IR Receiver Optical Filter Bandwidth: Matching Passband to Emitter and Detector
This guide treats the filter as part of a chain rather than a part number: emitter spectrum on one side, detector response on the other, and the optical path in between. Getting the IR receiver optical filter bandwidth right is a matching exercise, and the numbers that matter are ratios, not absolutes.
Quick Answer: What the Filter Bandwidth Actually Buys
A narrow passband admits less unwanted light, which raises the ratio between signal and background. It also admits less of the emitter's own output, shifts with angle and temperature, and costs more to manufacture. The right bandwidth is therefore the widest one that still delivers the required ambient light rejection under the worst ambient the product will meet.
Start from the ambient, not from the filter catalogue. If a receiver is only ever used indoors under filtered light, a 50 nm passband may be generous. If it faces direct sun, the same 50 nm may be the difference between a working channel and a saturated one.
What an IR Optical Filter Does in the Receiver Chain
Silicon responds to a very wide range of wavelengths, from the visible through to roughly 1100 nm. That breadth is what makes silicon photodiodes cheap and what makes them vulnerable: visible light is inside their response, so daylight is not a nuisance at the edge of the band, it is a large signal at the centre of it.
An IR bandpass filter placed in front of the detector removes most of that visible energy before it becomes photocurrent. It cannot remove energy inside its own passband, and it cannot help if the ambient light happens to be strong exactly where the emitter sits. That last limitation is why filter choice and emitter choice have to be made together.
There is a second, quieter benefit. Filtering reduces the total photocurrent, which lets the transimpedance stage use a larger feedback resistor for the same output swing. The result is a lower noise floor, and a channel that works at a longer range than an unfiltered one with the same emitter. Wavelength selection on the emitter side is treated in the guide to 940nm vs 980nm IR LED, and the two decisions should be read as one.
Center Wavelength and Passband Width: The Only Two Numbers That Matter First
Two specifications define a filter well enough to start a design: where the band sits, and how wide it is. The center wavelength should be placed on the emitter's peak, or slightly to the short-wavelength side of it, because emitter peaks drift upward with temperature while filter bands drift in the opposite direction in many coating designs.
| Passband width | Ambient admitted | Emitter energy captured | Practical consequence |
|---|---|---|---|
| Very narrow (tens of nm) | Low | Low, sensitive to drift | Best daylight rejection, tightest alignment and thermal tolerance |
| Moderate (tens to ~50 nm) | Moderate | Most of the peak | The usual compromise for outdoor sensing |
| Wide (100 nm and above) | High | Nearly all of the peak | Forgiving of drift, poor rejection of bright ambient |
Note what the table implies: widening the passband helps the emitter side of the budget and hurts the ambient side. Neither column can be optimised alone, which is why a bandwidth specified without a stated ambient condition is not a specification at all.
The shape of the emitter peak sets the floor on how narrow the filter can usefully be. A device with a broad emission peak loses proportionally more when the filter narrows, and the full-width at half-maximum figure is the number that quantifies it, as explained in LED FWHM spectral bandwidth.
Angle of Incidence and Temperature Shift
Thin-film filters are interference devices, so their passband position depends on the angle at which light crosses them. Tilt the filter by twenty degrees and the band moves toward shorter wavelengths; tilt it more and the effect accelerates. In a receiver with a wide field of view this means the effective center wavelength is not a single value but a range, and the edges of that range are where signal is lost.
Temperature adds a slower drift in the same direction or the opposite one, depending on the coating stack. A filter specified at room temperature and installed behind a window that reaches high temperature in sun may be offset by several nanometres in service. The consequence is not a failure but a margin reduction, and margin reductions are what turn a working prototype into a field complaint.
Both effects argue for keeping the passband wider than the emitter's thermal drift alone would require, and for placing the band asymmetrically: more room on the side toward which the emitter drifts. Optical alignment across the whole field of view is covered in the guide to IR LED polar radiation pattern, because a receiver that only works on axis is a receiver with an alignment problem rather than a filter problem.
Detector Responsivity and Where the Passband Must Sit
Silicon responsivity rises with wavelength up to roughly 900 to 950 nm and then falls sharply toward the band gap. This means two emitters with equal radiant output do not produce equal photocurrent: one at 850 nm may generate noticeably more signal than one at 940 nm in the same receiver, purely because of where the detector is more sensitive.
That asymmetry changes the filter decision. A channel built around a shorter-wavelength emitter has more detector response to work with, so it can afford a narrower filter and gain rejection. A channel built around a longer-wavelength emitter has less response, so a narrow filter may cost more signal than the ambient rejection is worth.
The practical rule is to compute the product, not compare the factors. Emitter power inside the passband, multiplied by filter transmission, multiplied by detector responsivity at that wavelength, gives the signal available. Ambient power inside the same passband, multiplied by the same transmission and responsivity, gives the competing background. The ratio of those two numbers is the design margin, and a receiver with a good ratio at one wavelength can have a poor one at another with the same filter.
Ambient Sources in Practice
Daylight is the dominant source in outdoor products, and its spectrum is broad and smooth: it fills whatever passband you choose. What changes with bandwidth is how much of it arrives, which is why daylight rejection improves roughly in proportion to the narrowing of the band.
Artificial sources behave differently and are often worse per unit of visible brightness. Incandescent lamps radiate strongly in the near infrared. Some fluorescent ballasts and dimmed LED drivers emit modulated components that fall inside the modulation band of the receiver, which no optical filter can remove. Halogen security lighting, welding arcs and direct sunlight reflected from glass are the cases that most often defeat a filter that looked adequate on paper.
For a receiver intended for industrial sensing, the useful sequence is to list the sources the product will actually see, then check each one against the passband rather than against the visible brightness. The sensing use cases in 850nm vs 940nm IR LED for industrial sensing give a starting set of ambient conditions to test against.
Validating With Real Lighting Before Fixing the Bandwidth
Bench validation under a lab lamp proves very little. The test that predicts field behaviour uses the actual light sources, at the actual distance, with the actual window or lens in place, and it measures the receiver output with the emitter on and off rather than only with the emitter on.
- Record the DC level with the emitter disabled under each ambient source; this is the background the filter has to suppress.
- Record the modulated amplitude with the emitter enabled; this is the signal the filter has to preserve.
- Compute the ratio at the worst ambient, not at the average one.
- Repeat at the temperature extremes the product will see, since both the filter band and the emitter peak move.
- Repeat at the extremes of the field of view, because angle shift changes the effective center wavelength.
A ratio that holds at the worst-case corner is a design; a ratio that holds only in the lab is a hope. When the measured ratio falls short, the first lever is usually not a narrower filter but a different combination of emitter wavelength and detector, which is why filter and emitter should be selected as one decision rather than two.
Frequently Asked Questions
How to choose IR receiver optical filter bandwidth for outdoor use?
Narrow enough to keep the signal-to-background ratio above your detection threshold under direct sun, and no narrower. Specify the ambient condition with the number, or the figure means nothing.
Can I use a visible-light-blocking filter instead of a bandpass filter?
A long-pass filter removes visible light but leaves a very wide infrared window open, so it rejects far less daylight than a bandpass device. It is adequate only where the ambient is weak.
Does a wider passband always mean a stronger signal?
It admits more emitter energy, but it admits proportionally more ambient as well. The ratio between signal and background often gets worse, not better.
Why does my receiver work on the bench and fail behind the product window?
The window tilts the filter and shifts its passband, and it may add its own absorption. Validate with the final optical stack rather than with the bare receiver.
Is ambient light rejection IR sensor performance mainly an optical problem?
Optics set the ceiling, and electrical design decides how close you get to it. Modulation, threshold hysteresis and integration time all act after the filter has done its work.
Do I need a different filter for a 940nm emitter than for an 850nm one?
Usually yes, because the band has to follow the peak and because detector responsivity differs at the two wavelengths. Using one filter for both accepts a margin loss at one of them.
Conclusion: Match the Band, Then Measure the Ratio
Filter bandwidth is not a quality grade that can be maximised. It is a matching parameter whose correct value is set by the emitter peak, the detector response and the worst ambient the product will meet, in that order. Read as a chain, the three decisions resolve together; read separately, they produce a receiver that passes on the bench and disappoints in the field.
If you are pairing an infrared emitter with a filtered receiver and want the wavelength and drive conditions checked against your optical stack, send the application details to the HOUKEM team, or review the emitter families on the HOUKEM IR LED product page first.
