I need to design a led driver of up to 1A, to drive some infrared and near infrared LEDs.

Due to the nature of my application, the intensity of the light should be as stable as possible, as fluctuations in the radiation intensity cannot be allowed.

So far I have considered using Buck Converters as the LM3405 or the TPS92515 from Texas Instruments, however on any DC / DC switching converter, there is always a ripple at the output.

1.- Does this mean, that there will be variations in the intensity of the light ?

2.- If so, Is it better to go for a Linear Regulator solution? (using a MOSFET, Op amp etc)

  • 1
    \$\begingroup\$ The better solution depends entirely on how much noise or ripple or voltage variations your target circuit can deal with without performance being affected. This is down to you to state hence, 1) yes, 2) maybe. \$\endgroup\$
    – Andy aka
    Mar 11, 2020 at 15:42
  • \$\begingroup\$ Why does the intensity need to be so stable? Are you experiencing visual artifacts, perhaps from photography? Or electrical artifacts, such as false triggers from beam interruption? \$\endgroup\$
    – rdtsc
    Mar 11, 2020 at 17:10
  • \$\begingroup\$ "I need to design a led driver of up to 1A" - Do you intend to vary the LED current? Does the long-term brightness need to be constant? \$\endgroup\$ Mar 11, 2020 at 18:02
  • \$\begingroup\$ I need the led intensity as stable as possible since it will be used for a medical application. The current will be fixed, but since different LEDs have different Vf and If, I wrote this line "I need to design a led driver of up to 1A", Now the application is somehow like an oximeter where IR and NIR LED is used, and the absorption of the hemoglobin chromophores is measured, but for legal reasons I can not completely describe it. However, in a phisiology level, fluctuations in the radiation intensity cannot be discriminated from those due to changes in chromophore concentrations. \$\endgroup\$
    – MoisesDS
    Mar 12, 2020 at 10:06

4 Answers 4


As the name implies, a SMPS relies on the principle of transferring energy from input to the output cyclically, and this is accomplished via a switching device and a corresponding driver. Thus, there will always be high frequency components (main switching frequency + harmonics + higher order frequency components) at the output of your power supply, but it is up to the designer to design how large their amplitude are. This is done by simply filtering the unwanted signals out.

Now to the my suggestion:

If you are dealing with low power applications, a LDO would be the best pick, since it produces less noise (no switching). However, a SMPS would definitely be more efficient for higher power applications at the cost of requiring a bit more effort to design properly.


The ripple on a buck converter will have a negligible effect on LED intensity, and will be far to high a frequency to be visible in any case. Note that such a converter could also be set up to operate as a constant current source (also with negligible, high frequency ripple), providing greater control of the intensity over all conditions.

  • \$\begingroup\$ Considering that he's trying to drive IR LEDs, it's unlikely that they're worried about the intensity ripple being visible to the human eye. For scientific purposes (a much more likely application), it's entirely possible that minute current variations will be detrimental to whatever they are trying to do. \$\endgroup\$
    – Jarrett
    Mar 11, 2020 at 17:35

You could add a power lowpass filter to remove the fundamental and higher frequencies. There are a ton of examples and calculators if googled. The simplest may be a series resistor of a low value (maybe 1 Ohm) and a parallel capacitor of say, 1000µF. The resistor drops a little power while operating, but the cap conducts AC, causing the AC component (ripple) to be forced across the resistor instead of whatever comes after it (LED's in this case.) The larger the capacitance (and lower the ESR rating) the better this will work.

Note that some switching supplies have a limit to the amount of capacitance they can drive before it affects their stability.

There are also inductor-capacitor or LC filters, and while they do filter better, they also have a tendency to oscillate. This may be corrected with proper damping but honestly, all this is likely overkill for powering LED's. The amount of light intensity change due to ripple is going to be very small even without filtering.


I'm inferring that LED stability is important for your application. The ripple will definitely affect the output with a noise spectrum of harmonics.

My suggestion? Use both. Use DC-DC to make the primary voltage, followed by an LDO (perhaps wired as a current regulator?) to post-regulate the ripple.


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.