Screening history - 2026-06-11
Multi-mirror screening history: superseded by the optimized prescription.
This page records how four-, five-, six-, and seven-mirror seed layouts were screened against the same packaging, detector-energy, tolerance, focus, and beam-obstruction questions. The screening ended when the obstruction gate showed every seed layout was physically self-obstructed; a full prescription optimizer then produced the current unobscured four-mirror design.
Checkpoint result
The legacy seed branches remain useful test cases, but the obstruction screen blocks them from prescription readiness.
Four-mirror folded branch
The four-mirror branch was chosen for full prescription optimization because it was the lowest-complexity screened branch that fit the package and mass screens. The optimized version (10 June 2026) is now the reference design; this seed layout is self-obstructed and kept only as screening history.
95.5 mm package span; 0.343 kg mass proxy; edge RMS 1026.5 um nominal, 691.1 um after refocus, and 563.0 um after the bounded prescription step. Minimum unobstructed fraction after that step is 1%.
Four CSV / Four focus CSV / Four prescription CSV / Four obstruction CSV / Four JSON
Five-mirror reimaging branch
The five-mirror branch was the second screened family. It fits inside the envelope and carried the near-flat relay idea into the trace stack, but its seed layout is self-obstructed; it would need full prescription optimization to re-enter the trade.
103.9 mm package span; 0.349 kg mass proxy; edge RMS 932.0 um nominal, 627.3 um after refocus, and - um after the bounded prescription step. Minimum unobstructed fraction after that step is 0%.
Five CSV / Five focus CSV / Five prescription CSV / Five obstruction CSV / Five JSON
Six-mirror dual-triplet branch
The six-mirror branch was the third screened family. It kept the package and mass proxies inside the payload envelope while carrying the two-subgroup reimaging idea into the trace stack, but its seed layout is self-obstructed; it would need full prescription optimization to re-enter the trade.
111.7 mm package span; 0.354 kg mass proxy; edge RMS 1581.2 um nominal, 1188.9 um after refocus, and 1091.1 um after the bounded prescription step. Minimum unobstructed fraction after that step is 0%.
Six CSV / Six focus CSV / Six prescription CSV / Six obstruction CSV / Six JSON
Seven-mirror cold-stop relay branch
The seven-mirror branch was the highest-mirror-count screened family. It kept the package and mass proxies inside the payload envelope while carrying a foreoptics-plus-relay grouping into the trace stack, but its seed layout is self-obstructed; it would need full prescription optimization to re-enter the trade.
118.6 mm package span; 0.359 kg mass proxy; edge RMS 3496.4 um nominal, 3060.9 um after refocus, and 3114.7 um after the bounded prescription step. Minimum unobstructed fraction after that step is 0%.
Seven CSV / Seven focus CSV / Seven prescription CSV / Seven obstruction CSV / Seven JSON
Same first-stack screen
The candidates use the same field points and detector-energy proxies.
Four-mirror folded branch
| Field | Throughput | RMS spot | EE80 radius | 2x2 energy | 16x16 energy | MTF 0.5 Nyquist | Status |
|---|---|---|---|---|---|---|---|
| 0.00 deg | 86% | 1721.5 um | 2177.2 um | 0% | 0% | 0% | image-spread limited; low MTF proxy |
| 0.75 deg | 100% | 1450.0 um | 1787.1 um | 0% | 1% | 0% | image-spread limited; low MTF proxy |
| 1.50 deg | 64% | 1026.5 um | 1294.8 um | 0% | 3% | 0% | vignetting risk; image-spread limited; low MTF proxy |
Five-mirror reimaging branch
| Field | Throughput | RMS spot | EE80 radius | 2x2 energy | 16x16 energy | MTF 0.5 Nyquist | Status |
|---|---|---|---|---|---|---|---|
| 0.00 deg | 38% | 915.9 um | 1024.0 um | 0% | 2% | 0% | vignetting risk; image-spread limited; low MTF proxy |
| 0.75 deg | 92% | 1131.5 um | 1428.4 um | 0% | 1% | 0% | image-spread limited; low MTF proxy |
| 1.50 deg | 64% | 932.0 um | 1177.5 um | 0% | 3% | 0% | vignetting risk; image-spread limited; low MTF proxy |
Six-mirror dual-triplet branch
| Field | Throughput | RMS spot | EE80 radius | 2x2 energy | 16x16 energy | MTF 0.5 Nyquist | Status |
|---|---|---|---|---|---|---|---|
| 0.00 deg | 37% | 1451.7 um | 1762.9 um | 0% | 2% | 0% | vignetting risk; image-spread limited; low MTF proxy |
| 0.75 deg | 92% | 2076.5 um | 2609.4 um | 0% | 0% | 0% | image-spread limited; low MTF proxy |
| 1.50 deg | 64% | 1581.2 um | 1991.9 um | 0% | 0% | 0% | vignetting risk; image-spread limited; low MTF proxy |
Seven-mirror cold-stop relay branch
| Field | Throughput | RMS spot | EE80 radius | 2x2 energy | 16x16 energy | MTF 0.5 Nyquist | Status |
|---|---|---|---|---|---|---|---|
| 0.00 deg | 37% | 3240.4 um | 3895.4 um | 0% | 0% | 0% | vignetting risk; image-spread limited; low MTF proxy |
| 0.75 deg | 92% | 4707.1 um | 5867.8 um | 0% | 0% | 0% | image-spread limited; low MTF proxy |
| 1.50 deg | 64% | 3496.4 um | 4381.5 um | 0% | 0% | 0% | vignetting risk; image-spread limited; low MTF proxy |
Transmission proxies use one window allowance and 95 percent per mirror reflection. The 2x2 detector-energy rows remain too low, so these are measured branches to improve, not released prescriptions.
Beam path gate
The seed-family image rows are now flagged when another surface blocks the active beam.
| Candidate | Nominal min unobstructed | Nominal worst blocked | Best-step min unobstructed | Best-step worst blocked | Obstruction CSV | Gate |
|---|---|---|---|---|---|---|
| Four-mirror folded branch | 6% | 94% | 1% | 99% | Four obstruction CSV | Blocked: mirror or detector geometry clips the active beam; optimize spacing/folds before using the image-quality rows. |
| Five-mirror reimaging branch | 0% | 100% | 0% | 100% | Five obstruction CSV | Blocked: mirror or detector geometry clips the active beam; optimize spacing/folds before using the image-quality rows. |
| Six-mirror dual-triplet branch | 0% | 100% | 0% | 100% | Six obstruction CSV | Blocked: mirror or detector geometry clips the active beam; optimize spacing/folds before using the image-quality rows. |
| Seven-mirror cold-stop relay branch | 0% | 100% | 0% | 100% | Seven obstruction CSV | Blocked: mirror or detector geometry clips the active beam; optimize spacing/folds before using the image-quality rows. |
This screen uses rays centred on the physical 40 mm entrance aperture, then checks every beam segment against every other mirror disk and the detector rectangle. It is separate from mirror clear-aperture clipping and applies the no-vignetting rule to self-obstruction.
Tolerance and clearance gate
Clearance can be bought with larger mirrors, but image quality still has no WFE margin.
| Candidate | Relative yield | P90 edge RMS | P10 throughput | WFE gate | Clearance scale | Scaled package | Scaled 2x2 energy | Gate |
|---|---|---|---|---|---|---|---|---|
| Four-mirror folded branch | 0% | 2184.7 um | 62% | No WFE margin | 1.1x | 100.3 mm | 0% | Screening only: WFE has no margin, so optimize image quality before detailed tolerancing. |
| Five-mirror reimaging branch | 100% | 1009.3 um | 62% | No WFE margin | 1.1x | 108.6 mm | 0% | Screening only: WFE has no margin, so optimize image quality before detailed tolerancing. |
| Six-mirror dual-triplet branch | 100% | 1886.0 um | 62% | No WFE margin | 1.1x | 116.3 mm | 0% | Screening only: WFE has no margin, so optimize image quality before detailed tolerancing. |
| Seven-mirror cold-stop relay branch | 30% | 5451.0 um | 62% | No WFE margin | 1.1x | 123.1 mm | 0% | Screening only: WFE has no margin, so optimize image quality before detailed tolerancing. |
The relative perturbation yield is measured against each branch's current edge-field spot, so it is a stability screen rather than a detector-readiness pass. The WFE gate must improve before detailed mount tolerances are meaningful.
First-pass optimization
Refocus sharpens the edge field, while aperture relief remains a separate throughput trade.
| Candidate | Best focus | Best aperture scale | Best edge RMS | RMS gain | Best throughput | Best 2x2 energy | Clearance-limited RMS | Focus CSV | Gate |
|---|---|---|---|---|---|---|---|---|---|
| Four-mirror folded branch | -3.0 mm | 1.1x | 691.1 um | 33% | 75% | 0% | 691.1 um | Four focus CSV | Screening only: refocus and aperture relief improve edge RMS, but 2x2-pixel detector energy still fails. |
| Five-mirror reimaging branch | -3.0 mm | 1.1x | 627.3 um | 33% | 75% | 0% | 627.3 um | Five focus CSV | Screening only: refocus and aperture relief improve edge RMS, but 2x2-pixel detector energy still fails. |
| Six-mirror dual-triplet branch | -3.0 mm | 1.0x | 1188.9 um | 25% | 64% | 0% | 1277.9 um | Six focus CSV | Screening only: refocus and aperture relief improve edge RMS, but 2x2-pixel detector energy still fails. |
| Seven-mirror cold-stop relay branch | -3.0 mm | 1.0x | 3060.9 um | 12% | 64% | 0% | 3316.4 um | Seven focus CSV | Screening only: refocus and aperture relief improve edge RMS, but 2x2-pixel detector energy still fails. |
The sweep is deliberately bounded: four aperture scales and nine sensor offsets are tested against the existing branch surfaces. It is a handoff screen for prescription optimization, not a final optical optimization.
Bounded prescription step
Small curvature-group changes have measured edge-field consequences.
| Candidate | Aperture scale | Fore radius | Relay radius | Focus | Edge RMS | RMS gain | Edge throughput | 16x16 energy | Max field RMS | Prescription CSV | Gate |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Four-mirror folded branch | 1.1x | 1.0x | 0.9x | -4.0 mm | 563.0 um | 45% | 75% | 11% | 1113.2 um | Four prescription CSV | Screening only: curvature, aperture relief, and focus improve edge RMS, but 2x2-pixel detector energy still fails. |
| Five-mirror reimaging branch | 1.1x | 1.0x | 1.1x | -4.0 mm | - um | - | 75% | - | 456.4 um | Five prescription CSV | Screening only: curvature, aperture relief, and focus improve edge RMS, but 2x2-pixel detector energy still fails. |
| Six-mirror dual-triplet branch | 1.1x | 1.0x | 1.1x | -4.0 mm | 1091.1 um | 31% | 75% | 1% | 1211.9 um | Six prescription CSV | Screening only: curvature, aperture relief, and focus improve edge RMS, but 2x2-pixel detector energy still fails. |
| Seven-mirror cold-stop relay branch | 1.1x | 1.0x | 1.1x | -4.0 mm | 3114.7 um | 11% | 75% | 0% | 3841.2 um | Seven prescription CSV | Screening only: curvature, aperture relief, and focus improve edge RMS, but 2x2-pixel detector energy still fails. |
This is not a full optimizer. It is a controlled handoff screen: two aperture scales, three fore-group radius scales, three relay radius scales, and two focus offsets per branch. The best rows reduce edge RMS, but 2x2-pixel detector energy remains below the gate for every measured family.
Surface branch
Each branch is traceable to an architecture builder.
Four-mirror folded branch
telescope_explorer.src.architecture.build_4m_folded
Clear apertures: 64.0 mm, 42.0 mm, 40.0 mm, 36.0 mm
Field points: 0.00 deg, 0.75 deg, 1.50 deg
Needs maturation: detector-energy and image-spread rows still need optimization, and obstruction must be cleared before this branch can become a prescription.
Five-mirror reimaging branch
telescope_explorer.src.architecture.build_5m_reimaging
Clear apertures: 64.0 mm, 42.0 mm, 40.0 mm, 36.0 mm, 34.0 mm
Field points: 0.00 deg, 0.75 deg, 1.50 deg
Needs maturation: detector-energy and image-spread rows still need optimization, and obstruction must be cleared before this branch can become a prescription.
Six-mirror dual-triplet branch
telescope_explorer.src.architecture.build_6m_dual_triplet
Clear apertures: 64.0 mm, 42.0 mm, 40.0 mm, 36.0 mm, 34.0 mm, 32.0 mm
Field points: 0.00 deg, 0.75 deg, 1.50 deg
Needs maturation: detector-energy and image-spread rows still need optimization, and obstruction must be cleared before this branch can become a prescription.
Seven-mirror cold-stop relay branch
telescope_explorer.src.architecture.build_7m_cold_stop_relay
Clear apertures: 64.0 mm, 42.0 mm, 40.0 mm, 36.0 mm, 34.0 mm, 32.0 mm, 30.0 mm
Field points: 0.00 deg, 0.75 deg, 1.50 deg
Needs maturation: detector-energy and image-spread rows still need optimization, and obstruction must be cleared before this branch can become a prescription.
Where this led
The screening handed off to a full prescription optimizer, which closed the four-mirror branch.
This checkpoint series moved four-, five-, six-, and seven-mirror branches from literature leads to screened candidates with tolerance, clearance, focus, bounded curvature-step, and beam-obstruction gates. The obstruction gate ended the series: none of the seed layouts had an unblocked beam path, so bounded tweaks of these surfaces could never become fabrication candidates.
A full prescription optimizer (all curvatures, conics, fold angles, spacings, and focus free) then produced an unobscured conic-only four-mirror design that fits the payload envelope and reaches the diffraction scale on axis. That design, not anything on this page, is the current reference; see the optics hub for its readout. The five-, six-, and seven-mirror families remain unexplored under the obstruction gate and would need the same treatment to re-enter the trade.