Ambient Light Interference in PPG: Rejection Techniques and Sensor Design

Ambient light is the second most significant noise source in photoplethysmography after motion artifacts, and effective rejection requires a coordinated approach spanning optical design, analog circuit architecture, and digital signal processing. The pulsatile cardiac component of a PPG signal typically represents only 0.5-2% of the total photodetector output, while ambient light contamination from sunlight, fluorescent lighting, or LED illumination can produce photocurrents 10-1000 times larger than this cardiac signal. Without robust rejection, ambient light degrades heart rate accuracy, corrupts SpO2 measurements, and limits the environments where PPG-based devices can function reliably.

This guide covers every major technique for ambient light rejection in PPG systems, from passive optical shielding to active electronic subtraction, with quantitative performance data and practical design guidance for researchers and engineers building PPG-based sensing systems.

Sources of Ambient Light Interference

Understanding the spectral and temporal characteristics of ambient light sources is essential for designing effective rejection strategies. Each source has distinct properties that determine which rejection techniques are most effective.

Solar Radiation

Sunlight is the most challenging ambient light source for PPG sensors. The solar spectral irradiance covers the entire operating range of PPG sensors, from green (525 nm) through near-infrared (940 nm), with an intensity of approximately 1000 W/m^2 at ground level under clear-sky conditions. At the photodetector, even a small gap between the sensor housing and skin can admit enough solar radiation to saturate the transimpedance amplifier.

The critical issue with sunlight is not just its intensity but its variability. Body movement during outdoor exercise creates time-varying light leakage at the sensor-skin interface, introducing low-frequency noise components (0.5-10 Hz) that directly overlap with the cardiac signal band. Maeda et al. (2011) documented that ambient light variations during outdoor walking produced noise amplitudes 5-20 times larger than the PPG pulsatile component at the wrist, making heart rate extraction impossible without effective rejection.

Fluorescent Lighting

Fluorescent lamps produce optical flicker at twice the AC mains frequency: 100 Hz in regions with 50 Hz power and 120 Hz in regions with 60 Hz power. While these frequencies are well above the cardiac band (0.5-4 Hz), they can alias into lower frequencies if the PPG sampling rate is not sufficiently high relative to the flicker frequency. According to Nyquist, sampling rates below 200-240 Hz risk aliasing fluorescent flicker into the cardiac band.

Older magnetic-ballast fluorescent lamps produce a near-sinusoidal intensity modulation at 100-120 Hz with modulation depths of 30-40%. Modern electronic-ballast fluorescents operate at 20-80 kHz, producing much less visible flicker but still generating detectable optical modulation at the photodetector. Compact fluorescent lamps (CFLs) can exhibit complex multi-frequency flicker patterns.

LED and PWM-Driven Lighting

Modern LED room lighting is typically driven by pulse-width modulation (PWM) at frequencies ranging from 100 Hz to 400 Hz for dimming control. Some LED drivers operate at even higher frequencies (1-10 kHz). The sharp rectangular waveform of PWM produces harmonics extending well beyond the fundamental frequency. Intensity modulation depths of LED lighting can reach 100% (full on-off modulation), making PWM-driven LEDs a significant interference source.

Smart lighting systems with adaptive dimming add another complication: the PWM frequency or duty cycle may change dynamically, making fixed-frequency notch filtering ineffective. Lee et al. (2014) characterized the spectral signatures of common indoor lighting sources and found that LED PWM harmonics could extend into the low-frequency region below 10 Hz under certain dimming configurations.

Display Backlights

For wrist-worn wearables, the device's own display backlight can be a significant ambient light source. OLED and LCD backlights typically use PWM dimming at 200-500 Hz, and light can couple to the PPG photodetector through internal optical paths within the watch case or through reflections off the skin. This is a design-specific issue that requires careful attention to internal optical isolation.

Passive Optical Rejection Methods

Passive rejection techniques attenuate ambient light before it reaches the photodetector, reducing the dynamic range requirements on the analog front-end and improving overall signal-to-noise ratio.

Optical Barriers and Housing Design

The first line of defense against ambient light is the physical design of the sensor housing. An opaque, light-tight enclosure that maintains consistent contact with the skin prevents external light from reaching the photodetector through direct or scattered paths.

Key design elements include opaque side walls between the LED and photodetector cavities to prevent direct LED-to-detector optical coupling, a compliant skin-contact surface (often silicone or elastomer) that conforms to the skin surface to eliminate air gaps where ambient light could leak in, and raised optical barriers or ridges around the photodetector aperture that block oblique-angle ambient light.

The effectiveness of mechanical sealing is highly dependent on the body site and activity level. At the wrist during exercise, sensor lift-off and lateral sliding can periodically break the optical seal, admitting bursts of ambient light. Castaneda et al. (2018) noted that sensor-skin coupling variability is a primary contributor to PPG signal degradation in wearable devices, affecting both motion artifact and ambient light contamination. Robust housing design can reduce ambient light leakage by 20-40 dB compared to an unshielded sensor, but cannot eliminate it entirely during dynamic conditions.

Optical Bandpass Filters

Narrowband optical bandpass filters placed over the photodetector transmit light at the LED wavelength while rejecting ambient light at other wavelengths. Since ambient light is broadband (covering the full visible and near-infrared spectrum), and the PPG LED emits in a narrow band (typically 20-30 nm full-width half-maximum), a well-designed optical filter can reject 90-99% of ambient light energy while passing the majority of the PPG signal.

For green PPG at 525 nm, a bandpass filter with a 30 nm passband (510-540 nm) transmits approximately 85-90% of the LED emission while blocking the remaining solar spectrum. The rejection ratio depends on the filter technology: thin-film interference filters achieve out-of-band rejection of 40-60 dB (OD 4-6), while absorptive color glass filters provide 20-30 dB rejection with broader transition bands.

The choice of filter technology involves tradeoffs. Interference filters provide sharp spectral cutoffs and high rejection but are angle-sensitive: their center wavelength shifts to shorter wavelengths at oblique angles of incidence, potentially misaligning with the LED emission. For reflectance-mode PPG where the detector receives light from a wide angular cone (scattered from tissue), this angular sensitivity must be considered. Absorptive filters are angle-insensitive but have broader transition bands and lower peak rejection. Multi-layer hybrid approaches combining interference and absorptive elements offer the best performance.

Modern PPG sensor modules increasingly integrate optical filters directly into the photodetector packaging. The OSRAM SFH 7072, for example, includes a green-pass filter integrated above the photodiode, providing combined ambient rejection and wavelength selectivity in a compact package suited for wearable device designs.

Spatial Filtering and Detector Geometry

The geometric arrangement of LEDs and photodetectors affects ambient light sensitivity. Placing the photodetector in a recessed well or behind a light guide that accepts only light arriving from specific angles (determined by the tissue scattering geometry) can preferentially reject ambient light entering from oblique angles while accepting back-scattered PPG signal light.

Multi-detector configurations with differential readout provide another spatial rejection mechanism. If two photodetectors are placed at different distances from the LED, they receive different amounts of PPG signal (because signal amplitude decreases with LED-detector separation) but similar amounts of ambient light (which is spatially uniform on the sensor scale). Subtracting the far detector signal from the near detector signal cancels the common-mode ambient light while preserving the differential PPG signal. Tremper and Barker (1989) described early implementations of differential optical detection for clinical pulse oximetry to improve ambient light rejection.

Active Electronic Rejection Methods

Active rejection techniques use electronic timing and signal processing to separate the PPG signal from ambient light contamination in the time domain.

Ambient Light Subtraction (ALS)

Ambient light subtraction is the most widely implemented active rejection technique in modern PPG systems. The operating principle relies on time-division multiplexing of LED-on and LED-off measurement phases.

In a typical ALS sequence, the system first samples the photodetector output with the LED turned off. This LED-off sample captures only ambient light. The system then turns on the LED and takes a second sample. This LED-on sample contains both the PPG signal and ambient light. By subtracting the LED-off sample from the LED-on sample, the ambient light component is removed, leaving only the PPG signal.

For this subtraction to be effective, the ambient light must remain approximately constant between the LED-off and LED-on samples. This requires the two samples to be taken in rapid succession, typically within 50-500 microseconds. If ambient light changes significantly between samples (due to rapid flicker or sensor movement), a residual error remains after subtraction.

The timing architecture typically operates as follows: an LED-off sample of 10-100 microseconds, a settling time of 5-20 microseconds for the transimpedance amplifier to respond to the LED turn-on transient, an LED-on sample of 10-100 microseconds, and then the LED turns off again. This cycle repeats at the desired PPG sampling rate (typically 25-500 Hz). The duty cycle of LED operation is typically 1-10%, which also reduces average LED power consumption.

Modern analog front-end (AFE) chips implement ALS in hardware with programmable timing. The Texas Instruments AFE4404, widely used in wearable PPG applications, provides three-phase sampling (ambient, LED1, LED2) with programmable timing down to 1-microsecond resolution and integrated sample-and-hold circuits that perform the subtraction in the analog domain before ADC conversion. Maxim Integrated's MAX86150 and Analog Devices' ADPD4101 offer similar capabilities. These AFE architectures are covered in detail in our analog front-end design guide.

ALS typically achieves 40-60 dB of ambient light rejection under static conditions. Performance degrades when ambient light changes rapidly (faster than the sampling interval) or when fluorescent/LED flicker frequencies produce significant variation within the LED-off to LED-on interval. Quantitative characterization by Winokur et al. (2012) demonstrated that ALS maintained PPG signal integrity under indoor fluorescent lighting with photocurrents up to 10 microamperes, but performance degraded above 50 microamperes of ambient photocurrent without additional optical filtering.

Modulated LED Drive and Synchronous Detection

An advanced extension of ALS uses modulated (typically square-wave) LED drive at a carrier frequency well above the ambient light interference band, combined with synchronous demodulation (lock-in detection) at the receiver. This approach is conceptually similar to a lock-in amplifier and provides narrowband rejection of all signals not at the modulation frequency.

The LED is modulated at a carrier frequency of 1-10 kHz. The photodetector output contains the PPG signal modulated onto this carrier plus ambient light (concentrated at DC and low frequencies for sunlight, and at 100-400 Hz for artificial lighting). A synchronous demodulator multiplies the detector output by the same carrier frequency and low-pass filters the result, extracting only the signal correlated with the LED modulation. All ambient light components at frequencies other than the carrier frequency are rejected.

This technique provides superior rejection compared to simple ALS because it effectively creates a very narrow detection bandwidth (determined by the low-pass filter after demodulation) centered on the LED modulation frequency. Rejection ratios of 60-80 dB are achievable. The tradeoff is increased circuit complexity, higher LED drive frequency requirements, and the need for precise timing synchronization between the LED driver and demodulator. Patterson et al. (2009) demonstrated synchronous detection achieving ambient rejection exceeding 70 dB in a reflectance-mode pulse oximeter, enabling accurate SpO2 measurement under direct surgical lamp illumination.

Correlated Double Sampling (CDS)

Correlated double sampling is a circuit technique borrowed from imaging sensor design that reduces low-frequency noise and DC offsets, including ambient light DC components. In CDS, each measurement consists of a reset sample (baseline reference) followed by a signal sample, and the difference is taken. This cancels any DC or slowly varying components common to both samples, including ambient light DC offset and 1/f noise in the amplifier.

CDS is particularly effective when combined with ALS, as it removes both the ambient DC component and the amplifier's own DC offset and low-frequency noise, leaving only the AC PPG signal and high-frequency noise that can be removed by subsequent filtering.

Digital Signal Processing Approaches

When hardware-level rejection is insufficient, digital post-processing techniques provide additional ambient light attenuation.

Adaptive Notch Filtering

For environments with known artificial lighting frequencies (100/120 Hz fluorescent flicker and their harmonics), adaptive notch filters can track and remove these specific frequency components from the digitized PPG signal. Adaptive implementations adjust the notch frequency and bandwidth in real-time to track changes in the interference frequency, accommodating variations in mains frequency and PWM dimming.

The key challenge is avoiding removal of PPG harmonic content that coincides with the notch frequency. For typical heart rates (50-200 BPM, or 0.83-3.33 Hz), the harmonic content extends up to 20-30 Hz and does not overlap with 100-120 Hz fluorescent flicker, making notch filtering safe for this interference source. However, some LED PWM dimming operates at frequencies low enough to overlap with PPG harmonics, where notch filtering risks distorting the PPG waveform morphology.

Multi-Channel Ambient Estimation

Some advanced PPG systems include a dedicated ambient light photodetector separate from the main PPG photodetector. This ambient-only channel is positioned to receive ambient light but not PPG signal (for example, facing away from the skin or shielded from the LED). The ambient channel signal is then used as a reference input to an adaptive filter (LMS or NLMS) that cancels ambient light components from the main PPG channel.

This approach is particularly effective for rejecting non-stationary ambient light variations (such as intermittent sunlight during outdoor exercise) because the adaptive filter continuously tracks the changing ambient light characteristics. The technique is analogous to accelerometer-referenced adaptive filtering for motion artifact removal, but applied to the optical noise domain.

Signal Quality Index and Segment Rejection

When ambient light interference exceeds the capability of rejection techniques, signal quality assessment provides a safety net by identifying and flagging or discarding corrupted segments. Signal quality indices (SQIs) for ambient light contamination include monitoring the DC level of the photodetector output (a sudden increase suggests ambient light leakage), measuring the noise floor in frequency bands above the cardiac range (20-50 Hz) where ambient light flicker may appear, and detecting saturation events in the ADC or transimpedance amplifier. Segments identified as having poor ambient light contamination can be excluded from heart rate or SpO2 calculations, accepting gaps in coverage to maintain measurement accuracy. This approach is essential for clinical-grade SpO2 estimation where inaccurate readings can have serious consequences.

Quantitative Performance Benchmarks

The effectiveness of ambient light rejection techniques can be characterized by several metrics.

Ambient light rejection ratio (ALRR): The ratio of ambient light photocurrent that can be tolerated before PPG signal quality degrades below a specified threshold. Optical barriers alone provide 20-40 dB. Adding optical bandpass filters increases this to 40-60 dB. Combining optical filtering with electronic ALS achieves 60-80 dB. Synchronous detection can push total rejection above 80 dB.

Maximum tolerable ambient photocurrent: For a typical wrist PPG sensor with a 3 mm^2 photodiode, direct sunlight can generate 50-200 microamperes of photocurrent. The transimpedance amplifier must handle this without saturation while maintaining sufficient resolution for the 10-100 nanoampere pulsatile PPG signal. This demands a dynamic range exceeding 60-80 dB, which is why ambient rejection before the amplifier (optical filtering) is so critical.

Heart rate accuracy under ambient interference: Well-designed PPG systems maintain heart rate MAE below 3-5 BPM under indoor lighting conditions (fluorescent, LED) and below 5-10 BPM under bright outdoor conditions with moderate motion. Without ambient rejection, outdoor accuracy can degrade to 15-30 BPM MAE or produce no valid readings. Bent et al. (2020) evaluated consumer wearables under varying ambient conditions and found heart rate accuracy degraded by 2-8 BPM in bright outdoor environments compared to controlled indoor settings.

Design Recommendations and Best Practices

For engineers designing PPG sensor systems, the following recommendations summarize the state of the art in ambient light rejection.

Layer the defenses. No single rejection technique is sufficient for all conditions. A robust design combines optical barriers (housing design), optical filtering (bandpass filter on photodetector), electronic subtraction (ALS in the AFE), and digital post-processing (notch filtering, signal quality assessment). Each layer addresses different aspects of the ambient light problem.

Design for worst-case outdoor conditions. Direct sunlight produces ambient photocurrents 100-1000 times larger than the PPG signal. The optical path, transimpedance amplifier dynamic range, and ADC resolution must accommodate this without saturation. If outdoor exercise use is a requirement, optical bandpass filtering is essentially mandatory.

Match filter passband to LED emission. The optical bandpass filter should be centered on the LED peak emission wavelength with a bandwidth just wide enough to accommodate the LED spectral width (typically 20-30 nm FWHM for green LEDs, 30-50 nm for infrared LEDs) and any angular shift in the filter transmission. Tighter filtering provides better ambient rejection but risks attenuating the PPG signal if wavelengths are not well matched.

Minimize the ALS timing interval. The effectiveness of ambient light subtraction depends on ambient light being constant between LED-off and LED-on samples. Shorter intervals (faster LED switching) improve rejection of rapidly varying ambient sources. Modern AFEs support switching intervals below 50 microseconds, which is sufficient for rejecting ambient variations up to several kilohertz.

Validate across lighting environments. Testing should include direct sunlight (1000 W/m^2), indoor fluorescent lighting (magnetic and electronic ballast), LED room lighting with PWM dimming at various duty cycles, mixed lighting (window plus overhead), and darkness as a baseline reference. Standardized test protocols from IEC 80601-2-61 provide guidance for clinical pulse oximeter ambient light testing. For a comprehensive view of how these techniques integrate with overall PPG signal processing pipelines, see our algorithm guide.

Emerging Technologies

Several emerging approaches promise improved ambient light rejection in future PPG systems.

Integrated photonic filters. Wafer-level deposition of multi-layer interference filters directly on photodetector arrays enables narrowband filtering without separate optical components, reducing module thickness and improving angular performance.

Frequency-domain PPG. Operating the PPG measurement entirely in the frequency domain, with continuous sinusoidal LED modulation and coherent detection, provides inherently narrowband measurement that rejects all out-of-band interference including ambient light. This approach is gaining interest for next-generation wearable sensors.

Computational ambient cancellation. Machine learning models trained on paired ambient-contaminated and clean PPG data can learn complex, nonlinear ambient light patterns and remove them more effectively than linear subtraction. Lee et al. (2019) demonstrated a CNN-based ambient light removal approach that improved SNR by 8-12 dB compared to conventional ALS, though computational requirements currently exceed what most wearable processors can support.

Micro-optic light guides. Structured light guides and micro-lens arrays above the photodetector can control the angular acceptance cone, preferentially accepting back-scattered PPG light from tissue while rejecting ambient light entering from outside the sensor-skin contact area. These approaches are being explored for next-generation wearable sensor modules with improved ambient immunity.

Frequently Asked Questions

How does ambient light affect PPG measurements?

Ambient light reaches the photodetector alongside the PPG signal, adding a large DC offset and introducing AC noise from artificial lighting (fluorescent flicker at 100-120 Hz, LED PWM at 100-400 Hz) and natural light variations (body movement under sunlight). This noise can saturate the analog front-end, reduce dynamic range available for the pulsatile cardiac signal, and inject spectral components that overlap with the heart rate band (0.5-4 Hz), degrading heart rate and SpO2 accuracy.

What is ambient light subtraction in PPG?

Ambient light subtraction is a hardware-level technique where the PPG system alternates between LED-on and LED-off sampling phases. During the LED-off phase, only ambient light reaches the photodetector. This ambient-only sample is subtracted from the LED-on sample (which contains both PPG signal and ambient light), isolating the PPG component. Modern analog front-end chips like the TI AFE4404 and Maxim MAX86150 implement this in hardware with precise timing control, achieving 40-60 dB of ambient rejection.

Can PPG sensors work in direct sunlight?

Yes, but performance is significantly challenged. Direct sunlight can produce photocurrents 100-1000 times larger than the PPG signal at the photodetector, risking analog front-end saturation. Effective outdoor operation requires a combination of optical barriers (opaque housing, skin-contact seals), optical bandpass filters matched to the LED wavelength, high-dynamic-range transimpedance amplifiers, ambient light subtraction, and increased LED drive current. Well-designed wearables maintain heart rate accuracy within 3-5 BPM even in bright outdoor conditions.

What optical filters are used in PPG sensors to reject ambient light?

PPG sensors use narrowband optical bandpass filters centered on the LED emission wavelength with typical passbands of 20-50 nm. For green PPG (525 nm), a filter passing 510-540 nm rejects the majority of broadband ambient light while transmitting the PPG signal. Infrared PPG sensors use filters centered near 940 nm. These filters are implemented as thin-film interference coatings deposited directly on the photodetector or on a separate glass substrate above it. Combined with ambient light subtraction, optical filtering can achieve total ambient rejection exceeding 80 dB.