BBSRC · In-chamber fluorescence for Crossbeam + Katana
In-chamber fluorescence for the Crossbeam and Katana volume-EM stack. The selected optical baseline is now NA 0.55 with Firefli Red. The Rev G CAD and Blender results below still describe the earlier NA 0.42 configuration.
Updated 17 September 2026 · Optical decision: 10 September · Rev G audit: 6 September · Sensor study: 7 September
Purpose
The instrument needs a small fluorescence channel that fits the chamber envelope, uses a long-working-distance objective, and does not collide with the microtome, detectors or stage travel. The design feeds a horizontal camera arm through a dichroic cube and a 45° prism above the objective, with reciprocal LED excitation on the same shared bore.
This page distinguishes the selected optical baseline from the implemented Rev G geometry and its existing Blender simulations. Revised optics, aperture checks and chamber clearance remain to be completed before a final print release.
Latest baseline — NA 0.55 & Firefli Red
Decision dated 10 September 2026 · Added to the main repository 17 September · Not yet applied to CAD or Blender
The starting objective is now NA 0.55. The working candidate is the Mitutoyo M Plan Apo 50× with 13 mm working distance and a 4 mm focal length. Its identity and physical envelope must be confirmed before mechanical changes. With the existing ideal 100 mm tube lens, the nominal system magnification is 25×.
| Parameter | Rev G CAD / simulation | Selected baseline |
|---|---|---|
| Objective catalogue magnification | 20× | 50× candidate |
| Numerical aperture | 0.42 | 0.55 |
| Working distance | 20 mm | 13 mm |
| System magnification with ideal f100 tube | 10× | 25× |
| Object pixel size | 0.345 µm | 0.138 µm |
| Object field | 662.4 × 414 µm | 264.96 × 165.60 µm |
| Objective-front datum | Z73.58 mm | Z66.58 mm proposed |
The initial sample is Firefli Red-labelled polystyrene, with reported excitation/emission peaks at 542/612 nm. It supersedes the provisional Nile Red channel for initial testing. Fluorescence has been reported after preparation, with high background; quantitative performance is still unverified.
Provisional spectral targets are a green LED around 530–545 nm, excitation around 540/20 nm, a 565–570 nm dichroic transition, and emission around 610/40 nm. These are targets for evaluation, not a purchased or validated filter set. The former 505 nm long-pass dichroic is unsuitable for routing green excitation in this configuration.
The shorter working distance moves the objective front 7 mm closer to the sample. Confirm the actual objective envelope, lid and knife clearance, focus assumptions and illumination geometry; then rerun the aperture audit. The old simulations do not validate this baseline.
New sample-image review
The supplied EVOS image combines red-pseudocoloured RFP fluorescence and transmitted light, acquired with a 20× / NA 0.45 objective. Visible red signal supports testing the selected NA 0.55 objective, but the merged 8-bit RGB image cannot establish fluorescence background, photon SNR or detection limits.
Next measurements need separate original fluorescence and transmitted-light frames, an identically prepared blank, the installed filter-cube identity and particle size. Compare contrast above measured background before choosing the final spectral components.
Layout
Mechanicals — existing Rev G
Rev F vignetted off-axis image and illumination bundles at the shared/image bore, illumination approach and dichroic clear aperture. Rev G enlarges those openings and shifts the tube-lens / camera and LED seats so the sampled pupil rays clear. The 6 September audit records no solid–solid interferences among the 34 assembled parts after the correction; the nine-part print layout regenerated cleanly. These are recorded checks, not a fresh validation of the live CAD.
| Item | Rev F | Rev G |
|---|---|---|
| Shared / image bore | Ø23 mm | Ø40 mm |
| Illumination approach bore | Ø24 mm | Ø40 mm |
| Dichroic clear rectangle | 32 × 22 mm | 46 × 32 mm |
| Dichroic glass envelope | 35.6 × 25.2 × 1.1 | 50.8 × 35.6 × 1.1 |
| Tube-lens / filter envelopes | Ø25–30 mm | Provisional Ø50.8 mm |
| Field stop | Ø1.8 mm | Ø3.6 mm @ f40 conjugate |
Existing Rev G 3D model & light-path
This model still uses NA 0.42 / f10 / WD20. It has not been updated for the selected NA 0.55 objective or Firefli Red channel.
Interactive Rev G optical audit mesh (assets/rev_g_optical.glb). Rays were traced in Blender through specular prism reflection, planar-window and dichroic refraction (Snell), finite aperture tests and ideal thin lenses. Excitation is reciprocal from the objective pupil toward the field stop. An independent mesh ray-cast against dimension-matched aperture sections confirmed zero collisions for Rev G on the sampled set. Orange = image, blue = excitation, red = failed (none in this slim export).
Model from assets/rev_g_optical.glb; optional rays in mm. Geometric clearance only.
Blender simulations — Rev G sensor view
The newer Blender scene places the active camera at the designed sensor position and uses an LED-driven synthetic fluorescent specimen. It shows ideal image formation, the effect of moving the specimen out of focus, and an LED-off control. These are NA 0.42 / nominal 10× results. They have not been regenerated for the selected NA 0.55 / nominal 25× baseline; this study did not change CAD dimensions.
- Focus: a 20 µm reference dot measures 58 pixels (20.01 µm), checking the implemented image scale.
- Defocus: raising the sample 10 µm spreads the signal and reduces the smallest test-dot peak from about 0.580 to 0.0116 in relative linear units.
- LED off: all pixels are zero in the model’s dark control. Real detector noise is not included.
- Illumination: the assumed LED and diffuser produce a calculated relative irradiance range of 0.922–1.024 across the sample grid, normalised to the centre.
The simulation uses ideal optical transfer and assumed LED/diffuser properties. It does not predict absolute brightness or detection limits, and does not include diffraction, real-lens aberrations, spectral response or camera noise. The existing aperture-clearance audit is a separate check.
Rev G design & simulation files
Onshape and GitHub links require an account with project access. Open the sensor scene in Blender 4.5 or later and press F12; the linked study explains the LED and focus controls.
Caveats
This is a geometric clearance result, not a validation of image quality, LED radiance, field uniformity, photon budget, spectral filters, ghosts, MTF, print tolerance, thermal drift or vacuum compatibility. Ø50.8 lens and filter solids are provisional envelopes — not purchased parts. Pupil planes used in the sensitivity sweep are assumptions; Mitutoyo’s proprietary prescription is unavailable. Do not treat ideal thin-lens focus as evidence of real-lens performance. Real tube-lens BFD, relay prescription and SEM/vacuum fit remain open before final print release.