Building an 8” Stevick-Paul — Notes for the Workshop

Companion to DESIGN.md. Aimed at an instructor advising a student who wants an 8” Stevick-Paul (SP) rather than the usual Newtonian.

1. Should this student build it?

Verdict: an 8” SP is an advanced project — several times the work of an 8” Newtonian, with one genuinely hard step (the convex secondary) and one skill that Newtonians never teach (off-axis tilted alignment). It is an excellent second or third telescope and a poor first telescope.

Quick scorecard:

	8” Newtonian	8” Stevick-Paul
Precision surfaces	1 powered + 1 flat	2 powered + 1 powered + 1 flat
Convex surface to test	no	yes (hardest step)
Central obstruction	~16–25%	none
Alignment	star test, well documented	geometric, no standard procedure
Build support/kits/parts	enormous	a handful of instruments, none
Typical build time (first timer)	weeks–months	months–years, high risk

Decision guide:

First mirror / first scope → steer to an 8” f/6–f/8 Newtonian. Build the SP as a follow-on.
Experienced, motivated, wants the challenge → build the SP, ideally folded, f/12–f/15, A = 0.4–0.45, spherical primary (see §2 and §5).

2. Concrete prescriptions (8”, radius ratio A = 0.4)

Both use: primary diameter D = 8”, clear field d = 0.62”, central beam diameter A·D = 3.2”. Minimum secondary diameter 3.57” (use a 4.25” blank); minimum tertiary 4.81” (use a 5” blank). Axis distances are a1 = 6”, a2 = 12”, a3 = 3” in both designs (for A = 0.4, 8”, a1 is essentially independent of f-ratio).

8” f/12 (recommended all-round)

Mirror	Diameter	Radius	Figure	Tilt	Spacing to next surface
Primary	8.0”	192.0”	paraboloid	2.97°	e1 = 57.6”
Secondary	4.25” (3.57 min)	−76.8”	spherical, convex	5.20°	e2 = 76.8”
Tertiary	5.0” (4.81 min)	76.8”	spherical, concave	2.23°	e3 = 38.4” to focus
Diagonal	1⅝” minor axis	flat	flat	42.77°	residual cone to focus
Focal plane	1.0” design field	—	—	tilt ≈ 8.9°	—

Derived: k = 1.75; T1 = C/N with C = 35.67; T2 = k·T1; T3 = T2 − T1; T4 = 45° − T3; fp tilt ≈ 4·T3. Exact raytraced tilt runs ~0.2° above the formula.

8” f/15 (easiest optics, longest tube)

Mirror	Diameter	Radius	Figure	Tilt	Spacing to next surface
Primary	8.0”	240.0”	sphere OK	2.38°	e1 = 72.0”
Secondary	4.25” (3.57 min)	−96.0”	spherical, convex	4.16°	e2 = 96.0”
Tertiary	5.0” (4.81 min)	96.0”	spherical, concave	1.78°	e3 = 48.0” to focus
Diagonal	1⅝” minor axis	flat	flat	43.22°	residual cone to focus
Focal plane	1.0” design field	—	—	tilt ≈ 7.1°	—

f/12 vs f/15 tradeoff: f/15 removes the need to parabolize the primary and gives a flatter, more forgiving focal plane, at the cost of a longer instrument and larger spherical-mirror radii (bigger blanks, longer grinding strokes). f/12 is more compact and usable; you pay with a (small) parabolizing job and a slightly steeper field.

Fold option: following Novack’s FSP, the flat can fold the light and eliminate the separate diagonal, giving a compact package. Novack judged a folded f/12–f/15 with three spheres the easiest off-axis scope that can actually be made at reasonable speed.

3. Why the Stevick-Paul cannot be scaled to faster focal ratios

The blockage is not aberration correction (that is the design’s whole strength, perfect to third order) — it is the tilt of the focal plane, which is intrinsic to using the Paul system off-axis and which grows as the cone gets faster.

Focal-plane tilt scales as 1 / f-number. From the design equations (A = 0.4, 8”): T1 = C/N, T2 = 1.75·T1, T3 = 0.75·T1, and fp tilt ≈ 4·T3 ≈ 3C/N ≈ 107/N degrees. So:

f/#	Primary tilt T1	Focal-plane tilt (≈3C/N)
f/15	2.38°	~7.1°
f/12	2.97°	~8.9°
f/10	3.57°	~10.7°
f/8	4.46°	~13.4°
f/6	5.95°	~17.8°
f/5.6	6.37°	~19.1°

Stevick deliberately caps designs at f/10, keeping focal-plane tilt under 10°.

Eyepieces can’t tolerate a steeply tilted image in a fast cone. A fast cone has only a small depth of focus, so a tilted focal plane smears the field unless the eyepiece itself is tilted by the same angle. At large tilt angles the eyepiece’s own aberrations and the field/exit-pupil mismatch degrade the view. On the f/6.2 prototype the eyepiece had to be kept square, and even then the top and bottom of the field were out of focus — which is exactly why Stevick recommends f/10+.
Image anamorphism (differential magnification across the tilted field) also grows with tilt. Stevick keeps it under 2%, which again forces a slow f-ratio.
Field curvature becomes relatively worse as f/ratio drops. The one residual third-order aberration — a gently curved Petzval surface — is fixed in absolute terms by the mirror radii. As the cone gets faster the depth of focus shrinks, so the same curvature eats a larger fraction of the sharp zone.
The only escape is refractive correction. An image-tilting / field-flattening lens lets you go faster, but adds glass, alignment, and chromatic issues to an otherwise all-reflecting design. As Novack put it, unless you use an image-tilting lens, the fastest practical SP for visual use is about f/10. (f/5.6 is quoted as the diffraction-limited floor, purely theoretical for visual use.)

Note the primary tilt and mirror diameters do not force slowness — radii actually shrink with f-ratio. The f-ratio floor is entirely the tilted, fast focal plane versus the eyepiece.

4. Optical performance vs conventional 8” Newtonians

Assume a well-made SP at f/12 and a good 8” Newtonian at f/6–f/8 with a ~16–25% secondary obstruction.

Where the SP wins

No central obstruction → better contrast. A ~20% obstruction pushes energy into the diffraction rings and lowers mid-spatial-frequency MTF by roughly 10–15%. For low-contrast planetary detail the unobstructed SP is genuinely better, and there is no spider diffraction spike.
Third-order perfection. All Seidel aberrations (spherical, coma, astigmatism, distortion) go to zero except Petzval curvature. Within its usable field the SP is free of the coma that limits a fast Newtonian’s off-axis image (an f/6 Newtonian’s diffraction-limited field is only a few arcminutes; f/8 is better but still comatic).
Perfectly achromatic (all-reflecting), same as a Newtonian.

Where the Newtonian wins

Flat, untilted, normally-oriented focal plane. No eyepiece tilt, no anamorphism.
Light throughput. 4 reflections vs 2: with 96% enhanced coatings, ~0.85 vs ~0.92 transmission; with standard ~89% aluminum, ~0.63 vs ~0.79. The SP gives away roughly 7–16% of its light to the extra two reflections, and each surface is another thermal/alignment liability.
Usable field. The SP is designed with ~½-magnitude vignetting at the field edge and a tunnel-vision diagonal nestling the beam; a Newtonian with a properly sized secondary has an unvignetted field.
Practicality, cost, robustness, support. A ~4 ft f/6 Newt tube on a Dob mount versus a 4-mirror tilted train at ~8 ft (f/12) focal length. Newtonians win massively on cost, time, thermal stability, and the availability of help.

Tie

On-axis sharpness: both reach the diffraction limit when well made. At 8” aperture the SP’s theoretical advantages over a good Newtonian are real but modest and mostly show up as contrast on planets and coma-free imaging over a wider (but vignetted and tilted) field.

Bottom line for the student: the SP is optically elegant and does deliver the unobstructed, coma-free image, but at 8” the payoff over a careful slow Newtonian is incremental while the build cost, risk, light loss, and field compromises are large. Its real draw is the design itself and the challenge — not a decisive performance win at this aperture.

5. Milestone checklist / project plan

Phase 0 — Design & layout

Choose f/ratio (start f/12, A = 0.4). Record target numbers from §2.
Raytrace / verify with a tilted-component program (Stevick’s Spot / WinSpot, or a modern TCT raytracer). Confirm focal-plane tilt and free cone.
Make a full-scale drawing on paper: mirror centers, principal-ray path, mirror surfaces at right angles to bisected ray angles, tick marks for clear-field diameters. (Stevick considers this mandatory — the diagonal placement and free cone are measured off the drawing.)
Decide: standard 4-mirror vs folded (FSP). Decide make-vs-buy per mirror.

Phase 1 — Primary

Grind/polish 8” f/12 primary to 192” r.o.c.
f/12 → parabolize (correction is tiny; shadows are soft — use a narrow slit).
f/15 → may be left spherical (verify SA within Rayleigh limit).
Test (Foucault / Ronchi). Note: SA tolerance equals a Newtonian of the same f-ratio.

Phase 2 — Concave tertiary FIRST

Grind/polish the 5” concave sphere to r.o.c. 76.8” (f/12) or 96” (f/15).
Null/test it as a sphere (it is testable exactly like a Newtonian mirror).
This becomes the test plate for the secondary — do it before the convex.

Phase 3 — Convex secondary (the hard step)

Grind the 4.25” convex to the mating radius.
Polish against the finished tertiary as a test plate; bring it in until it shows straight fringes (Newton’s-ring / interference test). This requires a monochromatic source and careful contact testing.
Cross-check: both surfaces must be the same radius and both good spheres.

Phase 4 — Diagonal flat

Make/obtain a flat, 1⅝” minor axis (major-axis/1.4). Ritchey-Common against the spherical primary is an amateur-proven route.

Phase 5 — Structure & mirror cells

Build tube/structure holding mirrors at the spacings (e1, e2, e3), tilts (T1–T3), and axis offsets (a1 = 6”, a2 = 12”, a3 = 3”).
Provide tip/tilt on every mirror (the tilts lock to the primary tilt, so you need independent adjusters).
Decide exit: across the parallel secondary–tertiary beam (42.77° diagonal), out the side panel (diagonal on its side, focuser near the altitude axis), or folded.

Phase 6 — Baffling

Fit a light baffle around the primary’s converging cone, sized to just pass the central (3.2”) beam, fitting the notches where light paths cross.
Blacken mirror edges; baffle the diagonal; verify no ray reaches within 1” of field center (⇒ 2” stray-light-free field).

Phase 7 — Coating

Coat all surfaces (96% enhanced recommended — 4 reflections make throughput matter). Blacken edges after coating.

Phase 8 — Assembly, alignment, first light

Assemble to the full-scale drawing; set tilts and offsets geometrically.
Focus test: verify the focal plane tilt; tilt the eyepiece holder to match (~8.9° at f/12), or accept center-only sharpness at higher tilt.
Star test; check edge-of-field vignetting (~½ mag by design).

Phase 9 — Document & iterate

Record the as-built prescription and any deviations (ground radii rarely hit target — recompute tilts/spacings from the actual primary r.o.c., as Stevick did when his 8” came out f/11.75 instead of f/12).

6. Tooling & testing summary

Required: full-scale drawing surface, spherometer, Foucault/Ronchi tester with a narrow slit, monochromatic source (or laser) for the convex/concave contact test, an interference/contact test setup for the mating spheres, and a way to measure tilt angles precisely (protractor/levels/fixtures).
No standard collimation tool applies. You align to geometry, not to a defocused star; the fixed tilt relationships mean the primary tilt is the master.
Testing the convex is the single biggest risk and the reason to finish the concave tertiary first.

7. Risk register & fallbacks

Risk	Mitigation / fallback
Convex secondary can’t be tested/matched	Finish concave tertiary first; use it as test plate; fall back to buying a custom convex sphere
Primary misses target r.o.c.	Recompute all tilts/spacings from actual value (normal, expected)
Focal-plane tilt too harsh at chosen f/#	Move to a longer f/# (f/12 → f/15) or add an image-tilting lens
Alignment never converges	Build rigid, draw-first, tip/tilt everywhere; consider the folded layout for compactness
Diagonal shadows the collimated beam	The flat must ride tangent between the collimated and converging beams (only a fraction of an inch of clearance) — derive its position from the full-scale drawing and check clearance. See `TRACE.md` §5
Cost/time overrun	Buy the primary; reduce aperture; build a Newtonian first

8. References

D. Stevick, “The Stevick-Paul Off-Axis Reflecting Telescope,” ATMJ Issue 3, pp. 4–9 (1993).
D. Stevick’s design pages (archived): https://spider.seds.org/scopes/dstevick/stevpaul.html (theory, design procedures, system layout) and folded.html / fsp.html (Novack’s folded f/10).
See DESIGN.md in this directory for the full equation set and source notes.