incoming light, place the primary vertex at V = (f,0,0) and its future unfolded focus at F = (0,0,0). All coordinates here are in millimetres. The surface is:
x=f-\frac{y^2+z^2}{4f},\qquad \mathbf n=\left(1,\frac{y}{2f},\frac{z}{2f}\right).
n is a normal vector, not necessarily a unit vector. For unit incoming direction i, the reflected direction r is:
\mathbf r=\mathbf i-2\frac{\mathbf i\cdot\mathbf n}{\mathbf n\cdot\mathbf n}\mathbf n.
This is the equal-angle reflection law in vector form: the tangential component stays unchanged and the normal component reverses. For f = 600 and axial input \mathbf i=(1,0,0), every clear ray points towards F. Our independent check is the path length from the plane x = − 100 to mirror point M and then to F:
∣MF∣ = 2f − x, (x + 100) + ∣MF∣ = 2f + 100 = 1300 mm.
It is the same for every zone, not just the drawn marginal rays. The parabola’s curvature near the vertex is 1/(2f), so the radius of curvature at the vertex is R = 2f = 1200 mm. This does not make the whole parabolic surface part of one sphere. The surface and R/F relationship in the source.
For comparison, a sphere with the same vertex and R=1200 mm sends an axial ray at 70 mm height to x ≈ 1.02345 mm instead of zero. It crosses the axis closer to the mirror. This calculates geometrical spherical aberration; it does not measure a real product’s quality.
The small mirror changes the route, not the convergence
Place a flat reflector at 45° to the main axis in the returning bundle. It turns the bundle by about 90° towards a side opening. The wavefront is already converging; an ideal plane simply turns the entire geometry to another location. Stellafane: the secondary’s function.
Our example has a 150 mm primary and a 600 mm focal length: f/D=4, or f/4. The axial distance from the primary vertex to the secondary centre is 450 mm; the outgoing focus is another 150 mm from there. With the inside tube wall 90 mm from the axis, the focus extends 60 mm beyond it. It is accessible outside, but this does not specify a ready-made focuser: wall thickness, attachment, travel and the sensor or eyepiece focal-plane position must be included.
The secondary is usually elliptical: an ellipse tilted into position gives an almost circular projection for the axial bundle. Its quoted size is the minor axis, rather than its longest diameter. Too large, and it blocks unnecessary light; too small, and it clips the converging cone or illuminated field. Its required size is determined with the focus position, rather than chosen separately for the appearance of a small “spider.” Stellafane: secondary sizing.
The flat turns an already converging bundle
On-axis star · α=0
Off-axis star · α=0.01 rad
Hypothetical D=150, f=600 mm; the flat has a 48 mm minor axis and lies in x+y=150. Equal x and y scales; glass behind the reflecting surface is schematic. Lines are exact geometrical traces of six clear rays; the blocked central ray is omitted. No spider, diffraction or atmosphere. Focus y=150 mm; wall y=90 mm.
A numerical enlargement of the focus: x coordinate at y=150 mm
| y at x=600, mm | On axis, mm | Off axis, mm |
|---|---|---|
| -70 | 150.000 | 143.931 |
| -50 | 150.000 | 143.963 |
| -30 | 150.000 | 143.986 |
| 30 | 150.000 | 143.992 |
| 50 | 150.000 | 143.974 |
| 70 | 150.000 | 143.945 |
A meridional section, not a complete star spot. All rays coincide in the second column; the third spans about 0.060 mm, invisible at tube scale.
Branch · Folding the pathOptional · you can skip this
Why the focus moves to (150, 150)
In the section, the small mirror’s plane is x + y = 150. Reflecting point F = (0,0) across this line gives F′ = (150,150). The reflector centre is C = (150,0), so the axial segments are:
VC = 600 − 150 = 450 mm, CF′ = 150 mm.
The optical focal length has not become 450 mm because the bundle turns. In this model, a 48 mm minor axis corresponds to a major axis 48√(2) ≈ 67.88 mm. Each oblique ray is also checked for interception by the finite ellipse. An infinite mathematical plane cannot stand in for a real mirror.
There is no central axial ray in the drawing: it would encounter the opaque secondary before reaching the primary. Only clear zones are shown. Thin reference axes do not mean that light passes through mirrors.
Why a star at the field edge can have a tail
A star to one side also sends an almost parallel bundle, but at an angle to the axis. A paraboloid focuses the axial bundle exactly; that is not a promise of perfect imaging throughout the field. Coma makes an off-axis star image asymmetric. Changing focus alone cannot bring all these rays to one point; astigmatism and field curvature may contribute too. Arizona: off-axis limits of the paraboloid.
The second panel of the same diagram uses an angle of 0.01 rad, about 0.573°. Its table enlarges the focus region numerically: six meridional rays reach different x coordinates spanning about 0.060 mm. This is a section of an exact geometrical calculation, not the full three-dimensional spot, diffraction pattern or a photograph of a star. A wide sensor can include regions where these errors are more noticeable; a suitable corrector adds its own optics and spacing requirements.
Obstruction is a different mechanism. The secondary and supports remove part of the aperture; the boundaries of the open pupil determine its diffraction pattern. Thin vanes hold the mirror but can produce spikes around bright stars. Those are neither coma nor scratches on the primary. Supports need to be thin in projection while remaining rigid. Stellafane: supports and diffraction.
Branch · Light budgetOptional · you can skip this
32% of the diameter does not mean 32% of the area is lost
For a circular primary D=150 mm and circular secondary projection d=48 mm, the open geometrical area fraction is:
(A_(open))/(A_(primary)) = 1 − (((d)/(D)))^(2) = 1 − (0.32)^(2) = 0.8976.
The secondary itself blocks 10.24% of the area, rather than 32%. If, hypothetically, both reflections have a coefficient of 0.90, the power fraction after these two reflections and this obstruction is:
T = 0.8976 × 0.90 × 0.90 = 0.727056 ≈ 72.7%.
Multiply the two reflections because each passes on its fraction of the preceding power. Stellafane: successive mirror losses. The 90% values are chosen for this example, not measured; vanes, field effects, scattering, eyepiece and detector are excluded.
Do not insert D−d=102 mm as a “new diameter” in every resolution equation. An annular pupil has a 150 mm outside dimension and its own diffraction pattern. Equal area does not make it identical to a smaller solid circle; the limitations of the aperture–image relationship are discussed in the article on resolution and signal.
Which parts must be made and supported
A modern mirror divides the work between the substrate, which maintains shape, and the thin coating, which reflects light. Borosilicate glass expands less than ordinary plate glass, helping during temperature changes, but a glass name does not certify polishing accuracy. A typical coating is vacuum-deposited aluminium with a protective layer; its particular composition and spectral performance come from the manufacturer. Stellafane: substrate and coating.
The coating belongs on the front optical surface. In a household mirror coated behind glass, light traverses the substrate first, introducing refraction, dispersion and unwanted reflections. It is not an equivalent substitute for a telescope’s flat mirror. A thin protective dielectric over metal is different from thick glass ahead of a rear coating. Edmund: first and second surfaces.
The primary’s sequence is: blank → curvature generation → fine grinding and polishing → figure testing and correction → coating → cell. Each stage checks a different property: a shiny finish indicates low scattering, but does not establish a correct paraboloid by itself. Detailed mirror manufacture and testing require a separate treatment; here the key is to avoid ordering a coating as a substitute for correcting the figure. Stellafane: mirror-making and testing sequence.
From front opening to rear cell
- Front ring
- Spider and secondary
- Tube and focuser
- Primary: front coating
- Supports and adjustment
Assembly schematic, without dimensional scale or rays. Parts are separated for reading; the mirror must not rest on its reflecting face. Cells support the substrate, preserve airflow and allow tilt. The number of support points depends on mirror diameter and thickness.
Stellafane makes its rear cell from plywood, bolts and springs: springs maintain tension and bolts adjust tilt. Thin metal spider vanes face the incoming light edge-on while maintaining rigidity in the required directions. A part’s material follows its load, rather than its resemblance to a mirror.
During layout, first reconcile the primary, tube and required external focus; then trial-fit the cell, secondary and focuser. In this project, the rear cell can slide along the tube to test the optical layout before permanent attachment. After fitting, all assemblies must be secure before fine collimation. A trial position is no permission to tilt unsecured optics.
| Assembly | Its purpose | What to check after assembly |
|---|---|---|
| Primary’s rear cell | Supports weight without significant bending and allows tilt | Position retained during pointing; ventilation; no pressure on the reflecting face |
| Spider and secondary holder | Keep the flat in front of the focuser | Rigidity, position/tilt adjustment and secure retention |
| Focuser | Moves the eyepiece or sensor to the required plane | Available travel and no tilt under working load |
| Tube and stray-light shielding | Maintain relative positions and limit unwanted illumination | Rings or internal projections do not clip the working bundle and field |
| Air path | Allows the mirror to approach ambient temperature | Ventilation stays open; changes of sharpness with time are separated from focuser movement |
Cell materials, adjustment and airflow are described in Stellafane’s construction; the relationship between focuser height, secondary size and field is discussed in its tube design. This is a functional table, not a drawing for unknown parts. A large thin mirror may need a different support system from a small thick one; counting screws cannot replace a bending calculation.
Alignment changes direction; squeezing changes shape
First position the secondary under the focuser. Next aim the focuser axis through it at the primary’s centre; then tilt the primary to bring its optical axis to the required output. Changing one assembly can require repeating an earlier step. Stellafane: collimation sequence.
These are three different tasks. A pleasing central reflection does not prove the whole field is illuminated, and the return of one laser spot alone does not establish every component’s correct position. A real fast Newtonian (with a small f/D) may have a designed secondary offset; demanding concentricity of every visible outline without considering the geometry is wrong. A designed offset: Sky-Watcher Quattro, p. 8.
Tilting the secondary moves the outgoing focus
Flat: 45°
Flat: 46°
The same hypothetical telescope. The horizontal scale is enlarged 6.36 times relative to the vertical, equally in both panels; dashed lines are geometrical axes, not rays. A 1° tilt gives F′=(144.765, 149.909) mm instead of (150, 150). This is focus displacement by an ideal flat, not simulated coma from a tilted primary or a star-test photograph.
In this model, tilting the little flat by 1° changes the central geometrical axis direction by 2° and displaces the focus sideways by about 5.235 mm. The ideal primary stays on axis: the flat has moved its point focus away from the focuser axis, rather than spoiled it. Tilting the primary so the sensor views an off-axis field of the paraboloid is another problem, with possible aberrations. The drawn comparison is not a photograph of a star test.
Excessive cell pressure really can alter an optical surface. Tightening retainers is no cure for blur: pinching creates another error. Check secure retention without deformation separately from tilt adjustment; the secondary must still remain unable to fall onto the primary. Stellafane: mounting without pinching.
A poor image needs a separate check of its cause
| Observation | What could explain it | A check that separates causes |
|---|---|---|
| Sharpness improves up to the stop, but focus is never reached | Required plane beyond focuser travel | Compare focus, attachment and sensor positions; do not begin by regrinding the mirror |
| Sharp centre, stretched edge stars | Off-axis errors, including coma for a simple paraboloid | Compare the same star in the centre and near the edge without changing the setting |
| Illumination falls or is asymmetric | Secondary size, position or a projection clips the field | Check the visible aperture from the working focal plane and trace geometry |
| Sharpness and position change under camera weight | Focuser or attachment flexure | Compare component positions with and without working load |
| Image changes during cooling | Thermal changes in the optics or air; the atmosphere also changes | Repeat over time, recording focus and conditions; do not claim one cause is proven |
These are working hypotheses, not diagnoses from one symptom. Collimation, focusing and surface testing answer different questions. Our geometrical model has no thermal currents, sensor noise or real atmosphere, so it cannot establish that these are absent in an observation.
Visual use needs a suitable eyepiece, exit-pupil position and stable pointing. Photography needs the sensor at the focal plane, an appropriate illuminated field, rigidity and tracking during exposure. The same mirror layout does not guarantee that every tube and mount suits both tasks.
The side eyepiece now has a reason: the primary creates the image, the flat gives access to it, and the mechanics preserve relative positions. Before mirror manufacture or a tube on a Dobsonian mount, try an available paper check: mark the primary, unfolded focus, secondary and external focus. Identify the blocked rays and the sensor’s required position. That needs neither bought parts nor a view through a telescope.