Tutorials, characterization procedures, and methodology references for photonic engineering.
When to use a wavemeter and when to use an optical spectrum analyzer to measure a laser: what each reports, accuracy and resolution compared in picometers and megahertz, single-mode and multimode sources, side-mode suppression, tuning and mode hops, with worked numbers at 1550 nm.
How a photonic chip gets made in a shared foundry run: what a process design kit provides, how a multi-project wafer (MPW) shuttle is organized, the steps from design to tape-out to delivered dies, what to put on a first chip, and the common reasons first silicon disappoints.
The steps in designing a silicon photonic chip, in order: choosing the platform and waveguide, simulating components, building the circuit from compact models, designing for variation and temperature, layout and verification, tape-out, and testing first silicon, with the site's pages for each step and worked numbers for a ring resonator.
How a Raman spectroscopy setup is built and calibrated: choosing the excitation wavelength (532, 785 or 1064 nm), laser and filter requirements, collection optics, the spectrometer's resolution and coverage with a worked example, detectors, wavenumber calibration with silicon and lamp lines, and common problems such as fluorescence and sample heating.
Technical questions commonly asked in interviews for photonics, optical and laser engineering roles, grouped by fundamentals, fibers, lasers, detectors and noise, integrated photonics and lab practice, each with a short worked answer, the numbers to quote and links to the full explanation.
What each line on a lens, mirror, window or filter specification means and how to judge whether it is good enough: dimensions and tolerances, material, focal length, surface flatness and irregularity, transmitted and reflected wavefront, surface quality, roughness, wedge and centration, clear aperture, coatings and damage threshold, with worked numbers.
Procedure for measuring the noise equivalent power (NEP) and specific detectivity (D*) of a photodetector: measuring the dark noise spectrum, removing the amplifier's contribution, measuring responsivity, normalizing to bandwidth and area, with a worked InGaAs example in which the feedback resistor, not the photodiode, sets the NEP.
Procedure for checking an optical power meter between calibrations: zeroing, comparison with a reference meter, the wavelength setting, linearity by the superposition method, fiber connector repeatability, and a simple uncertainty budget, with worked numbers in percent and dB.
How to choose a laser power or energy meter: photodiode, thermopile and pyroelectric sensors compared, with power ranges, wavelength coverage, speed, aperture and damage limits, worked examples for wavelength-setting errors, overfilled apertures and pulsed beams, and a checklist.
The path a bit takes through an optical link, from the host's SerDes through FEC, the modulation format, the driver and the laser or modulator, the fiber, the photodiode and transimpedance amplifier, to clock recovery and decoding, with the numbers for a 100 Gb/s PAM4 lane: 53.125 GBd, 18.8 ps symbols, the RS(544,514) FEC and the 9.5 dB PAM4 penalty.
How silicon photonic wafers are tested optically before dicing: the optical probe station, fiber probes over grating couplers, per-die alignment and its share of test time, reference structures, wafer maps of loss and resonance, probe calibration, and the failure modes of an automated flow.
Reference to the main photodetector types and how to choose between them: silicon, InGaAs and germanium photodiodes, PIN and UTC structures, avalanche photodiodes, SPADs, superconducting nanowire detectors, photomultipliers, thermal detectors and cameras, compared by wavelength range, sensitivity, speed and gain.
Reference to the optical fiber types in use: the OM1 to OM5 multimode grades with their modal bandwidths, Ethernet reaches and jacket colors, the ITU-T single-mode fiber families from G.652 to G.657, typical parameters of standard single-mode fiber, specialty fibers, and how to choose between them.
Reference to the main types of laser, grouped by gain medium: gas lasers (HeNe, CO₂, argon-ion, excimer), solid-state lasers (Nd:YAG, Yb:YAG, Ti:sapphire), fiber lasers, semiconductor lasers (Fabry-Perot, DFB, DBR, VCSEL, external-cavity, quantum cascade), dye lasers and parametric sources, with their wavelengths, how they operate in time, and what each is used for.
What nonlinear optics is and where it matters in practice: the nonlinear polarization, second-order effects (harmonic generation, sum and difference frequencies, parametric amplification, the Pockels effect), phase matching and quasi-phase matching, third-order effects (Kerr, self-phase modulation, four-wave mixing, Raman, Brillouin), with worked numbers for PPLN and optical fiber.
Procedures and choices for measuring femtosecond and picosecond laser pulses: the intensity autocorrelator and its calibration, the deconvolution factor and what it hides, checking against the spectrum with the time-bandwidth product, interferometric autocorrelation, and when FROG, SPIDER or d-scan is needed.
Procedure for measuring PER of polarization-maintaining fiber and components, and PDL of fiber-pigtailed devices: why a single rotating-polarizer reading can hide a misaligned launch, the heat or stretch method, and the all-states and Mueller four-state methods for PDL with a worked example.
Procedure for measuring the insertion loss and return loss of fiber-pigtailed and connectorized components: reference methods and which connections they include, launch conditions, the continuous-wave reflectometer method with background subtraction, and the errors that dominate at the 0.1 dB and 50 dB level.
How pulsed time-of-flight and FMCW coherent lidar measure distance, with the range, resolution and Doppler equations, worked numbers, the choice of 905 nm or 1550 nm, detectors, scanning methods, and the strengths of each approach.
A datacenter module name encodes its data rate, form factor, reach class, lane count and fiber. How to decode 400G QSFP-DD DR4, 800G OSFP 2xFR4 and 1.6T DR8, and which connector and fiber each one needs.
Methods for measuring the refractive index of bulk glass and crystals, liquids, thin films and waveguides: minimum deviation with a prism, critical-angle refractometers, the prism coupler, spectroscopic ellipsometry, and interference fringes, with the accuracy of each, worked examples, and the difference between phase and group index.
Procedure for measuring the frequency response and 3 dB bandwidth of a photodiode or optical receiver: the calibrated-modulator network-analyzer method, the optical heterodyne method with its expected RF power, the impulse-response method, de-embedding the RF path, and the bias and photocurrent dependences that change the answer.
Procedure for measuring optical signal-to-noise ratio with an optical spectrum analyzer: the 0.1 nm reference bandwidth, the IEC interpolation method and its bandwidth corrections, integrating wide coherent channels, and the in-band methods needed when filtering or dense spacing hides the noise floor.
Procedure for measuring the small-signal electro-optic bandwidth of Mach-Zehnder and phase modulators with a network analyzer: the setup and calibration, bias and drive level, removing the photodiode's response, the electrical and optical 3 dB conventions, and the velocity-mismatch and electrode-loss limits that set the result.
Procedures for measuring a lens's focal length on the optical bench: the distant-object and collimated-laser methods for back focal length, autocollimation, the Bessel displacement method that needs no principal-plane location, the magnification method, and the errors that limit each.
Procedures for measuring the attenuation of optical fiber in dB/km by the cutback method and by OTDR, the spectral attenuation curve, and macrobend loss by mandrel wrapping, with launch conditions, a worked cutback example, and the errors that dominate each measurement.
Procedure for measuring the far-field divergence of a laser beam: the two-point method and why it underestimates inside the Rayleigh range, the focal-plane lens method that works on any bench, width definitions, half versus full angle, and a worked example.
Procedure for collimating the output of a fiber or laser diode: what a collimated Gaussian beam can and cannot do, choosing the focal length from the beam size wanted, how far away the waist can be placed, setting the lens distance, and checking collimation with distance measurements or a shear plate.
Procedure for coupling a collimated laser beam into single-mode fiber: choosing the lens focal length to match the mode field diameter, the 1 dB tolerances in offset, angle and focus with a worked example at 1550 nm, beam walking with two mirrors, back-propagation, APC fiber, and verifying the efficiency.
How to focus a laser beam to a chosen spot: the focused spot size from wavelength, focal length, input beam diameter and M², the trade against depth of focus, choosing and orienting the lens, peak irradiance and fluence at the focus, damage thresholds, and how to measure the spot.
How fiber, CO₂ and Nd:YAG lasers compare for materials processing: wavelength and absorption, focused spot size and depth of focus, beam delivery, efficiency, and which laser suits metal cutting, welding, marking, wood, acrylic, glass and copper, with worked numbers.
How light gets between optical fiber and a silicon photonic chip: the mode-size problem in numbers, edge couplers, grating couplers and their alternatives, the active alignment loop, fiber-array roll alignment, and how to measure and de-embed coupling loss. A guide to the site's detailed procedures.
Reference to the optical fiber connector types in use: LC, SC, FC, ST, MU and E2000 single-fiber connectors, MPO multi-fiber connectors and their polarity, the very small form factor SN, MDC and CS connectors, ferrule sizes, polish and color codes, typical losses, and which connector suits which job.
Procedure for inspecting and cleaning fiber connector end faces before every mate: what a particle on the core costs in insertion and return loss, choosing an inspection tip, the dry and wet-dry cleaning sequence, bulkheads and transceiver ports, MPO connectors, and when to stop cleaning and replace.
How to choose between the main electromagnetic simulation methods for photonic devices: eigenmode solvers, eigenmode expansion, the beam propagation method, FDTD, and circuit-level models, with what each assumes, what it costs, typical meshes and run sizes, and which to use for waveguides, couplers, tapers, gratings, rings and circuits.
Procedure for attaching a V-groove fiber array unit to a photonic integrated circuit: array specifications and their tolerances, the alignment arithmetic for edge and grating couplers, choosing and curing the epoxy, the loss budget per channel, and the Telcordia tests the joint has to survive.
How far a ring resonance moves per kelvin on silicon, silicon nitride, silica, and lithium niobate, computed from the thermo-optic coefficient and group index; what a heater costs per free spectral range and why undercut buys efficiency at the price of speed; thermal bistability under high circulating power; and the locking methods that keep a ring on its channel, with the numbers that decide between them.
The arithmetic behind a laser diode lifetime specification: the Arrhenius temperature acceleration and the current acceleration exponent, worked with realistic activation energies; how a few thousand hours of stress on a few dozen devices becomes a million-hour mean time to failure and what confidence that number carries; the failure criteria that define a failure; what burn-in catches and does not; and how to derate a laser from the claim to an operating point.
How the front and back facet reflectivities of a Fabry-Perot laser diode set its mirror loss, threshold gain, slope efficiency, and the split of power between the two facets, worked through with computed numbers for uncoated, 10/90, 5/95, and 2/95 pairs on a 1 mm pump laser and a 300 µm telecom chip, and what the choice costs in facet intensity, feedback sensitivity, and monitor photodiode signal.
What a kink in a laser diode's light-current curve is, the lateral-mode and beam-steering mechanism that produces most of them in ridge lasers, how mode hops, polarization switches, and facet damage look different, how to read the signature, and how kink-free power is measured and specified.
Cutback, Fabry-Perot fringe contrast, and ring-resonator Q extraction measure the same propagation loss with different structures, different dominant uncertainties, and different failure modes. Worked numbers for each, a decision rule for which to use at what loss level, and what to expect when two of them disagree.
Why a spectrophotometer cannot verify an OD 6 filter, why the light source has to be spectrally cleaner than the filter under test, and a laser-and-detector procedure with an error budget that reaches OD 7: source clean-up, detector linearity by a calibrated attenuator chain, baffling, the stacked-filter floor test, and how to report the result.
Where the common-mode rejection ratio of a balanced photodetector comes from (responsivity mismatch, path-delay mismatch, bandwidth mismatch), the formula that gives CMRR as a function of frequency with worked numbers, what a given CMRR buys against local-oscillator intensity noise, and a step-by-step measurement and trimming procedure.
An honest assessment of optical computing versus optical interconnect: why matrix multiplication in a photonic mesh looks free but is not once conversion and precision are counted, why the same companies pivoted from optical compute to optical interconnect, and where compute by light still fits.
A function-by-function account of the digital signal processing behind coherent optical links: chromatic dispersion compensation as a static filter, adaptive polarization demultiplexing, carrier frequency and phase recovery, timing, and FEC, and why shedding parts of this chain is what coherent-lite trades for power.
How an optical circuit switch (OCS) works, from the MEMS mirror array that steers each beam to the insertion loss it adds to the link budget; the millisecond-reconfiguration-versus-packet-switching tradeoff it forces; and the traffic patterns of AI training fabrics that make a transparent, protocol-agnostic switch pay off.
How a free-space optical link is accounted, and why it differs from fiber: geometric beam-spreading loss that grows with the square of range, the pointing problem, atmospheric attenuation and scintillation, and worked budgets for a terrestrial link, an inter-satellite link, and a deep-space downlink.
Procedure for measuring TDECQ on 100G and 200G per lane PAM4 optical transmitters: the reference receiver chain, capture requirements, pattern and clock recovery setup, the histogram-and-noise computation, companion metrics (OMA, ER, RLM), and the bench failure modes that corrupt the number.
At 1.6T a DSP pluggable runs 13.8 pJ/bit, LPO 7.9 and CPO 6.4. What LRO, LPO and CPO each remove, and what they cost in interoperability and serviceability.
How nested antiresonant nodeless fiber (NANF, DNANF) reached 0.091 dB/km, below solid silica's 0.14 dB/km floor. An air core also cuts latency and nonlinearity.
Why co-packaged optics keeps its lasers outside the package: the thermal, reliability, and yield arguments, the ELSFP and CW-WDM interface standards, the laser-count arithmetic of split-and-modulate architectures, quantum dot combs against Kerr microcombs, and a worked power budget from wall plug to modulator.
Procedure for measuring the half-wave voltage and EO bandwidth of Mach-Zehnder and phase modulators: DC transfer-curve fitting, the Bessel-null method, small-signal quadrature slope, and swept-frequency S21, with the dominant error sources and a worked example.
Procedure for separating fiber-to-chip coupler loss from device loss using back-to-back reference structures: loopback measurement, per-coupler spectra, spectral subtraction, alignment repeatability protocol, and Fabry-Perot ripple diagnosis, with worked numbers.
The three ways to split wavelengths: cascaded thin-film filters, arrayed waveguide gratings, and fiber Bragg gratings with circulators. Compared on loss, channel count, isolation, temperature behavior, and cost.
dn/dT of silicon nitride (2.45 × 10⁻⁵ K⁻¹ for LPCVD Si₃N₄ at 1550 nm) against silicon, silica, InP, GaAs, lithium niobate, and polymers, with sources; the measured ring-resonance drift and heater efficiency that follow from each; a silicon nitride versus silicon platform table; and the film, wavelength, and temperature caveats.
The three amplifier technologies compared on the numbers that decide between them: gain, noise figure, saturation power, bandwidth, and dynamics. With the use cases where each is the only sensible answer.
How much optical feedback a laser diode tolerates: the five Tkach–Chraplyvy regimes in dB, coherence collapse from about −39 dB, and the isolator budget.
Two bench methods for finding the actual temperature of a laser chip: the DFB wavelength-shift method and the forward-voltage method. Includes the pulsed-vs-CW logic, worked numbers, and thermal resistance extraction.
A working guide to the standard laser diode packages. What each is for, thermal paths, monitor and TEC contents, the wiring-variant traps, and how to choose for a bench experiment versus a product.
A working engineer's walkthrough of a laser diode datasheet: what threshold, slope efficiency, kink-free power, SMSR, RIN, and tracking error actually promise, the conditions-column traps, and the specs that are missing when a vendor hopes you won't ask.
A selection guide for the four standard linewidth measurements: DSH, heterodyne, frequency-noise discriminator, and coherent receiver. Covers the fiber-length math, what each method actually reports, and the mistakes that skew results.
The decision every PIC project makes once: edge coupling or grating coupling. Loss, bandwidth, polarization, wafer-level test, packaging, and the hybrid strategy most real programs land on.
How single-layer anti-reflection coatings work and how to design one: the quarter-wave thickness condition, the ideal-index rule, worked examples on glass and silicon, and when a single layer isn't enough.
Procedures for measuring the three primary parameters of photodetectors used in photonics (wavelength-dependent responsivity, noise-equivalent power, and 3-dB electrical bandwidth), with focus on the standard measurement configurations for InGaAs PIN photodiodes.
Complete characterization workflow for a distributed-feedback (DFB) semiconductor laser, integrating LIV, spectral, temperature-dependent, and noise measurements to extract the parameters required for telecom and sensing applications.
Reference for optical spectrum analyzers (OSAs) including diffraction grating, Fabry-Pérot, and Fourier-transform variants, resolution bandwidth and dynamic range specifications, and the standard measurement configurations for laser source characterization, OSNR measurement, and EDFA gain characterization.
Step-by-step procedure for turning a laser diode's LIV sweep into its wall-plug efficiency curve: the definition, why the peak sits at a few times threshold, the three factors that set it, a worked example with numbers, the measurement errors that bias it, and typical peak values from telecom DFBs to pump diodes.
Equipment selection, thermal mounting, electrical configuration, and noise mitigation for a benchtop laser diode characterization setup, with focus on the configuration choices that determine measurement quality.
Procedure for alignment of optical fiber to edge coupler structures on photonic integrated circuits using lensed fibers and inverse-taper couplers, with discussion of butt-coupling, anti-reflection cleaving, and the distinct failure modes compared to grating coupler alignment.
Comparison of pulsed and continuous-wave measurement modes for semiconductor laser LIV characterization, with quantification of self-heating bias, duty cycle requirements, and equipment configuration.
Comparison of the four standard methods for extracting threshold current from a semiconductor laser LIV curve, applied to a single dataset, with discussion of when each method is preferred and the systematic differences between them.
Procedure for measuring laser beam quality (M²) per ISO 11146 with a beam-profiling camera or a knife edge: second-moment (D4σ) widths and why baseline noise inflates them, the caustic scan through a focus, hyperbolic fitting, automated M² systems, and the dominant errors.
Measure a laser's relative intensity noise (RIN) in dB/Hz with a photodiode and spectrum analyzer: formula, shot-noise calibration, worked DFB example, errors.
Procedure for extracting loaded and intrinsic quality factors of integrated ring resonators from measured transmission spectra, including fit methodology, the all-pass / add-drop distinction, and resolution of the coupling regime ambiguity.
Procedure for measuring optical propagation loss in integrated photonic waveguides using the cutback method, including paired-device variants used in foundry process control, fit methodology, and uncertainty analysis.
Procedure for extracting slope efficiency and external differential quantum efficiency from laser diode LIV data, with discussion of measurement geometry, facet-collection considerations, and common extraction errors.
Procedure for extracting the characteristic temperature T₀ of a semiconductor laser diode from temperature-dependent threshold current measurements, with worked example and discussion of extraction errors.
Active alignment of a single-mode fiber to a silicon photonics grating coupler: height, 40 × 40 μm raster, polarization, fine search, expected loss, failure modes.