For an arc lamp or quartz tungsten halogen (QTH) source, collecting a large amount of light is only the first task. The collected radiation must also fit the downstream beam path, sample, spectrometer, fiber and detector. This tutorial uses geometrical optics for incoherent sources. Diffraction and coherent-beam effects are outside this model.
Start with the complete system
Trace the source, collector, filters or beam splitters, delivery optics and detector before optimizing a condenser. A collector can gather more flux while producing a beam that the next aperture or slit rejects. Useful throughput therefore depends on both position and angle, and on the spectral transmission of every component.
F-number and numerical aperture
A lens of clear diameter D and focal length f, with the source at its front focus, is commonly described by F/# ≈ f/D. This is the paraxial expression. For a collection cone of half-angle θ in refractive index n, the aperture convention used here is:
NA = n sin θ; F/# = 1/(2n sin θ) = 1/(2NA).
At high angles the cone definition is more useful than treating f/D as exact. The source regards the simple small-angle approximation as appropriate up to roughly 15°. In air NA cannot exceed 1; an NA of 0.95 is an example of a high-aperture microscope objective, rather than a typical condenser specification.
For a fixed source and otherwise comparable paraxial geometry, collected flux scales approximately as φc ∝ 1/(F/#)2. A smaller F-number is a faster collector. An iris changes the effective diameter and hence the collection. For a rectangular or square aperture, an equal-area circular diameter is more representative of collection than the diagonal: Deq = 2√(A/π).
These ideas also apply to mirrors. With an ellipsoidal collector, however, a quoted F-number may use the distance from its aperture to an external focus rather than the focal length convention of a collimating lens. Always establish what distances a stated F-number represents.
Collected light and useful light
A fast singlet can capture a large cone but introduce enough aberration that much of the output cannot be refocused into the required image. Later optics cannot recover an ideal image from a beam whose rays already have the wrong position–angle relationships. A slower collector or an aberration-corrected design may deliver more useful light through a small slit.
Historical design guidelines put a spherical singlet near F/4 for demanding imaging and around F/2 to F/1.5 for some arc-lamp condenser tasks. They are application guidelines, not hard limits. Correct lens orientation matters. Faster corrected systems generally require more elements or aspheric surfaces; For scale, the historical reference describes five or six components in F/1 camera lenses, four in a commercial F/0.7 aspheric collector, and ten or more in some microscope objectives approaching F/0.5. These are examples of design complexity and restricted field, rather than universal element-count requirements.
Real condensers: spectrum, temperature and source size
Spectral transmission
The lens material sets part of the usable wavelength band. In an infrared system, germanium can transmit the desired infrared while absorbing visible light from a broad-spectrum source. In a UV system, synthetic UV-grade fused silica can be important because the ultraviolet fraction of the source is limited. Material origin and cumulative short-wavelength exposure affect transmission through solarization. Crystalline quartz and amorphous fused silica should not be treated as identical materials.
Thermal loading
The first element of a fast condenser is close to the lamp and absorbs some incident radiation. Temperature changes alter refractive index and focal length, while thermal gradients and startup shock can fracture the optic. Material selection, mounting and cooling must be considered together. The historical source describes failure of a borosilicate crown element near 1000 or 1600 W lamps and use of a cooled fused-silica element instead. This illustrates a thermal-design issue; it is not a universal failure threshold for those lamp powers.
Finite-source collimation
A real filament or arc has finite dimensions. Each source point produces a differently directed beam, even with an ideal condenser. In the source’s 1 kW QTH example, a filament is 6 mm in diameter and 16 mm long. With a 50 mm focal-length condenser, the long dimension gives a half-angle scale of (16/2)/50 = 0.16 rad, about 9° or 160 mrad. “Collimated” output from this source therefore still contains a substantial angular distribution.
Aberrations and the collector position
Spherical aberration causes marginal rays of a singlet to behave differently from paraxial rays. A position that collimates the central cone may leave the outer rays converging. Moving the condenser slightly toward the source can provide a useful compromise, but the best setting depends on the source and the downstream acceptance.
Chromatic aberration follows from wavelength-dependent refractive index. A single setting cannot precisely collimate every wavelength of a broadband source. The source quotes visible-to-near-infrared index variations of about 1.5% for borosilicate crown and 3% for fused silica, with faster changes in the UV; these are illustrative broad-band figures without fixed endpoints. Shorter focal length at shorter wavelength can require moving the lens closer to the source, which also changes the collected cone.
One historical F/1.5 fused-silica example gives a marginal-ray divergence of 0.33° at 300 nm when spherical aberration is excluded. That figure depends on the original prescription and focus reference; it is not a general value for all F/1.5 lenses. Adjust focus for the spectral region that matters to the experiment.
Visible achromats can reduce color error but often contain cement unsuitable for intense UV or IR exposure. One approach is to collect with a suitable primary condenser, remove unwanted UV and IR, and form a well-defined secondary source using a lens and pinhole. A downstream achromat can then image this filtered secondary source. Low-intensity or distant visible sources may be handled directly by a suitable achromat.
Source directionality changes the collection estimate
Lamp radiance is generally not uniform in all directions. Positioning the collector toward the brightest useful direction can matter as much as its nominal F-number. For example, a historical Oriel 6333 100 W QTH lamp was described as emitting about 3600 lumens. Uniform emission would give about 94 lumens, or 2.6%, into the example F/1.5 collector; the stated directional result was about 140 lumens. These values illustrate why total lamp lumens alone do not establish collected output.
Reflective alternatives
| Reflector | Ideal action | Limitation to consider |
|---|---|---|
| Spherical concave | Collects and focuses light in a simple reflective arrangement. | Spherical aberration, finite field and thermal loading remain relevant. |
| Paraboloidal | Collimates radiation from its focus, or focuses an incoming axial collimated beam. | Extended or displaced sources introduce aberrations. |
| Off-axis paraboloidal | Provides access to a displaced focus using a segment of a paraboloid. | Avoids central source/detector obstruction, but still requires controlled source extent and alignment. |
| Ellipsoidal | Transfers rays from one focus to the other and can surround a source over a large collection angle. | Points away from the exact first focus are magnified and defocused; high collection does not imply a small sharp image. |
Reflection eliminates refractive chromatic focal shift, but the coating still has a wavelength-dependent reflectance. Mirrors can therefore simplify broadband focusing without providing wavelength-independent transmission or immunity to heat.
Why F-number matching is insufficient
The same lens at three different source distances has the same f/D but collects different solid angles. A coupling design should describe both the object-side collection and image-side acceptance, with the actual source and image sizes. Ratios of object or image distance to aperture may be useful geometric descriptions, but should not be confused with an intrinsic lens F-number.
Suppose a fiber feeds a monochromator. Increasing the output F-number to match the monochromator’s accepted cone also increases the image size if the optical invariant is preserved. More light may then miss the narrow entrance slit. Angle matching by itself cannot optimize the transmitted power.
Geometrical extent, optical extent and radiance
For an ideal paraxial system in air, with no internal stop clipping the accepted bundle, the source and image satisfy:
AsΩs = AiΩi = G.
A is the relevant source or image area and Ω its accepted solid angle. G is geometrical extent, often called étendue or throughput. Between media of different index, the conserved optical extent includes n2AΩ. At large angles the general definition includes the directional projection factor; the simple AΩ expression here is a paraxial model.
If the accepted source has uniform radiance L and the total optical transmission is t, then φ = tGL for the ideal accepted bundle. Losses, aberrations and clipping make φ ≤ tGL the practical bound. Transmission includes mirror reflection losses as well as transmissive elements.
For one transverse plane, the corresponding paraxial invariant is n1h1θ1 = n2h2θ2. In the same medium it reduces to h1θ1 = h2θ2. Sine-angle forms are used beyond the small-angle approximation. A slit instrument must be considered in both width and height planes because the two accepted bundles can differ greatly.
Find the bottleneck before enlarging the collector
Every real system has a limiting accepted extent, Glim. It may come from a spectrometer slit and acceptance angle, a fiber core and NA, or detector area and angular acceptance. Detector area can also affect noise, so increasing it is not automatically beneficial to measurement quality.
The useful upper-bound expressions become φ = tGlimL and, spectrally, φλ = tλGlim,λLλ. Extent may be approximately wavelength independent in a given geometry, while source radiance and transmission are not.
- Identify the limiting component and increase its accepted extent if the application permits.
- Choose collection optics that at least fill that accepted extent. A smaller collection extent loses useful power.
- Avoid collecting far beyond what the system accepts unless there is another reason to do so. The surplus can heat apertures or contribute stray light.
- For fixed accepted extent, increase the source radiance in the required spectral band. Source area and collection angle can be traded, but radiance cannot be created by passive optics.
Historical collector examples quote 1.17 sr for F/0.7 and 0.66 sr for F/1. These are values for the original collection configurations; they should not be treated as universal exact conversions between F-number and solid angle.
Spectral radiance comparison
The following historical reference values are averages over nominal source areas, in W cm−2 sr−1 nm−1. Local bright regions can exceed those averages. Source power labels for lamps refer to electrical input, while the HeNe example refers to optical output.
| Source | Nominal dimensions or condition | 300 nm | 400 nm | 500 nm | 633 nm |
|---|---|---|---|---|---|
| Oriel 6333 QTH, 100 W input | 4.2 × 2.3 mm | 0.0015 | 0.013 | 0.034 | 0.065 |
| Oriel 6253 xenon arc | 0.5 × 2.2 mm; input power not given in the table | 0.085 | 0.16 | 0.21 | 0.24 |
| 3400 K blackbody | Temperature example | 0.0037 | 0.03 | 0.08 | 0.15 |
| Sun | Extraterrestrial reference; source table lists 1.4 × 1012 mm nominal diameter | 0.75 | 2.1 | 2.9 | 2.2 |
| HeNe, 2 mW optical output | 0.63 mm beam diameter; 4 × 10−3 nm linewidth assumed | 0 | 0 | 0 | 108 |
The laser row illustrates spectral concentration and depends strongly on the assumed linewidth. It does not extend the incoherent ray model to laser diffraction. Zero entries indicate no line contribution in this simplified comparison, rather than a measured detector background.
Worked example: source magnification
A condenser of focal length f1 collimates light from a source, and a secondary lens of focal length f2 reimages it. In the small-angle model, h2 = h1θ1/θ2 = h1f2/f1. The intervening nominally collimated path is convenient for filters and beam splitters, but does not remove the source-size constraint.
Worked example: filter versus monochromator
Consider isolating a 20 nm band centered at 365 nm from an extended source. The historical comparison uses an Oriel 56430 narrow-band filter and an Oriel 77250 1/8 m monochromator with a 1200 lines/mm grating. These are third-party reference examples.
| Quantity | Filter | Monochromator |
|---|---|---|
| Acceptance half-angle | About 15° | About 7.7° |
| Accepted solid angle | About 0.21 sr | About 0.057 sr |
| Accepted area | 22.9 mm diameter, about 4.1 cm² | 3.16 mm slit width × 12 mm height, about 0.38 cm² |
| Extent G | 0.86 sr cm² | 0.022 sr cm² |
| Transmission-weighted extent GT | 0.17 sr cm², using T = 0.2 | 0.007 sr cm², as given for this configuration |
For this recalculated educational comparison, the solid angle of a circular cone is Ω = 2π(1 − cos θ). It gives approximately 0.057 sr at 7.7°; the historical text’s 0.57 sr is a decimal error, inconsistent with its own stated extent. The comparison therefore retains the internally consistent 0.057 sr and G ≈ 0.022 sr cm².
The transmission-weighted ratio is 0.17/0.007 ≈ 24, so the filter can pass about 24 times as much light for this extended-source example, provided the detector accepts the emerging bundle. It does not establish the same advantage for a small asymmetric arc or for every desired spectral resolution. Slit size, bandwidth, source geometry and downstream acceptance must all be included. Even monochromators with similar F-numbers can have different useful extent at the same resolution if their usable slit areas differ.
