EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: Corporate666 on April 04, 2013, 04:11:27 am
-
I am trying to drive 12 high-power LED's from a 12V source (each LED has a Vf of about 2.95) and be able to individually control the brightness of each LED. I want to control the LED's with a constant current circuit, but 12 CC circuits would be complicated and expensive at 1A drive currents these LED's use.
I was thinking of splitting the LED's into strings of three, and having 4 constant current drivers. Then switch each LED in and out of it's string and let the CC driver compensate for whatever the forward voltage is at the moment with whatever LED's are in the string.
The LED driver appears to be able to respond in some microseconds to changes in voltage/current demanded of it. I would be dimming the LED's with a PWM frequency of about 200Hz, so I would think this could work OK.
But I've never tried it. Has anyone done anything like this before? Any pitfalls?
If the idea is sound, has anyone got some recommendations on the best approach to switching the LED's in and out of the string? I can think of a few ways to do it, but not sure what best practice would be.
-
Seems like a feasible idea, but you will need to make a voltage shifter of a floating gate drive for mosfet, if you want to dim for example middle led in a string. In some LED dc-dc converters - while the current is of course constant - the absolute voltage in respect to ground can vary quite a lot (especially in low side switched topologies)
-
Seems like a feasible idea, but you will need to make a voltage shifter of a floating gate drive for mosfet,
Not necessarily - you could just make the gate drive voltage swing high enough to cover all permutations of LED Vf
I did an RGB dimmer like this a few years ago - I think I've subsequently seen a Nat.Semi appnote with a similar circuit.
An issue is if you want to do PWM control like this, the dynamic response time of the CC circuit can become significant.
-
Seems like a feasible idea, but you will need to make a voltage shifter of a floating gate drive for mosfet,
Not necessarily - you could just make the gate drive voltage swing high enough to cover all permutations of LED Vf
That's right, but only while any possible combination of source potential + Mosfet Vgs won't exceed +/- 20V which is typically what channel-gate oxide layer can withstand. Obviously this can be prevented with a ~15V zener between source and gate and some series gate resistor, but this adds even more parts to whole circuit and when mosfet driver is in off state and mosfet source is positive with respect to GND, the zener can get forward-biased and fry your mosfet driver.
As I said: definitely feasible, but would need some serious thought.
-
I guess the main problem is the voltage spikes that will arise across the LEDs, how large they are and how long the LEDs have to withstand them. It might be hard to calculate the worst case solution so you will have to put in a large design safety factor.
I don't think it's a good solution.
Just do 12 channels if you need 12 channels. To do it in micros you will need 12 pwm channels and optionally 12 A2D channels. ie probably More than one micro.
Or do it in analogue.
For your solution you will need 12 pwm outputs anyway plus the 4 led drivers.
-
But I've never tried it. Has anyone done anything like this before? Any pitfalls?
It looks feasible providing the switcher isn't upset by rapidly changing load voltage and the LEDs are not going to like any smoothing capacitance on the switcher output. Logic level MOSFETs with 12v gate drive look like the simplest solution to short out LEDs.
On the other hand a single switcher producing around 4v with some resistance in series with each LED and a simple MOSFET switching the low side would be less components at the cost of some efficiency and heat.