LEDs Are Photodiodes Too

[Editor’s note: Here is another really good one from the series “Rarely Asked Questions: Strange but True Stories From the Call Logs of Analog Devices,” guest authored by James Bryant. ]

Q: I read RAQ No. 45, “ Glass Diodes May See the Light – and Hum,” which discusses 100/120 Hz LF noise caused by a glass diode’s photosensitivity. I need a cheap photodetector, but a 1N4148 doesn't seem to work. How should I connect it?

A: 1N914/1N4148 diodes have enough photosensitivity to cause hum, but not enough to use as photocells. Their sensitivity is more in the infrared than the visible spectrum, and even in bright sunlight their photocurrent is only about 10 nA. Glass diodes are not a practical substitute for solar photovoltaic panels! Interestingly, an old-fashioned flashlight with an incandescent bulb excites two or three times more photocurrent in these diodes than direct sunlight, and a mains-powered 60 W incandescent bulb produces about 7% photocurrent modulation at 100 Hz. This suggests that the likely source of hum in RAQ No. 45 is from both incandescent and fluorescent lighting.

At about 2¢ each in large quantities, LEDs cost about five times as much as diodes, but they are much more sensitive as photocells. With the sun falling directly on it, the photocurrent of a red 5 mm LED (1000 mCd @ 20 mA) is over 20 μA. In the sunny tropics this might keep a clock battery charged. They’re not well suited for power generation, but LEDs are convenient photodetectors at about 10% of the price of purpose-made ones.

An LED’s spectral sensitivity depends on its color. They sense wavelengths shorter than or equal to their own emitted wavelength. This depends on the properties of the encapsulation. Light from colors that it absorbs will not reach the LED. White LEDs contain a phosphor to convert monochrome light to white light and do not make good photocells.

Manufacturers do not characterize LEDs as photocells, so minor design changes that have minimal effect on their behavior as LEDs may cause major changes in their characteristics as photocells. When using LEDs as photocells, characterize them yourself, and use conservative designs so that your circuit still performs well with any changes. This makes mass-produced circuits using LEDs as photocells demanding, but they are very useful in small batch or single system designs.

An elegant application is where an LED is driven by an analog microcontroller. The same LED may be used as a photodetector by disabling the digital output driving it and sensing its photocurrent output. If the microcontroller has dual-purpose analog input/digital I/O pins, like the ADuC7023 or the Atmel ATMega controllers used in Arduinos, this can be done with an LED and two resistors — and just one processor pin.

A semiconductor diode can be connected as a photocell in two ways: photovoltaic mode and photoconductive mode. Solar panels work in photovoltaic mode; light shines on them; the anode becomes more positive than the cathode; and a current proportional to the incident light flows in any circuit connected between the anode and the cathode. The diode is forward biased, and its capacitance is several times larger than its reverse-biased capacitance.

In photoconductive mode, light shining on a reverse-biased photodiode causes a photocurrent proportional to the incident light to flow. It is best to use photoconductive mode for AC signals as the frequency response is better, but light measurement in photovoltaic mode is very simple, as shown in simple photocell circuits.

— James Bryant has been a European applications manager with Analog Devices since 1982. He holds a degree in physics and philosophy from the University of Leeds. He is also C.Eng., Eur. Eng., MIEE, and an FBIS. In addition to his passion for engineering, James is a radio ham and holds the call sign G4CLF.


27 comments on “LEDs Are Photodiodes Too

  1. David Maciel Silva
    August 5, 2014

    This chat, if I can call it that clarified some doubts.

    My experience with photovoltaic cells, was not very good because I bought the substrate for mounting in route and they broke!

    But then I got a more enhanced feature, the system was not intended to energize a residence, was for an automatic welding (display with protection), which made a “burden” on a battery, thus increasing the useful life of the system.

    The light was captured by the PV panel through a circuit “charger pump” increased tension and made the display dark.

    It worked. but it was not very efficient!

    I believe that a functional scheme is presented in the figure below, for residential applications:


    The project explained that it was something very close to this:




  2. Davidled
    August 6, 2014

    I am wondering Power efficiency percentage in the summer time and winder time because sunset time of winter is very shorter than that of summer. Charger controller is very big. Engineer may design the small charging controller for this scheme.

  3. Steve Taranovich
    August 6, 2014

    @DaeJ–The says:

    A very safe rule says: calculate 30% safety buffer (especially in off-grid systems). Then if an 20A charge controller, withstands up to 52Voc (in an 24V battery system), the parameters of desired PV panels should be calculated ca -30%, about 14A/36Voc, with respect to the system safety. Another part is system efficiency, if highest efficiency is required, choose the MPPT charge controller.

    See this article 

  4. David Maciel Silva
    August 7, 2014

    I do not think I can really diminuito the charger, since demand a reasonable amount of energy.

  5. Davidled
    August 7, 2014

    12 V Bulb may be connected to battery instead of charging controller. I am wondering what type charging controller is used. If brand name/part number is given, it might be reviewed.

  6. Victor Lorenzo
    August 8, 2014

    Thanks Steve for bringing us this story. I think some of us have made this kind of “wrong component use” in the past. I enjoyed playing with decaped MP41 and P423 germanium transistors as light sensors many years ago.

    It will be very interesting reading more stories like this one.

  7. David Maciel Silva
    August 8, 2014

    You can not always get everything right 

    But if these components used for what function?

  8. Victor Lorenzo
    August 8, 2014

    On application was as a small light sensing cell producing about 100mV when exposed to light for an automatic blinking light beacon. Another application was for the target sensor part in a light shooting gun game.

  9. Steve Taranovich
    August 8, 2014

    Thanks Victor—I'm glad you liked it. I hope that Planet Analog can be the Designer Engineer's source of highly diversified, high tech information for their designs. We ant to make their life easier

  10. etnapowers
    August 9, 2014

    Is the 100mV voltage a DC value? Does it depend from the impacting light energy value?It's a very interesting feature.

  11. etnapowers
    August 9, 2014

    I wonder if the cells can be put in series to create a DC stable voltage of 1~2 V

  12. eafpres
    August 11, 2014

    @Victor–“I think some of us have made this kind of “wrong component use” in the past.”

    I know of a couple of these wrong uses that could backfire.

    It turns out that at some frequencies, you can use a wire-wound inductor ss an antenna.  The probem is that while the manufacturer might guarantee the inductancce to the stated tolerance, they don't guarantee the characteristic impedance or resonant freuqncy when those are far out of the range specified for a component.  What you then have is a very high-Q component and if it varies much at all, suddenly the antenna doesn't work and the poor component must live out its days as a lowly inductor.

  13. eafpres
    August 11, 2014

    @Steve–I'm sure you know but I hope some of the others here will go and read the comments about this article on the sister site All LED Lighting.  Here is a link to a comment I mad there:

    One time sensor

  14. JamesBryant
    August 13, 2014

    These issues are discussed in the article “Photodiodes and Other Light Sensors” which is linked to the RAQ on the ADI website, and will probably be linked from the copy here very soon. In the meantime visit

  15. Victor Lorenzo
    August 18, 2014

    @etnapowers, the 100mV was a DC voltage value, unloaded, and it was dependant on the impacting light (sun light) and its incident angle with respect to the die.

    Windowed transistors degraded too fast after mechanization, I guess it was due to hummidity and dust since I did not cover the window.

  16. Victor Lorenzo
    August 18, 2014

    @etnapowers, I never used them in series as the base-emitter and base-collector junctions acted as simple P-N junctions (diodes) when not exposed to light.

  17. eafpres
    August 26, 2014

    Does anyone know the photo-sensitivity of IR diodes used for active illumination for night vision cameras?

    Applications in security would be interesting if the LEDs could generate enough current when a warm body was close enough, then the system could power up and take images.  This would allow a much longer battery life.

  18. geek
    August 31, 2014

    “Applications in security would be interesting if the LEDs could generate enough current when a warm body was close enough, then the system could power up and take images.  This would allow a much longer battery life.”

    @eafpres1: If I understand correctly, the savings in battery life would come about because the sensors to detect a human body won't be running all the time and instead the LED would generate the current and power the camera? Or is there something more to it?

  19. geek
    August 31, 2014

    “I hope that Planet Analog can be the Designer Engineer's source of highly diversified, high tech information for their designs. We ant to make their life easier”

    @Steve: We can even have a section to capture blogs related to the innovative applications that designers in this community have come up with in their experiments with curcuits. That would appeal to a lot of enthuisiasts like me who'd like to get ideas for innovative gadgets and devices.

  20. eafpres
    August 31, 2014

    @tubair–“won't be running all the time and instead the LED would generate the current and power the camera?”

    What I have in mind is there are 2 modes; sleep mode where the camera is off but there are sensors to detect heat signature then wake the system to active mode, where the camera is on and is using the on-board battery.

    So the battery is used only for taking images, otherwise the system acts like a passive sensor.

  21. Steve Taranovich
    August 31, 2014

    @tzubair—Absolutely, we do have some really good experts blogging for us here on PA

  22. etnapowers
    September 8, 2014

    @Victor: nice post, thank you for the informations. I think you're right about the degradation, and I wonder if by a appropriate exposure to light the voltage produced can be stabilized and the overall reliability can be enhanced. the cells might be utilized as diodes in a circuit and as voltage source in case of exposure to lights, this feature can be useful to build photodetector application circuits.

  23. Victor Lorenzo
    September 8, 2014

    @Blaine, perhaps a thermopile based sensor combined with an ultra-low power microcontroller (we have several available from Microchip, Texas, Nxp and Freescale) would do the job for you. With different optics you could cover a wider or narrower area.

    The sensor output could be connected to one comparator input, one cheap SPI DAC could produce the reference thresshold and the CPU could be in low power state and wake-up on comparator output change.

  24. eafpres
    September 9, 2014

    @Victor–sounds like a plan.  It is amazing that things impossible a decade ago are solved by commodity parts today.  Kids toys have better sensors than spacecraft that went to the moon.  

    There does seem to be more going on in IR sensors these days.  They used to be for TV remotes and a few other remote controls; many of those are being replaced by RF, yet IR sensors now have roles in gesture recognition etc.

  25. etnapowers
    September 18, 2014

    @Victor:  I wonder if there is a chance to write an equation containing the dependence of the voltage from the incident angle and the intensity of impacting light, it could be very interesting.

  26. JamesBryant
    October 30, 2014

    If the LED manufacturer provides an equation for intensity v angle off-axis the same equation should be valid for sensitivity as a photocell. If he simply provides a chart of intensity v angle then you would have to scan (or measure manually) and use a curve-fitting program. [I'm not able to provide URLs so I haven't bothered to Google for one, but there are several out there.]

  27. etnapowers
    November 7, 2014

    As fitting software to interpolate the data I guess that excel is enough. the sensitivity might be evaluated by the derivative of the intensity curve, to evaluate the speed of reaction of the system to a variation of the angle of incidence.



Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.