I am performing eye safety calculations on a device and I have hit a roadblock. To perform the calculations I have been following the IEC 62471 "Photobiological safety of lamps and lamp systems" Standard.

I am using three IR LED arrays. One has 34 IR LEDs and the other two have 3 IR LEDs. The arrays utilizes the SFH 4243-Z IR LED from OSRAM.

OSRAM Provides an eye safety document in order to calculate if your set up is potentially hazardous to the eye however, their examples when calculating the burn hazard has me confused.

I will post an example they provide that is the most similar to my device. However I will ask my question first because the snippet may be confusing at first.


  • They have 20 X IR LEDS , which operate for > 1000 seconds
  • Each LED has a Radiant intensity of 160 mW/sr
  • In total the array produces a radiant intensity of 3.2 W/sr
  • When they perform the retinal burn hazard calculation they only use 160 mW/sr and not 3.2 W/sr. (I have highlighted this in red in the snippet below)
  • Because they only use 160mW/sr, the value of L_IR is calculated to 58.7 mW/mm²/sr, which falls within the exposure limit of 545.5 mW/mm²/sr.
  • Therefore, this would pass and be NO RISK
  • However, if this value is multiplied by 20 for the entire LED array (3.2 W/sr) you get 1174 mW/mm²/sr which would FAIL the limit
  • This calculation is used in this way for all of their examples.

Therefore, my question is: Why is the radiant intensity of only one LED used for the burn hazard calculation and not the entire LED Array? Is there a scientific reason for this?

Here is the example:

enter image description here enter image description here

Thank you

(For additional information Equation 10 has been appended below) enter image description here

For a description of my IR LED arrays, in relation to the eye, here is a simplified picture I have drawn for my setup. Each LED has a typical radiant intensity of 11 mW/sr

enter image description here

  • 1
    \$\begingroup\$ I'm not familiar with IEC 62471 so I'll let someone else answer that, but optically the reason they only use one LED is that your eye can only look at one of the diodes at a time. Therefore, for each area on the back of the retina, light can only arrive from one diode at a time. Adding more diodes will illuminate more points on the back of the retina, but for thermal effects you only care about power per area which does not increase if you add equal amounts of power and area at a time. \$\endgroup\$ Jul 13, 2020 at 13:56
  • \$\begingroup\$ @user1850479, that is very interesting. I would assume that if you doubled the amount of IR LEDs pointed towards the eye, the thermal effect would double as well since the area of the eye does not change. Would the radiant energy from each additional LED not constructively add to the rest of the LEDs? \$\endgroup\$ Jul 13, 2020 at 14:08
  • 1
    \$\begingroup\$ renesas.com/eu/en/doc/application-note/an1737.pdf defines the retinal , back-of-the-eye, hazard range as 0.011 radian or 0.063 deg \$\endgroup\$ Jul 13, 2020 at 14:11
  • \$\begingroup\$ @TonyStewartSunnyskyguyEE75, yes. This if I am not mistaken, is called the angular subtence. However, would this not mean that each LED would cover an area of the retina equivalent to an angle of 0.011? And if so, calculating for a worst case scenario, there is potential for overlapping on this area (Therefore, increasing the thermal effect of the array)? \$\endgroup\$ Jul 13, 2020 at 14:17
  • 1
    \$\begingroup\$ @ScottBruton Power/area on the cornea doubles, but power/area on the retina stays constant. If two spatially distinct objects could hit the same point on your retina at the same time, then no image would be formed on the retina and you would be blind :) \$\endgroup\$ Jul 13, 2020 at 15:29

2 Answers 2


The retina (back of the eye) depends on the angular subtend to each "spot" on the retina for spot burn hazards.

The cornea (front of the eye) hazard depends on accumulated IR light on the entire surface for causing cataracts.

The angular subtend, a of the LED here is 0.0015 rad.

For small angles, tan (a) = x/y = 0.3mm chip/200 mm range= 0.0015 rad.

The LED radiates from a 0.3mm chip in a 3 mm case and with no gap side by side would indicate a subtend spacing of 0.015 rad which is close but greater than the hazard range of the retina.

The retinal angular subtend safety hazard range is defined by this as 0.011 radians.

I would think you want a safety factor of at least 2 here, below the legal limit for a matrix source for thermal reasons.

  • \$\begingroup\$ thanks for you response. So there is actually a safety limit for the angular subtense? What would happen if the angular subtense was greater than the limit? Reason I am asking, is that I have 3 arrays. 1 array with 34 LEDs at a distance of 100mm, and 2 arrays of 3 LEDs, at a distance of around 20mm on the sides, angled towards the nose. This 20 mm distance could push the angular subtense over the safety limit. I will update my post with the setup of my arrays relative to the eyes to help visualize this. \$\endgroup\$ Jul 14, 2020 at 7:25
  • \$\begingroup\$ Additionally, what is the reason for calculating the actual angular subtense from the source. In the example above, the actual angular subtense of the source is never used. The limits of a = 0.0017 rad (100 us) and a = 0.011 (1000 s) are used in determining the limits, however, the actual calculated a = 0.0015 rad is never used in the calculations? \$\endgroup\$ Jul 14, 2020 at 8:49
  • \$\begingroup\$ I'm cross-eyed now. maybe others can focus better on the results ;) \$\endgroup\$ Jul 14, 2020 at 13:06

Why is the radiant intensity of only one LED used for the burn hazard calculation and not the entire LED Array? Is there a scientific reason for this?

Yes. The lens of the eye focuses the image of each LED in the array to a different place on the retina. In other words, there is no point on the retina that is receiving light from more than one LED at a time.

Contrast this with the unfocused light falling on the cornea, each point of which does in fact get light from all of the LEDs.

The main danger to the retina with IR is that the iris of the eye does not react to bright IR sources. In an otherwise dark environment, the pupil remains wide open, allowing more of the IR to pass through the aperture.

According to Wikipedia, the human pupil can be as large as 9 mm. But the number you really want to use is the "entrance pupil" -- the apparent size of the pupil as viewed from outside the eye, which is larger because it is magnified by the cornea. I would use 10 mm, a number that telescope and binocular manufacturers use when designing eyepieces. This is 2x the area of 7 mm.

  • \$\begingroup\$ Thank you for your response. So in this application, it will be in a dark environment when the IR illuminates the eye. What sort of safety factor would you suggest I use? I use a 7 mm pupil diameter in all calculations for worst case scenario just to be safe. \$\endgroup\$ Jul 14, 2020 at 7:17
  • \$\begingroup\$ Hey @Dave Tweed. So I am using 7mm because this is what the IEC 62471:2006 standard says I should use. "When the luminance of the source is low, i.e. infrared radiation is present with little or no visible stimulus, the EL is based on a 7 mm pupil diameter." Eye safety from OSRAM "For IR-A light the visual stimulus of the eye is very low. That means that the aversion response, which normally protects the eye from excessive continuous irradiation, does not work. As there is no trigger of the iris contraction, we have to calculate with the full 7 mm pupil diameter that collects the light." \$\endgroup\$ Jul 14, 2020 at 12:01
  • \$\begingroup\$ Well, if that's what the standard says, then go with it in terms of your compliance calculations. But I stand by what I said in terms of actual human pupils. \$\endgroup\$
    – Dave Tweed
    Jul 14, 2020 at 12:07
  • \$\begingroup\$ I 100% agree with you, and I am only using 7 mm for compliance with the standard. I really appreciate your input, and I will include 10 mm For worst case \$\endgroup\$ Jul 14, 2020 at 17:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.