The terms on filter and photodiode datasheets, each with the relation behind it. Where a relation has only a few inputs, the equation is live: change a value and the result follows.
Passband and wavelength
Center wavelengthCWL, λ0
The wavelength midway between the two half-maximum points of a bandpass filter, at normal incidence in a collimated beam.
Here λ50 means 50 % of peak transmission, not 50 % absolute. Tilting the filter moves the CWL to shorter wavelengths, see effective index; warming it usually moves the CWL to longer ones, see temperature coefficient.
The width of a passband between the two points where transmission falls to half of its peak value.
Also called the bandwidth. The width at 10 % of peak is a different, larger number, and a −3 dB width matches the FWHM only when it is referenced to the peak rather than to 100 % transmission. In a converging beam the measured FWHM grows, see ray center spread.
The highest transmission inside the passband, measured at or near the CWL.
Peak and average transmission over the passband are different numbers: close together for a flat-top multi-cavity filter, far apart for a single cavity. Both are quoted for collimated, unpolarized light at a stated angle of incidence.
Tpeak=λmaxT(λ)
Unit: % or fractionAppears as: Tpeak, Peak T, Tmax, Transmission
How transmission falls from the peak to the blocking: flat-topped with steep sides for multi-cavity filters, Lorentzian for a single cavity.
A filter of k coupled cavities follows the model below to a good approximation, with k = 1 the Lorentzian. The passband models on this site use k = 3. A larger k gives a flatter top and steeper sides at the same FWHM.
For a longpass or shortpass filter, the wavelength at which transmission crosses 50 %.
Edge steepness is the distance in nm from the 50 % point to the point where blocking reaches a stated level, often OD 3. Some datasheets quote it as a percentage of the edge wavelength instead.
The allowed deviation of the measured center wavelength from its nominal value, at a stated angle of incidence and temperature.
A tolerance of ±0.5 nm at 0° and 23 °C makes no claim about the CWL at 10° or 60 °C. The effective index and the temperature coefficient give those, and the shifts add to the tolerance in a budget.
λnom−δ≤λ0≤λnom+δ
Unit: nmAppears as: CWL tolerance, CWL ±, Center wavelength tolerance
The shift of the center wavelength per kelvin, in pm/K. A positive value moves the passband to longer wavelengths as the filter warms.
Dense sputtered coatings shift a few pm/K; porous evaporated coatings shift more and also move with humidity. Over a working temperature range the shift adds to the CWL tolerance.
Δλ=cΔT
pm/K
Range −50 to 50 pm/K
K
Range −150 to 150 K
Unit: pm/KAppears as: Temperature coefficient, Thermal shift, dλ/dT, CWL drift
Blocking expressed as a base-10 logarithm: OD 4 transmits 10⁻⁴ of the incident light, OD 6 transmits 10⁻⁶.
Each OD unit is a factor of ten in transmission. ODs add for absorbing layers in series; for reflective blockers in series they do not, see stacked OD.
The wavelength range over which a stated OD holds, usually with an exclusion band around the passband where blocking is not specified.
Read it as a pair of ranges, for example OD ≥ 6 from 200 to 515 nm and from 545 to 1100 nm; outside both the datasheet makes no claim. The range that matters is the spectral range of the receiver behind the filter.
OD(λ)≥ODminfor λ1≤λ≤λ2
Unit: nmAppears as: Blocking range, Blocking, Rejection band
The blocking of two filters used in series. For reflective blockers it is less than the sum of the two ODs, because light bounces between them.
With transmissions T1,T2 and reflectances R1,R2, the multiple reflections add incoherently as below. Tilting one filter or adding an absorbing element breaks up the cavity and brings the result close to the sum of the ODs.
The highest OD an instrument can resolve. A quoted OD can be no higher than the dynamic range of the spectrophotometer that measured it.
Above the limit the entry should read OD > limit, for example OD > 6, with the instrument stated. OD 8 on a datasheet measured with an OD 6 instrument is a design value, not a measurement.
ODmax=log10PfloorPref
Appears as: OD > 6, Measured to OD 6, Instrument limited
A single index that describes how fast a coating's passband moves with angle. A larger value means a smaller shift.
It belongs to the coating design, commonly 1.4 to 2.2 for dielectric bandpass filters, and can be fitted to centers measured at two angles. One index holds well to about 20° to 30°; beyond that s and p need separate values.
Focal length divided by entrance aperture diameter. It sets the half angle of the converging cone that reaches a filter near a focus.
The working f-number of the beam at the filter is what counts, not the value marked on the lens for an object at infinity. Each ray in the cone meets the filter at its own angle of incidence, see cone half angle.
The range of center wavelengths seen by the rays of a cone: the center at the shallowest ray minus the center at the steepest.
For a 532 nm filter with an effective index of 2.0 tilted 10° in an f/2 beam, the spread is 11.15 nm, wider than a 3 nm passband, so the passband smears instead of shifting. When the cone straddles the normal the shallowest ray is at 0°.
The two polarization states at oblique incidence: s has the electric field perpendicular to the plane of incidence, p has it in that plane.
Away from normal incidence a filter's passband splits: the p passband shifts further and widens, the s passband shifts less and narrows. Unpolarized light sees the average of the two; at 0° the distinction vanishes.
A beam whose rays are parallel, a plane wave in the limit. Most filter datasheet values assume collimated light at a stated angle.
A beam counts as collimated for a filter when the cone shift below, with α the divergence half angle in radians, is much smaller than the CWL tolerance δ.
Tested limits: every unit, or every unit of a sampled lot, is measured against them and rejected outside them.
A limit is only as useful as its test conditions and the sampling behind it, see lot sample. A limit set by design or characterization rather than by production test should say so.
The temperature, AOI, polarization, beam, bias, frequency and load under which a value was measured. Without them a value is not reproducible.
Filter values need angle, polarization, beam divergence and temperature. Photodiode values need bias voltage, temperature, wavelength, load and bandwidth. Values measured under different conditions cannot be compared directly.
Appears as: Conditions, Test conditions, @ 25 °C, VR = 10 V
A test plan in which a sample of parts from each coating run or production lot is measured, and the lot is accepted on the result.
For filters the sample is often a witness piece coated in the same run. Sample size and acceptance rule belong in the purchase specification, next to the limits they apply to.
Appears as: Lot tested, Sample tested, Witness sample
The letter or number and date that identify one version of a datasheet. Values change between revisions; the one cited in an order applies.
A revision history lists what changed and why. Purchase specifications cite the datasheet by revision so a later change does not alter what was ordered.
The central area of an optic over which its optical specifications hold. Usually smaller than the outside size by a stated margin.
A beam that overfills the clear aperture reaches the edge zone, where coating and surface values are not specified. Quoted as a diameter or as a percentage of the outside dimension.
Unit: mmAppears as: CA, Clear aperture, Useful aperture
Scratch-dig: a pair of numbers such as 40-20 that grade cosmetic surface defects, written in the notation of MIL-PRF-13830B.
The scratch number refers to visual comparison standards, not to a width in micrometers; the dig number is the largest allowed dig diameter in hundredths of a millimeter. Lower numbers are tighter: 60-40 is common commercial quality, 20-10 precision quality.
The deviation of a wavefront passing through the optic from its ideal shape, in waves at 633 nm over the clear aperture.
Quoted peak-to-valley or RMS, which are different numbers; P-V is commonly 3 to 5 times the RMS. λ/4 P-V at 633 nm is a common imaging-grade value over the clear aperture.
The highest laser fluence or intensity a coating withstands without damage, at a stated wavelength, pulse width and spot size.
Pulsed thresholds are in J/cm² with pulse width and repetition rate; CW thresholds are in W/cm² or W/cm. For a Gaussian beam the peak fluence is twice the average, with w0 the 1/e² radius. Thresholds at other nanosecond pulse widths scale roughly with the square root of pulse width.
Photocurrent per watt of incident optical power at a stated wavelength, in A/W.
Set by the quantum efficiency and the photon energy, so it rises with wavelength until absorption falls off toward the band gap. With λ in nm, 1239.84 is hc/q in nm·V.
The fraction of incident photons that produce collected electrons, from 0 to 1 or in percent.
External quantum efficiency includes the reflection loss at the surface; internal quantum efficiency excludes it. Datasheets usually mean external. It converts to responsivity as below.
The current a photodiode passes with no light, at a stated reverse bias and temperature.
It rises with bias and roughly doubles every Td degrees, commonly 8 to 10 °C for silicon. It sets the shot noise floor of a reverse-biased photodiode.
The capacitance of the depletion region. It falls as reverse bias widens the region and sets bandwidth and noise gain with the front end.
A parallel-plate model gives the relation below, with A the junction area and W the depletion width. Quoted at a stated bias and frequency, often 1 MHz.
The slope resistance of a photodiode near zero bias, typically measured at ±10 mV. It sets the Johnson noise in photovoltaic use.
Values in the GΩ range are common for small silicon photodiodes and fall as temperature rises. It matters most at zero bias and low light, where Johnson noise from the shunt dominates.
The time for the output to go from 10 % to 90 % of its final value after a step in light.
For a single-pole response it follows from the −3 dB bandwidth as below. Slow diffusion tails and the front end can make a real step slower than the estimate.
The frequency at which the response falls to 0.707 of its low-frequency value, half the electrical power.
Set by the RC of the junction capacitance and the load, by carrier transit time and by the front end. The RC limit is below; with a transimpedance gain amplifier the op amp's gain-bandwidth enters too.
f−3dB=2πRLCj1
Unit: MHzAppears as: f−3dB, BW, Bandwidth, Cut-off frequency
The optical power that gives a signal equal to the noise in a 1 Hz bandwidth, in W/√Hz. Lower is better.
It is the total current noise divided by the responsivity, so it applies at one wavelength. A system value includes shot noise, Johnson noise and the amplifier's noise. The default 18.36 fA/√Hz is the quadrature sum of the shot noise of 1 nA and the Johnson noise of 1 GΩ at 25 °C.
NEP=Rin
fA/√Hz
Range 0.001 to 1e9 fA/√Hz
A/W
Range 0.001 to 2 A/W
Unit: W/√HzAppears as: NEP, Noise equivalent power
Thermal current noise of a resistance, here the shunt or feedback resistance, in A/√Hz.
It falls as the resistance rises, so a high shunt resistance and a large feedback resistor make a quiet low-light front end. It adds in quadrature with shot noise.
in=R4kT
GΩ
Range 1e−6 to 1000 GΩ
°C
Range −60 to 150 °C
Unit: A/√HzAppears as: Johnson noise, Thermal noise, in
The output voltage per unit input current of a transimpedance amplifier (TIA), set by the feedback resistor, in V/A or Ω.
A larger Rf gives more signal and less Johnson noise but less bandwidth. The estimate below holds for a compensated op-amp TIA, with GBW the op amp's gain-bandwidth and Cin the total input capacitance including the junction capacitance.
f−3dB≈2πRfCinGBW
Unit: V/AAppears as: Gain, Rf, Transimpedance, V/A