Safety
Safety is a top priority at FaceFocusVR – but ultimately, responsibility for safe usage lies with the user.
FaceFokusVR does not accept liability for any damage resulting from the use or misuse of the hardware.
While infrared radiation is a natural part of our environment (approximately 50% of sunlight consists of infrared), excessive exposure can be harmful. The hardware includes multiple safety mechanisms designed to ensure that IR emissions remain far below internationally recognized safety limits.
Because transparency is important to me, I want to make it 100% clear that the system—like any other DIY eye-tracking setup I know of—has not been independently certified for compliance with formal safety standards; however, my calculations, in-depth testing, measurements, and consultations with professors in relevant fields have confirmed—with a significant safety margin—that all applicable exposure limits are well respected.
Still, no system is 100% secure. Therefore, it’s also your responsibility to:
Never attempt to replace, disable, or modify safety-related components.
Monitor how your eyes feel during use.
If you notice any unusual warmth or discomfort in your eyes, stop using the device immediately and assess the situation.
A slight warmth may occur due to the added cameras, but with the cooling fan running properly, your eyes should feel normal—aside from a gentle breeze if you’re not used to having a fan inside your VR headset.
Discontinue use immediately if you experience symptoms such as dark spots in your vision, dry eyes, or excessive eye strain.
If you’re unsure about anything safety related—calculations, circuits, safety—please contact me. I’ll be happy to help.
Fundamentals and Scientific Foundations
Basics
Infrared (IR) radiation is commonly used in eye-tracking systems to illuminate the eye without being visible to the human eye. While IR radiation is a natural part of our environment—approximately 50% of solar radiation is infrared—prolonged or excessive exposure can be harmful, particularly to sensitive areas like the eyes.
To ensure safe use, guidelines have been established that define exposure limits based on factors such as the affected body part (e.g., eye, skin), the wavelength of the radiation, and the duration of exposure. My safety assessment is primarily based on two authoritative sources:
The ICNIRP (International Commission on Non-Ionizing Radiation Protection) is an independent organization that provides scientifically backed guidelines on the health effects of non-ionizing radiation, including infrared. Their exposure limits are widely recognized and used internationally.
EN 62471 is a European standard that provides detailed criteria for evaluating the photobiological safety of lamps and lamp systems, covering ultraviolet (UV), visible, and infrared (IR) radiation. It specifies exposure thresholds to prevent thermal and photochemical damage to the eyes and skin.
These sources form the foundation of my calculations, design decisions, and safety precautions to ensure that IR exposure from the system remains well below harmful levels. Since the content and limits of both references are essentially identical, only the EN 62471 standard will be explained in detail below. The ICNIRP sources are cited without further elaboration.
EN 62471
EN 62471 evaluates photobiological hazards from optical radiation in the range of 200–3000 nm. In the context of this project, continuous infrared exposure at 860 nm for more than 10 seconds is relevant, specifically involving thermal hazards to the retina, thermal effects on the lens, and thermal damage to the skin.
Infrared radiation hazard exposure limits for the eye
IR radiation can be absorbed by the outer and inner structures of the eye, including the cornea and lens, leading to localized heating. Since IR radiation is invisible and does not trigger natural protective reflexes, the eye is particularly vulnerable to unintentional overexposure. For exposure durations longer than 1000 seconds, EN 62471 sets strict irradiance limits to prevent acute thermal injury and minimize long-term degenerative effects like cataractogenesis:
- To avoid thermal injury of the cornea and possible delayed effects upon the lens of the eye (cataractogenesis), ocular exposure to infrared radiation, EIR, over the wavelength range 780 nm to 3000 nm, for times greater than 1000 s, shall not exceed [EN 62471 4.3.7]:
$$ E_{IR} = \sum_{780}^{3000} E_{\lambda} \times \Delta \lambda \leq 100 \ \quad\quad \left[\frac{\text{W}}{\text{m}^2}\right] \quad \quad \text{for } (t > 1000 \ \text{s}) $$
$$ E_{IR} \leq 100 \ \frac{\text{W}}{\text{m}^2} = 10 \ \frac{\text{mW}}{\text{cm}^2} \quad\quad \text{ for } (t > 1000 \ \text{s}) $$
Where:- Eλ is the spectral irradiance,
- Δλ is the bandwidth,
- t is the exposure duration,
- EIR is the infrared irradiance (total IR radiation power per unit area over the wavelength range 780–3000nm).
Retinal thermal hazard exposure limit — weak visual stimulus
Unlike the general radiation limits that primarily address thermal effects on the front parts of the eye, the retinal thermal hazard exposure limit focuses on the risk of damage to the retina caused by infrared radiation. Even when the visual stimulus is weak or barely noticeable, the radiation can be concentrated onto the retina, causing localized heating and potential injury. Because the retina is particularly sensitive to temperature increases, EN 62471 establishes strict exposure limits for short durations to protect retinal cells from irreversible thermal damage.
- For an infrared heat lamp or any near-infrared source where a weak visual stimulus is inadequate to activate the aversion response; the near infrared (780 nm to 1400 nm) radiance, LIR, as viewed by the eye for exposure times greater than 10 s shall be limited to [EN 62471 4.3.6]:
$$ L_{IR} = \sum_{780}^{1400} L_{\lambda} \times R(\lambda) \times \Delta \lambda \leq \frac{6000}{\alpha} \quad\quad \left[\frac{\text{W}}{\text{m}^2\cdot\text{sr}}\right] \quad\quad \text{for } (t > 10\ \text{s}) $$
Where:- L is the spectral radiance,
- R(λ) is the burn hazard weighting function,
- Δλ is the bandwidth in nm,
- t is the exposure time in seconds,
- α is the angular subtense in radians.
Thermal hazard exposure limit for the skin
In addition to ocular safety, EN 62471 also addresses the risk of thermal injury to the skin caused by prolonged exposure to infrared radiation. Since the skin can absorb IR radiation over a broad area, excessive exposure may lead to surface heating, burns, or long-term tissue damage. However, for exposure durations exceeding 10 seconds over larger areas, the standard notes that pain is typically perceived before any actual tissue damage occurs. As a result, an individual’s natural aversion response due to discomfort generally limits exposure well before injury is possible. For this reason, thermal hazard exposure limits for the skin are not further considered.
-
[...] exposure limit is based on skin injury due to a rise in tissue temperature and applies only
to small areairradiation. Exposure limits for periods greater than 10 s are not provided. Severe
pain occurs below the skin temperature required for skin injury, and an individual's exposure
normally will be limited for comfort. Large area irradiation and heat stress are not evaluated
since this involves consideration of heat exchange between the individual and the
environment, physical activity, and various other factors, which cannot be applied in a product
safety standard, but must be evaluated by environmental heat-stress criteria. [EN 62471 4.3.8 (Note)]
ICNIRP
ICNIRP, the International Commission on Non-Ionizing Radiation Protection, is an independent organization that provides scientific guidance on the health effects of non-ionizing radiation, including infrared. In its 2006 and 2013 publications, ICNIRP addresses exposure limits and potential health impacts. The 2006 paper, available here, includes the relevant skin exposure limit in formula 4b on page 639 (or page 11 of the document). The 2013 paper, available here, presents an updated limit in formula 21 on page 88 (or page 18 of the document).
EN 62471 Compliance Evaluation
Datasheet Parameters
The key parameters used in the following calculations are taken from the LED’s official datasheet: CSL1501RW1
- Radiation Intensity: Maximum of 3.4 \(\frac{\text{mW}}{\text{sr}}\) at 30 mA
- Horizontal Emission Angle: 140°
- Vertical Emission Angle: 160°
The LED emits 3.4 milliwatts of optical power per steradian² when operated at a forward current of 30 mA. Since the LED is limited to 2.4 mA due to the board design, and the datasheet indicates that the radiation intensity decreases linearly, we can calculate the adjusted radiation intensity as follows:
\[ I_{\text{op}} = \frac{3.4 \ \frac{\text{mW}}{\text{sr}} \times 2.4 \ \text{mA}}{30 \ \text{mA}} = 0.273 \ \frac{\text{mW}}{\text{sr}} \]
² A steradian is a unit used to measure angles in three-dimensional space, similar to how a radian measures angles in a circle.
Solid Angle
In the context of light sources like LEDs, the solid angle describes the portion of space into which the light is emitted. A larger solid angle means the light is spread over a wider area, while a smaller solid angle indicates a more focused beam. For reference, the total solid angle surrounding a point in all directions (a full sphere) is 4π sr, which is approximately 12.57 sr.
Since LEDs do not emit light in a perfect cone in just one direction but rather in more of an elliptical pattern, the usual formula for calculating the solid angle based on a circular cone is not entirely accurate. Instead, it’s better to approximate the emission using an elliptical model that takes into account the different horizontal and vertical beam angles. This can be done using the formula:
\[ \Omega = 4 \arcsin \left( \sin \left( \frac{\theta_x}{2} \right) \cdot \sin \left( \frac{\theta_y}{2} \right) \right) \]
where $\theta_x$ and $\theta_y$ are the horizontal and vertical beam angles, respectively, given in radians.
It’s important to note, however, that this formula is still an approximation. In reality, the light distribution of LEDs is not perfectly rectangular as the formula might suggest, but rather elliptical or somewhat smoother in shape. The formula deliberately simplifies the geometry for ease of calculation, but despite this, it provides a sufficiently accurate estimation of the solid angle.
When the beam angles from the datasheet are applied in the formula, the resulting calculation yields a solid angle of approximately 4.73 steradians:
\[ \Omega = 4 \arcsin \left( \sin \left( \frac{160^\circ \cdot \pi}{360} \right) \cdot \sin \left( \frac{140^\circ \cdot \pi}{360} \right) \right) = 4.73 \,\text{sr} \]
Illuminated Area
To calculate the area illuminated by each individual LED, the Oosterom-Strackee formula can be used after a suitable transformation. This formula has already been applied in combination with the horizontal and vertical beam angles to estimate the solid angle (steradian) covered by the LED.
However, by rearranging the formula, it can also be used to calculate the projected area of the light cone at a given distance. Specifically, it allows us to estimate the size of the illuminated spot, which is formed by the horizontal and vertical spread of the LED beam.
\[ \text{Width} = 2 \cdot h \cdot \tan\left(\frac{\theta_x}{2}\right) \quad ; \quad \text{Height} = 2 \cdot h \cdot \tan\left(\frac{\theta_y}{2}\right) \]
\[ A = \text{Width} \cdot \text{Height} = 4 \cdot h^2 \cdot \tan\left(\frac{\theta_x}{2}\right) \cdot \tan\left(\frac{\theta_y}{2}\right) \]
where $\theta_x$ and $\theta_y$ are the horizontal and vertical beam angles, respectively, given in radians, and $h$ is the distance from the LED to the illuminated surface.
Using the beam angles from the datasheet and a distance of 1 cm in the formula, the calculated illuminated area is approximately 63.33 cm².
\[ A = 4 \cdot (1\,\text{cm})^2 \cdot \tan\left(\frac{140^\circ \cdot \pi}{360}\right) \cdot \tan\left(\frac{160^\circ \cdot \pi}{360}\right) = 63.33 \,\text{cm}^2 \]
Total Power
Since the datasheet only specifies the LED’s intensity in \(\frac{\text{mW}}{\text{sr}}\), meaning the power emitted per unit solid angle, but not the total emitted power, the actual power emitted by the LED must be calculated by multiplying this value by the LED’s total solid angle. This yields the total power in milliwatts emitted over the LED’s full emission angle, which is necessary to determine the LED’s power over a specific area (Irradiance) in the next step.
\[ P = I_{\text{op}} \times \Omega = 0.273\,\text{mW/sr} \times 4.73\,\text{sr} = 1.29\,\text{mW} \]
Irradiance
Irradiance is the measure of the radiant power received by a surface per unit area. It quantifies how much energy falls onto a given area. In simpler terms, it tells you how „intense“ the light is on a surface. To calculate it, the total radiant power incident on the surface is divided by the surface area.
Using the previously calculated values, the irradiance from 7 LEDs at a distance of 1 cm can be expressed as:
\[ E = \frac{7 \times P}{A} = \frac{7 \times 1.29\,\text{mW}}{63.33\,\text{cm}^2} = 0.14\,\text{mW/cm}^2 = 1.4\,\text{W/m}^2 \]
Radiance
The previously calculated values focused on irradiance, which relates the LED’s emitted power to the illuminated area on a surface at a given distance. While irradiance is important for evaluating the overall light intensity received by a surface, the EN 62471 standard—particularly regarding the Retinal Thermal Hazard Exposure Limit under weak visual stimulus—requires consideration of radiance as well.
Radiance describes the radiant power emitted by the source per unit projected emitting area and per unit solid angle. This makes it a key parameter for assessing potential retinal hazards, as it captures not only how much power is emitted, but also how concentrated the light is in specific directions. Unlike irradiance, which depends on the receiving surface, radiance is intrinsic to the source’s emission characteristics and better reflects the risk of thermal damage to the retina from focused or directional light.
Radiance can be calculated using the formal definition:
\[ L = \frac{d^2\Phi}{\cos(\beta) \cdot dA \cdot d\Omega} \quad \left[\frac{\text{W}}{\text{m}^2 \cdot \text{sr}}\right] \]
This formula describes how much radiant power \( d^2\Phi \) is emitted from a projected area element \( dA \) of the source into a given solid angle \( d\Omega \), with \( \beta \) being the angle between the surface normal and the emission direction. Importantly, the area \( dA \) refers to the source’s emitting surface—not the illuminated target—and the cosine term \( \cos(\beta) \) accounts for the projection of the surface in the emission direction.
To simplify the calculation, we assume \( \beta = 0^\circ \), meaning the observation direction is perpendicular to the emitting surface. This represents the worst-case scenario, as the radiance is maximized when the viewing angle aligns directly with the surface normal, resulting in \( \cos(\beta) = 1 \):
\[ L = \frac{d^2 \Phi}{dA_{\mathrm{led}} \cdot d\Omega} = \frac{P}{A_{\mathrm{led}} \cdot \Omega} = \frac{1{,}29\,\mathrm{mW}}{0{,}5\,\mathrm{mm}^2 \cdot 4{,}73\,\mathrm{sr}} \approx 545 \, \frac{\mathrm{W}}{\mathrm{m}^2 \cdot \mathrm{sr}} \]
Radiance Limit set by EN 62471
The radiance limit set by EN 62471 is not a fixed value; instead, it depends on the angular subtense of the light source, measured in radians. The angular subtense describes how large the light source appears to the observer — basically, the angle subtended by the source at the eye.
To calculate the limit allowed by EN 62471 for exposures longer than 10 seconds, you simply divide 6000 by the angular subtense aaa (in radians):
\[ L_{\text{limit}} = \frac{6000}{a} \quad \left[\frac{\text{W}}{\text{m}^2 \cdot \text{sr}}\right] \]
To calculate the angular subtense \( a \), both the physical size of the light source and its distance from the observer’s eye must be known. The angular subtense in radians is defined as the angle subtended by the source, based on its characteristic dimension \( d \) (e.g., diameter or side length) and the viewing distance \( r \). For the 0402 LEDs used here, where \( d = 1\,\text{mm} \) and \( r = 1\,\text{cm} \), the angular subtense is:
\[
a = \frac{d}{r} = \frac{1\,\text{mm}}{10\,\text{mm}} = 0.1\,\text{rad}
\]
This results in a limit for exposure durations longer than 10 seconds of:
\[
L_{\text{limit}} = \frac{6000}{a} = \frac{6000}{0.1} = 60{,}000\,\frac{\text{W}}{\text{m}^2 \cdot \text{sr}}
\]
Compliance Check
The calculated irradiance for 7 LEDs at a distance of 1 cm is approximately 1.4 \(\frac{\text{W}}{\text{m}^2}\) (0.14 \(\frac{\text{mW}}{\text{cm}^2}\)). This value is about 70 times lower than the safety irradiance limit of 100 \(\frac{\text{W}}{\text{m}^2}\) (10 \(\frac{\text{mW}}{\text{cm}^2}\)) for prolonged exposure, meaning, based on the calculations, it is well within safe limits.
I also observed similarly low irradiance values when measuring the LEDs.
The radiance of the LED was calculated to be about 545 \(\frac{\text{W}}{\text{m}^2 \cdot \text{sr}}\). According to the EN 62471 standard, the radiance limit for the given angular subtense of 0.1 radians is 60,000 \(\frac{\text{W}}{\text{m}^2 \cdot \text{sr}}\). This means the LED’s radiance is roughly 110 times lower than the safety threshold.
In conclusion, both irradiance and radiance are far below the EN 62471 safety limits.
Hardware Safety Considerations
There are two primary safety considerations in the implementation.
First, the host device—typically the Valve Index—must be protected against any potential faults originating from the custom circuit board. While it is highly likely that the Index includes internal protection mechanisms, the circuit board has been designed with comprehensive safeguards to ensure that it cannot interfere with or cause damage to the host device under any circumstances.
Second, infrared radiation must be minimized to minimize any risk of eye injury. As demonstrated in the calculations above, the current supplied to the LEDs—and thus their emitted radiation—needs to be tightly constrained, ensuring that exposure remains well within safe limits.
Host Device Safety Measures
To sell hardware within the EU, it must pass so-called EMC tests. Simply put, EMC (Electromagnetic Compatibility) determines how much one electrical circuit is allowed to interfere with another—the less interference, the better. These interferences are evaluated, among other things, through standards and measurements in accordance with EN55022 AV/QP. My hardware has successfully passed both measurements with flying colors.
To achieve compliance with the EMC tests, a two-stage EMC filter was integrated. This filter effectively reduces high-frequency noise by using a combination of inductors and capacitors to block or attenuate unwanted signals while allowing the desired signals to pass through, ensuring optimal performance and adherence to emission standards.
The second safety mechanism is the integration of a polyfuse. In the event of a short circuit within the hardware, or if the current draw exceeds the 1 A limit of the Valve Index’s USB port for any reason, the polyfuse activates to protect the device by limiting the current to 1 A. Once the fault condition is resolved and the current returns to a safe level, the fuse automatically resets.
Eye Safety Measures
To ensure safe operation and prevent excessive infrared radiation, the hardware includes three independent safety mechanisms that limit the current—and therefore the LED output power:
Software-Based Current Limiting
The LED current is adjustable via software, allowing precise control over the brightness. A maximum limit is enforced in the firmware, ensuring the LEDs remain within safe operating ranges under normal conditions.
Hardware Limiting via the AW9967DNR LED Driver
The LED driver (AW9967DNR) has a built-in current limiter that sets a hard cap per output channel. This hardware safeguard ensures that even if the software fails or misbehaves, the current cannot exceed predefined safe values.
Polyfuse Protection (10 mA per IR Ring)
Each IR LED ring is protected by its own polyfuse rated to trip at around 10 mA. If, for any reason, the hardware current limit fails, the polyfuse limits the current by significantly increasing its resistance. Once normal conditions are restored, the fuse resets automatically.
These layered protections are designed to work independently of each other, with particular emphasis on ensuring current limiting does not rely solely on software. This is critical for safety and is guaranteed by the hardware-based limit of the AW9967DNR.
Finally, any excessive IR output would be immediately visible in practice: overexposure in the eye-tracking application would result in a washed-out or unusable image—clearly indicating that something is wrong and prompting immediate correction.