Need help to design led drivers with dimming capabilities

[sijeo.philip]

Hi,
I am trying to design a LED Driver of 12W(24v/0.5A) output , where I can keep the voltage constant and vary the current. i.e. I would like to understand if we can vary the current by introducing the current control loop in the SMPS /Driver itself...
i.e. Inner control loop which would supply the desired current to the LEDs (Thus enabling dimming )
and outer voltage loop that would be keeping the voltage constant at minimum current to maximum current.
I am attempting to understand if we can achieve dimming of LEDs without introducing transistors in the LED string and use PWM to get the dimming effect.

 

[red_alert]

You just have to check for the output current only. The voltage doesn't care, as the LED will self adjust the voltage across its terminals.

By using a PWM signal and using the output current as feedback signal you could accomplish your task (LED dimming).

Think about LED as a simple diode - you don't have to set the voltage across its terminals.

 

[dick_freebird]

A PWM controller which breaks out the current mode
control section (like error amp out is pin-strapped back
to ISET input) could work nicely. Or if the error amp can
work with a reference that goes down to ground. These
are not super common especially in more integrated
DC-DC / POL regulators but you may find some.

Many illumination LED drivers offer PWM dimming
(esp. the cheapest ones as this is easiest to do).

 

[sijeo.philip]

if I leave the voltage unmonitored/unregulated , reducing the current might drop the voltage and increasing the current might increasing the voltage, leading to undervoltage or overvoltage condition.. which might be harmful to the LED string I assume. Is it possible to find a proven/working design in which current controlled dimming is done which I can consider as a reference

 

[red_alert]

If you keep the LED current under the maximum admisible value I _guarantee_ there will be no overvoltage across its pins.

The forward voltage (from any LED datasheet) just helps you to calculate the LED current. By example, if you run the LED from a constant voltage source (5V), by substracting the forward voltage (2.5V) from 5V then divide it by the desired LED current (20mA) you get the LED series resistor.

You can't control a LED forward voltage (nor any other diode drop voltage); you're actually controlling its forward current.

 

[chuckey]

The voltage drop across the LED is a function of the current through it and its temperature. LEDs are specified at a current, the forward voltage drop is a notional figure, and gives the designer an idea of what the likely maximum value is likely to be. And it gives chancers an opportunity to parallel diodes.
Frank

 

[asdf44]

As others said, you can't have two competing control loops and you only really want the current loop. Drive the current and let the voltage do what it wants to do.

Are you against going linear? With 24V a linear circuit isn't a terrible option if you're driving a string of LED's.

Here is a TL431 based circuit which creates a fixed current (25mA here) where brightness is controlled by the PWM input.

Another option would be to skip the PWM and instead bias the TL431's feedback. Just insert a resistor between R1 and the TL431 ref pin and then bias that with a variable voltage (more voltage = less current).

It's linear, but it only needs 2.5V of the 24V available (1.24V if you opt for lower voltage TL431 variants). Though the key to efficiency would be stuffing in multiple LED's that would come close to using up the remaining available 21.5V. The mosfet eats up unused voltage and will get very hot if this amounts to more than a couple volts with 0.5A going through it (at best you probably still need a heatsink).

 

[dick_freebird]

I believe that if you chop it fast enough, the persistence
of vision (and phosphor "hang time", on a white LED)
will let you go without an inductor. You might be able
to get away with a simple emitter ballasted NPN switch
or something simple as that, driven by a logic gate, on
the low side and varying the pulse width of that fixed-
"on"-current train could get it done. Emitter resistor
sets the current as (VlogicH-Vbe)/Re. If you have some
other thing that would source a variable logic pulse width,
this might be your lowest incremental cost approach.

 

[red_alert]

@dick_freebird is right - that's the usual LED dimming circuit (a simple chopper with a variable pulse width).

The only design parameters are the maximum LED current (at 100% mdulation) and the minimum chopping frequency (to avoid flickering).

Few useful links:

**broken link removed**

https://www.fairchildsemi.com/datasheets/FA/FAN5346.pdf

 

[treez]

these controllers are nice for you and give you advice.
by the way you do not give your input voltage

if you have low vdc in then what about zxld1370, zxld1360 or hv9910b, or one of the supertex ones, lm3402 by ti.com, lm3409.....lm3407...etc etc

If you do pwm dimming then if you are using hysteretic control you don't need a series fet for pwm dimming, but if you have a feedback loop then you do need it otherwise the inductor current overshoots every time you come back on.

 

[sijeo.philip]

Thanks a lot for the advice and suggestions... Did really give me a direction.. will get back with more findings /queries from my side.

 

[leehying]

I = V/R, V constant, R(i.e. LEDs) constant, how can I vary?!

You may need a fast transient DCDC to step its output between e.g. 1V (LED off) and 3V(LED on).

--------
MPEW (Manage Power in an Efficient Way)
We are a team good at power and battery management IC/IP design.
Mail leehying@163.com for more.

 

[red_alert]

I = V/R, V constant, R(i.e. LEDs) constant, how can I vary?!

The internal resistance of a LED is not constant! It is just not too obvious (being nonlinear) but its characteristic could be analyzed using the classical U/I formula.

Look at a tipical diagram of LED forward voltage vs forward current and you could calculate its internal resistance for a specific forward current.