4-mirror
plate_scale_efl, no_vignetting, physical_clearance_model, detector_plane_rms, flat_image, package_span.
RFS-064 measured result - 2026-07-27
None of the bounded four- through seven-mirror families clears the 100 mm F/1.8 gates.
None of the four bounded optimized families passes the common gates. The four-mirror row remains closest on plate scale and is the only row with non-zero minimum acceptance throughput. The seven-mirror cold-stop-relay row measures 6.0 mm EFL, 0.0% minimum throughput, and a 226.3 mm package span; the four-mirror row remains at 94.1 mm EFL and 11.8% minimum throughput.
Identical target and optimizer path
The seven-mirror relay misses plate scale, throughput, and package gates.
| Architecture | EFL mm | EFL error | Min throughput | RMS range um | Focus spread um | Package mm | Mass kg | Verdict |
|---|---|---|---|---|---|---|---|---|
| 4-mirror | 94.1 | 5.9% | 11.8% | 1060.3-6027.7 | 26686.6 | 170.5 | 0.406 | fail |
| 5-mirror | 66.6 | 33.4% | 0.0% | 3659.5-11298.8 | - | 195.7 | 0.452 | fail |
| 6-mirror | 36.5 | 63.5% | 0.0% | 11247.8-18086.9 | - | 137.5 | 0.482 | fail |
| 7-mirror | 6.0 | 94.0% | 0.0% | 9863.4-13031.9 | - | 226.3 | 0.419 | fail |
5-mirror
plate_scale_efl, no_vignetting, physical_clearance_model, detector_plane_metrics_reported, detector_plane_rms, chief_ray_angle_reporting, package_span.
6-mirror
plate_scale_efl, no_vignetting, physical_clearance_model, detector_plane_metrics_reported, detector_plane_rms, chief_ray_angle_reporting, flat_image.
7-mirror
plate_scale_efl, no_vignetting, physical_clearance_model, detector_plane_metrics_reported, detector_plane_rms, chief_ray_angle_reporting, package_span.
Method boundary
A measured family result, not an impossibility proof.
All four rows used the same obstruction-aware differential-evolution path and reserved 17-ray sampling at 13 fields. Configured budget units were 400 geometry, 400 freeform, and 200 final polish. Actual evaluation counts differ with vector dimension and are retained per architecture in the JSON.
This is a deterministic negative result for the recorded start families and budgets, not proof that every possible four-, five-, six-, or seven-mirror architecture is infeasible.
Next queued work: RFS-065 hard-gated P2 shortlist from the measured same-target candidates.
Four-mirror metrics / Five-mirror metrics / Six-mirror metrics / Seven-mirror metrics
Four-mirror acceptance / Five-mirror acceptance / Six-mirror acceptance / Seven-mirror acceptance