LED Driver Switching Frequency. The Higher the better?

[Geeker]

I try to understand the LED Driver and design my own. Just wonder if the switching frequency of LED the higher the better? I know than with higher switching frequency you can use lower inductor. The efficiency will increase as well. How about the disadvantages? Are there any?

 

[OVNI]

I try to understand the LED Driver and design my own. Just wonder if the switching frequency of LED the higher the better? I know than with higher switching frequency you can use lower inductor. The efficiency will increase as well. How about the disadvantages? Are there any?

Switching supplies/drivers generate more noise than linear drivers but, as you point out, are much more efficient generating less heat. Switching noise might not be a problem depending on your application. If you're going to modulate, worth looking into, IMO.

 

[Cyparagon]

Disadvantages would be cost of the IC, and increased complexity of dealing with parasitics at high frequency. A 47 nanohenry trace does nothing to the circuit at 30KHz, but can cause problems at 1MHz.

If you're going to modulate, worth looking into, IMO.

No. (The vast majority of) switching converters are not fast enough to modulate. All modulated drivers I'm aware of are linear for that reason.

 

[OVNI]

No.

Well I just couldn't decide which one to pick :thinking: soooo ...

and lastly

:wave:

EDIT: One more because that's kind of how I reacted at first ...

 

[Benm]

There certainly is an optimum in the frequency you operate a switchmode supply at. To give a couple of considerations:

Increasing operation frequency will:

- reduce the size of the inductor required (good)
- increase the loss at the diode that follows it (bad)
- increase the loss at the switching component, either in the ic or exernal (bad)
- increase losses in places where you don't want inductive effects like pcb traces (potentially bad)
- requires a smaller input buffer cap (good) but demands lower ESR of that buffer cap (bad).

This gives a rather mixed total effect, and hence you see switchmode power supplies operating anywhere from tens of kHz to a few MHz depending on application.

There also is the practical thing that inductors tend to produce audible noise if the switch frequency goes all the way down into the audio range. From an elecronics perspective this is not a problem, but it may be annoying for users.

One application where is is quite clear is on trains. Given the large amount of energy needing regulations on several (european) train systems a switching frequency is chosen that is in the audible range. This frequency actually isn't constant either, but varies with how much power the train needs, and you can often hear it playing a bit of a tune when departing a station or when using regenerative breaking on arrival.

 

[Geeker]

There certainly is an optimum in the frequency you operate a switchmode supply at. To give a couple of considerations:

Increasing operation frequency will:

- reduce the size of the inductor required (good)
- increase the loss at the diode that follows it (bad)
- increase the loss at the switching component, either in the ic or exernal (bad)
- increase losses in places where you don't want inductive effects like pcb traces (potentially bad)
- requires a smaller input buffer cap (good) but demands lower ESR of that buffer cap (bad).

This gives a rather mixed total effect, and hence you see switchmode power supplies operating anywhere from tens of kHz to a few MHz depending on application.

There also is the practical thing that inductors tend to produce audible noise if the switch frequency goes all the way down into the audio range. From an elecronics perspective this is not a problem, but it may be annoying for users.

One application where is is quite clear is on trains. Given the large amount of energy needing regulations on several (european) train systems a switching frequency is chosen that is in the audible range. This frequency actually isn't constant either, but varies with how much power the train needs, and you can often hear it playing a bit of a tune when departing a station or when using regenerative breaking on arrival.

Thanks for answering in detail. From what you said it seem that switching at higher frequency even worse.

 

[Benm]

From an electrical standpoint it often is, though practical application at low power doesn't really incur that much of a problem with higher frequencies. Given the (physical) size of the inductor increases for lower frequencies it always is a trade off.

Also i forgot to mention that the output cap has to be larger for a lower frequency, so that adds to the size of the whole thing too.

In many real world applications you are constrained to some envelope the driver needs to fit it, and just choose a frequency that allows reliable operation for those dimensions - lower if you have space and want efficiency, higher if you don't have much space but can deal with some losses.

Most led and laser drivers designed for flashlights and such hosts are space constrained. In things where space doesn't matter that much (say, led floodlights) you will find drivers with lower switch frequencies.

 

[LaserFever]

From an electrical standpoint it often is, though practical application at low power doesn't really incur that much of a problem with higher frequencies. Given the (physical) size of the inductor increases for lower frequencies it always is a trade off.

Also i forgot to mention that the output cap has to be larger for a lower frequency, so that adds to the size of the whole thing too.

In many real world applications you are constrained to some envelope the driver needs to fit it, and just choose a frequency that allows reliable operation for those dimensions - lower if you have space and want efficiency, higher if you don't have much space but can deal with some losses.

Most led and laser drivers designed for flashlights and such hosts are space constrained. In things where space doesn't matter that much (say, led floodlights) you will find drivers with lower switch frequencies.

Thanks Ben, very great info to learn :beer:

 

[gunsimon]

subscribed!