OSA common mistakes
OSA · Understand the traps and how to avoid them.
Computing a thin-lens or Gaussian image position and reporting it as the final answer without checking whether the marginal ray actually clears every aperture, so vignetting or a field stop is missed.
Candidates treat paraxial imaging as a complete answer because it is the first formula set learned, and the exam often supplies a stop or clear-aperture dimension that must be applied afterward.
What to do instead
After every paraxial calculation, add a ray-height check at each surface or aperture: trace the marginal ray (height = y at the stop) and the chief ray through the stop, and state explicitly which aperture limits the axial and field bundles. Report both the paraxial image location and the aperture-limited field.
Confusing numerical aperture with f-number and using the wrong one in the diffraction-limited spot size, depth of focus, or coupling efficiency formula.
Both describe cone angle, and candidates memorize one relation (NA = 1/(2·f/#) in air) without tracking when the working medium index or the half-angle convention changes.
What to do instead
Write the definition chain at the top of the scratch work: NA = n·sin θ (half-angle), f/# = f/D, and NA ≈ 1/(2·f/#) only in air at small angles. Then select the formula that matches the given quantity, and convert explicitly before substituting numbers.
Selecting a glass or coating by index alone while ignoring dispersion, thermal expansion, or the wavelength range over which the quoted index applies.
Datasheet index values are usually quoted at the sodium d-line, and candidates carry that single number into a broadband or athermal design without checking Abbe number or dn/dT.
What to do instead
For every material choice, state the design wavelength, the Abbe number or partial dispersion, and the relevant thermal coefficient. If the problem spans a band, verify achromatization with the V-number condition and check that the coating or transmission curve covers the full band, not just the center wavelength.
Specifying a mirror or window coating without checking angle of incidence and polarization, then reporting reflectance at normal incidence as the system answer.
Coatings are usually tabulated at 0° or 45°, and candidates apply the normal-incidence number to oblique or s/p-split geometries.
What to do instead
Identify the incidence angle and polarization state first. For oblique use, quote separate s- and p-reflectance (or transmittance) values and check whether the design is specified for unpolarized average. Flag any coating whose specified angular range does not include the operating angle.
Balancing aberrations by adding elements or splitting powers without first identifying which aberration dominates and which conjugate or stop condition drives it.
Candidates reach for standard remedies (splitting a lens, adding a field flattener) as recipes rather than diagnosing spherical, coma, astigmatism, field curvature, or distortion from the layout and stop position.
What to do instead
Before proposing a fix, name the dominant aberration, its field or aperture dependence, and the stop position. Then apply the matching lever (bending for coma, splitting for spherical, meniscus or field flattener for curvature, stop shift for off-axis balance) and state the trade-off introduced elsewhere.
Treating a design as complete once the spot diagram looks small, without checking tolerancing, manufacturability, or the diffraction limit at the stated aperture.
Optimization software rewards small RMS spot radius, and candidates accept that as the endpoint rather than comparing to the Airy diameter or evaluating sensitivity to decenter and tilt.
What to do instead
Close every design answer with two checks: compare RMS spot to the diffraction-limited Airy diameter at the design f/#, and state the tightest tolerance (typically decenter or element tilt) with its effect on performance. If the spot is far below the Airy disk, say the system is diffraction-limited and tolerances dominate.
Mixing up laser gain, population inversion, and threshold conditions, or applying the Schawlow-Townes or slope-efficiency relations with the wrong units.
Rate equations are dense and candidates substitute numbers before confirming whether a quantity is per-unit-length, per-pass, or a dimensionless ratio.
What to do instead
Annotate every symbol with its units before substitution (gain coefficient per meter vs. gain per pass, cavity loss per round trip). Derive threshold from round-trip gain equals round-trip loss, and keep slope efficiency in power per power with the output-coupling fraction explicit.
Ignoring the difference between radiometric and photometric quantities, or between coherent and incoherent detection, when computing signal or SNR in a sensing problem.
Watts and lumens, and shot-noise-limited versus background-limited regimes, look interchangeable on paper, and candidates apply one noise model to every detector.
What to do instead
Classify the problem first: radiometric (W, W/sr) versus photometric (lm, cd), and coherent (local-oscillator shot noise) versus direct detection (signal plus background shot noise, thermal noise). Write the dominant noise term explicitly and justify why the others are negligible before computing SNR.
Reporting a measurement result without an uncertainty budget, or quoting resolution as accuracy in an interferometric or wavefront test.
Candidates conflate the smallest readable increment with the true error and skip contributions from reference surface, environmental drift, and fringe-analysis algorithm.
What to do instead
Always separate resolution, repeatability, and accuracy. Build a short uncertainty budget naming the reference-surface error, air-turbulence or thermal drift, detector or camera noise, and analysis bias, then combine in quadrature and report the result as value ± expanded uncertainty with coverage factor stated.
Answering a laser-safety or standards question with the nominal hazard class or MPE from memory without checking wavelength, exposure duration, and whether the beam is collimated or extended-source.
Exposure limits depend strongly on wavelength band, pulse duration, and source geometry, and candidates recall a single class table rather than the parameter-dependent rule.
What to do instead
State the wavelength, exposure duration (or pulse repetition), and source geometry before quoting any class or MPE. For pulsed sources, check both the single-pulse and average-power limits and apply the applicable correction factors. Cite the controlling standard clause rather than the class label alone.