I am doing Project on Wireless Optical Communication(Visible Light Communication). I am using Nicha NSPW510 DS White LED at transmitter and Vishay BPW34 PIN photodiode at receiver. The problem is with LED TURNOFF TIME. It is about 30 usec s. and i am not able to increase my speed beyond 10 kbps.(distance 1 - 2 meters). Could u please suggest any LED type, which supports upto 1 Mbps. ( I have tried to reduce turnoff time of LED by driving it into reverse bias region, but it did'nt helped me.)

  • 1
    \$\begingroup\$ I would suspect that this issue is more to do with either the LED driver circuit or the photodiode & receiver circuit than with the LED. The LED would be last on my list of suspects... \$\endgroup\$
    – brhans
    Commented Mar 29, 2016 at 11:13
  • \$\begingroup\$ Thanks @brhans. I am directly applying Digital signal (Logic 1 @ 3.7v and Logic 0 @ 0.0v) to LED. Is it wrong to do ? \$\endgroup\$ Commented Mar 29, 2016 at 11:20
  • 2
    \$\begingroup\$ @brhans it is a white led using a phosphor, so the time that the phosphor continues to glow after excitation has ceased will be much greater than a direct LED. \$\endgroup\$ Commented Mar 29, 2016 at 11:20
  • 2
    \$\begingroup\$ Test the hardware first. Take a signal generator and an oscilloscope. Give a square pulse to LED and observe the response of receiver. Go on increasing the frequency. See what max frequency you are getting. If you are able to get a higher frequency, then check your wiring. Make them short and neat. After that try optimizing your code. There is a possibility that your code loop is not executing fast enough. \$\endgroup\$ Commented Mar 29, 2016 at 11:24
  • 1
    \$\begingroup\$ @Sirraajithprakash its not so much wrong as probably not the best solution. I would use an N-channel MOSFET such as an 2N7002A. Your logic controls the MOSFET gate. You will also need a current limit resistor in series with the LED. \$\endgroup\$
    – Steve G
    Commented Mar 29, 2016 at 11:32

2 Answers 2


White LEDs are the wrong choice for communication. This is because they are a two-part system. Inside is a blue or near UV LED and a mix of phosphors. The LED excites the phosphors, which then emits various colors of light. The result is balanced to look roughly white.

The reason for the lag is that the phosphors will continue to re-emit the energy for a short time after the LED is turned off.

Use anything that is a bare LED. This also has the advantage of all the light being within a very narrow spectral band. That makes it possible to filter out much of the ambient light at the receiver.

IR LEDs are usually used for this since the ambient level at near IR tends to be lower than the visible level, IR LEDs are efficient, and materials are available that block most visible light but let the IR pass. Wavelengths are usually in the 900-1000 nm range.

  • \$\begingroup\$ @ Olin Lathrop . Sir, Phosphors LED will emit upto 30 microsecond ? . If it is so, i will use Blue LED of narrow spectral range. will it be okay ? \$\endgroup\$ Commented Mar 29, 2016 at 11:47
  • \$\begingroup\$ @Sirraajithprakash: Why still blue, and not IR? \$\endgroup\$
    – arc_lupus
    Commented Mar 29, 2016 at 18:45
  • \$\begingroup\$ @arc_lupus. ..Because, my aim is to use only "Visible light" for Communication. The same light source(an array of LED s) both for Vision and for Communication. \$\endgroup\$ Commented Mar 30, 2016 at 11:19

In addition to switching to bare LEDs rather than phosphor-coated, you may want to have a look into the literature on drive circuits. Here's an example to get you started High-Speed LED Driver for ns-Pulse Switching of High-Current LEDs Halbritter et al. (from Osram, so they should know a thing or two about their LEDs).

And here's another A LED-based calibration system for the external strings of the NT-200+ deep-water neutrino telescope on Lake Baikal, Vasiliev et al.

Both of these are about high-brightness LEDs and shape the current pulse. I may want to apply these techniques to driving reactions in an experiment.

Either the latter paper or another I can't find for now found significant variation in the speed of even nominally identical LEDs in the ns regime. Fast photodiodes are necessary but quite easy in comparison.

  • \$\begingroup\$ Thanks @Chris H. I got a very valuable information from the links you given. and also in the first paper "The circuit reduces significantly the optical rise and fall time of LEDs from 10/15 ns, respectively, down to 2.6 ns" . But i am trying to modulate LED in 1 usec range only. In that case can i drive LED directly ? as shown in Fig 1 (a) link High-Speed LED Driver for ns-Pulse Switching of High-Current LEDs \$\endgroup\$ Commented Mar 30, 2016 at 12:03
  • \$\begingroup\$ You may well be able to drive it directly with a FET and a well-designed driver (plus appropriate choices related to cabling/PCB traces) as in Fig 1b. If your rise/fall times are ~1% of your pulse width you should be OK for comms (correct me if I'm wrong there, not an expert). I think you'd need <0.5us pulses to drive 1Mb/s. \$\endgroup\$
    – Chris H
    Commented Mar 30, 2016 at 13:04
  • \$\begingroup\$ Thanks @Chris H . I will try with FET driver. And the Bit rate is given by Rb=1/Tb. Where Tb is duration of a "Bit" (either logic 1 or logic 0). So if my Logic 1 Pulse is of 1usec duration, that means i am transferring with 1Mbps. correct me if i am wrong. Thank you. \$\endgroup\$ Commented Mar 31, 2016 at 6:17
  • \$\begingroup\$ It's a long time since I studied this but you'll probably have trouble staying in sync if a series of 1s is a constant high level. In fact, unless you control all illumination at that wavelength (and filter the light well), you'll have trouble distinguishing your on and off states at all. IIRC from years ago you'd be better off using pulse edges as bits. Which is all to say that from the little I remember it's going to be harder than you think. \$\endgroup\$
    – Chris H
    Commented Mar 31, 2016 at 6:47
  • \$\begingroup\$ When i have given some voltage (or current) the LED starts illumination(logic 1 state). when no voltage(or current) it will not illuminate. That s what i am thinking. On-Off keying (Digital ASK). The problem is that how fast the LED will on (time for raise) and how fast it off (time for fall). The voltage pulses are very sharp and i had no problem with electrical waveforms. When i used white LED it gives 30 usec fall time. As Pete Kirkham said, White LED made of Phosphors will glow for a while after turn off. So i am thinking to work on LED part. and the problem is if i choose .... \$\endgroup\$ Commented Mar 31, 2016 at 9:51

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