Originally Posted by kchriste

Ok, let me explain it to you this way. Let's say you have 12V, a LED string with a Vf of 10V, and a 100Ω resistor in series. This means that there will be 2V across the resistor and 20ma will flow. That will mean there is 40mW wasted in the resistor.
Now, let's say you have 14V, a LED string with a Vf of 10V, and a 100Ω resistor in series. This means that there will be 4V across the resistor and 40ma will flow. That will mean there is 160mW wasted in the resistor with no PWM. Now if we PWM it at 50% that means that an average of 80mW would be wasted in the resistor and an average of 20ma will flow.
If you have 14V, a LED string with a Vf of 10V, and a 200Ω resistor in series then there will be 4V across the resistor and 20ma will flow. That will mean there is 80mW wasted in the resistor with no PWM.
As you can see, there is no power saving using PWM without an inductor.

I agree with all your explanations. Pls see the following explanation and correct it if you find necessary.
Take the same example.
At 12V input with out PWM current taken is 20mA. So power from 12V is 12x0.02 = 240mW.
At 14V with out PWM power = 14x0.04= 560mW
With 50% PWM power from 14V = 14x0.04/2 = 280mW only.(extra 40mW in the resistor. But total power has come down drastically)
Am I Right?

Originally Posted by audioguru

Leds are less efficient at high peak currents. They waste the extra power as heat.
Your current-limiting resistors also waste the extra power as heat.

A buck regulator stores power, it doesn't waste power.

Philips Lumileds make buck and boost converters that have the inductor built-in and they are very efficient. They sense and regulate the average current.

Yes. Here there is an important point. Whether these high brightness LEDs use for traffic lights are suitable for dynamic display. Their light O/P unlike some other type of LEDs may not increase much with increase in current from 20 to 40 mA. Then with this type of regulation, the lights will become dim when input voltage increases since duty cycle is is reduced.
In that case buck converter alone will be solution.

Originally Posted by kchriste

Yes, but your LEDs will be 1/2 as bright because the average current is now 20ma with 50% PWM vs 40ma without it.

With 12V also the average current was 20mA only. But it was continuous. Here it is 40mA with 50% duty cycle keeping the average as 20mA. So brightness should be same if the light O/P of LED varies proportionately with current (which may not be true in this case as audioguru pointed out).
Right?

Originally Posted by blueroomelectronics

Why not build a test circuit? A simple 555 timer and a variable resistor to limit current plus an LED and see how duty affects brightness.

Anyway I will be doing that with the MC and driver. But I put these doubts in the forum to get the opinion of people with practical experience so that I can save time and learn from their experience.
I had done dynamic display earlier and observed the brightness variation.

Originally Posted by audioguru

I use MV8191 red LEDs. The luminous intensity at 40mA is only 1.7 times the amount at 20mA so the brightness is not doubled. Your LEDs might be different.

From the data sheet of MV8191, it seems to me that intensity Vs If characteristics is suitable for dynamic display, but light O/p is less for traffic light application.

Originally Posted by Roff

If you still want to use a voltage regulator, why not use a buck-boost switching regulator?

Can you pls suggest any link to learn about the buck-boost regulator you suggested?

Originally Posted by ericgibbs

hi,
A very useful reference manual for SMPS.

Thank you very much. I just had a glimpse. Seems to be very useful. I will definitely use this as a reference.
What is this 'buck boost' actually? Is it same as fly back converter where we can have the o/p voltage higher than input voltage?

Originally Posted by ericgibbs

This and other Google links for buck boost converter describe the operation.

Buck-boost converter - Wikipedia, the free encyclopedia

Thank you very much