I have developed a product in which LEDs spectrum is sensed by CCD (TCD1304) using ADC of pic microcontrollers. It is being observed that LED spectrum changes with time. We concluded that it could be due to change in temperature and change in forward current (consequence of change in temperature).

The datasheet of one of the LED is over here. The LED is being driven using NPN transistors. I am using PWM to control the intensity of LED and the maximum current that flows through LED is 12 mA. The datasheet of LED shows change in intensity with the temperature at constant forward current. The graph which states that in given below and you can find it in the datasheet of the LED.

enter image description here

My question is after starting the system (and starting the LEDs), how to nullify or counter the change in intensity of LED with time. We know there are many products in which LEDs is being used and sensed. I just want to know how industries or experts tackle this type of problems and what are industries best practices to be followed to avoid or counter such effects?

The schematic of my LED circuit is:

enter image description here

Note: NC in the schematics means a zero ohm resistor.

UPDATE: We have tried few experiment with constant current source (LM334) and results were a bit better but still they were not acceptable. I have designed a new PCB with PWM-Controlled constant current source. I will test it in a week and update my question with the results. I liked the answer given by Misunderstood alot and will try LEDs suggested by him.

  • \$\begingroup\$ (1) What is an NC resistor? (2) Why have you shown two identical switches? \$\endgroup\$
    – Transistor
    Nov 6, 2017 at 8:56
  • \$\begingroup\$ NC is not connected Resistor. A Wire is placed at it's position. \$\endgroup\$
    – abhiarora
    Nov 6, 2017 at 9:29
  • \$\begingroup\$ There are two leds in my system. That's why I have two switches to control them. \$\endgroup\$
    – abhiarora
    Nov 6, 2017 at 9:29
  • 1
    \$\begingroup\$ NC in that case implies open-circuit if there is nothing connected. "Zero-ohm" or "LINK" would be much better. \$\endgroup\$
    – Transistor
    Nov 6, 2017 at 9:31
  • \$\begingroup\$ I have edited my question. \$\endgroup\$
    – abhiarora
    Nov 6, 2017 at 9:47

1 Answer 1


LEDs intensity and chromaticity do deteriorate over time.

This LED was designed in 1998 and when GaN LEDs were first being introduced. It has horrific temperature stability. A current blue LED will lose 10% max from 25°C to 150°C.

LEDs have come a long way in the past 20 years. Use a newer lighting LED.

I do not know enough about what you are doing with the LED to make any recommendation.

The best blue LEDs today are deep blue 450nm as they are used for most white LEDs and that is where the manufacturers R&D goes.


For full spectrum you could use a white LED. The white with the widest range I know of is the Luxeon Crisp White. You would have to add IR and may have to add UV (Crisp White has a peak at 405nm). enter image description here

LED engine, Lumiled, Cree, and OSRAM have Deep Blue to Far Red in their horticulture market.

LED Engine Inc. has some multi color LEDs Brochure: LuxiGen Brochure
They also have some interesting LEDs for Horiculture

The best option would be to make a custom PCB with all the available wavelength. I would use OSRAM they have wavelengths 450 470 528 590 617 623 660 730. OSRAM Horticulture Flyer
OSRAM Infrared Infrared LEDs

Lumileds has some new UV and IR:
LUXEON IR Domed Line Product Datasheet
LUXEON UV U Line Product Datasheet
LUXEON UV FC Line Product Datasheet

Can you suggest smaller LEDs?

Lumiled makes the smallest case (1.9mm x 1.9mm) LUXEON C Color Line Product Datasheet

There are only a few manufactures that make far red (≈720nm). When looking for LEDs with a wide spectrum, I first search for far red. The target market for far red is horticulture which requires high power LEDs. Very limited market.

If by smaller you are looking for less power these LEDs can be driven with as little as 10mA. Also the distance from the LED to the detector is a significant factor in radiated flux (inverse square law).

To see what is available, you can do a parametric search on Digikey
Digikey parametric search far red LEDs

Can you suggest the LED covering IR range as much as possible?

LUXEON IR Domed Line Product Datasheet

Typical normalized power vs. wavelength for LUXEON IR Compact Line

The 860nm and 940nm are standard parts. If you need to fill in the valley at 880nm, these emitters are binned by wavelength. The problem is they are special order and you would likely have to buy a reel of them and wait 10 weeks.

enter image description here

It would be difficult for me to chose LEDs for you as I do not know intensity and your other requirements.

The other option to find what you need is by doing a Digikey search by wavelength. From this link choose the wavelength in the "wavelength" parameter column. Then sort the results by whatever parameter is most important.

Digikey Infrared parametric search

  • \$\begingroup\$ Thanks for your answer. We want to cover wavelength from 380 nm to 950 nm. Do you know any LED or any suggestion? \$\endgroup\$
    – abhiarora
    Nov 13, 2017 at 16:56
  • \$\begingroup\$ See my update to my post. \$\endgroup\$ Nov 14, 2017 at 22:29
  • \$\begingroup\$ Excellent! Thanks for getting back to me with a great answer. The LEDs you have suggested are quite impressive. But they are quite large. Can you suggest smaller LEDs?? \$\endgroup\$
    – abhiarora
    Nov 16, 2017 at 8:45
  • 1
    \$\begingroup\$ Can you suggest the LED covering IR range as much as possible? \$\endgroup\$
    – abhiarora
    Nov 16, 2017 at 9:50

Your Answer

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

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