I am currently trying to make a circuit to drive an LED with smallest amount of fluctuation as possible.

Background: Using an LED and spectrometer together to analyze small molecules. I can use the spectrometer to record the intensity of a light over time and that is how I read noise in an LED. I have tried a variety of LED linear drivers, and I am sticking to 5mm LEDs. My circuit consists of one driver (right now im using MAX16815) and LED, and some bypass capacitors at the IN and OUT pins.


With a DMM, I found that the current decreases/increases for 15-20 minutes before reaching a steady state.

-How can I reduce this phenomenon?

I have a 35-100mA output 3.5% accuracy LED driver with a max 100mA 5mm LED. So far, I have been working on breadboard to do testing and analysis so there could be a source of error in there. I use an external resistor to control the current output of the driver. It currently takes 15 minutes to reach any current I choose with a standard deviation of 0.007 mA. I want to reach a standard deviation of 0.001 mA just to be sure.

What can I do to reduce variations in the intensity of the LED?


closed as unclear what you're asking by PlasmaHH, brhans, Trevor_G, Olin Lathrop, RoyC Jan 31 '18 at 15:38

Please clarify your specific problem or add additional details to highlight exactly what you need. As it's currently written, it’s hard to tell exactly what you're asking. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.

  • 5
    \$\begingroup\$ "What can I add/replace in my circuit" that question is really easy to answer since we all know what is in your circuit... \$\endgroup\$ – PlasmaHH Jan 31 '18 at 15:14
  • 3
    \$\begingroup\$ A good linear current source will provide better stability of the drive to the LED, but if you want that kind of stability you would really need some sort of feedback measurement of the light emitted itself. The issue is the temperature of the LED changes the emissions, you would need to close that loop to get fast and accurate steady state output. \$\endgroup\$ – Trevor_G Jan 31 '18 at 15:25
  • 2
    \$\begingroup\$ A linear constant current source controlled by a photodiode feedback would be ideal here. \$\endgroup\$ – winny Jan 31 '18 at 15:27
  • 2
    \$\begingroup\$ I was involved in creating LED-based light sources intended as "lamp references" for optical work. Even if you have a perfect current source, LED output varies with temperature. So they must be operated at a stable temperature. We chose to heat them to about 75C using a closed loop control. That solved the temperature drift problem. But LEDs themselves aren't consistent; even those cut from the same wafer aren't. We had to pre-stress thousands of them, holding them at a stable temperature, running them with a current source for days, monitoring the optical output and logging results. \$\endgroup\$ – jonk Jan 31 '18 at 16:27
  • 1
    \$\begingroup\$ We tested the LEDs for much longer than a few days. Those that stabilized in less than two days, were then stable over much longer periods. Those that didn't stabilize within two days, never were reliable even after "baking for weeks." So we determined that we could tell what we needed to know within two days and used that as a selection cutoff. \$\endgroup\$ – jonk Jan 31 '18 at 16:30

Search for CC sink or source. Or use a LM317 as such from 5V for 20mA.

LED brightness is a function of current and long term, aging.

Voltage has a NTC so any temp rise will reduce voltage -x mV/'C so constant voltage or even PWM with CV will affect intensity.

Even temperature rise affects efficacy to some extent. Consider multiple 20Cd <=30deg LED's operating at <10mA. Consider temperature stability and short term aging or burn-in process.

Only use PD + R , not photo transistor and block all stray light.

What stability range do you need from drift? > 30dB? > 60dB ?

How will you equalize emission curve?

Do you want to regulate output using dual PD detector in a closed loop? That is best way.

One using beam split closed loop feedback, Two for sample output detector. e.g. Using 1% emitter sample with fiberoptic.
( not shown )

I assume using Bragg Effect... then get plots like this.

enter image description here

However I might suggest PNJ4K01F or AMS-302 Panasonic logarithmic light detector with optical compensation...maybe for comparison.


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