Switch to LED Lights in my home

[cheerio]

A few month ago i started to look into using LED to minimize the power consumption of lights. I currently use Halogen lights which have this very nice natural light. We all know that White LED lights cannot produce the same light so i started looking for alternatives. I found them in RGB LED.

Plan:
Replace all Halogen lights with RGB LED lights without having any disadvanteges. I can save ~1/3 of the power.
To produce the same light i can switch the colour maaaany times a second so the human eye sees a nice spectral band of colours.
In addition i can mix the colours for any mood i am in.

Status: prototyping

Problems:
I use NPN Bipolar transistors that are Rated with 1A. The current drain on each of the 3 the LEDs is ~350mA. I calculated the Base resitor just fine but the Transistor gets too hot. My workaround is using MOSFETs. But they are more expensive and i want to use the Bipolar NPN that i own a few hundred of

Timeplan:
No real timeplan, just finish it in a few month


As always i like to read suggestions and posts of people that already did this.

[kam]

I've seen you quite often participating in pcbheaven, solving riddles and posting in the forum. Thank you for your participation! 

There are power transistors to solve your problem. A TIP142 for example can handle some 10 amperes and is relatively cheap.

Regarding the LED ilght... Replacing the lamps with LEDs, is not an easy job. You can use easily LEDs to make mood light, but you cannot work with LEDs, unless you pay a lot of money for expensive LED lamps.

[cheerio]

I did build prototype RGB LED lamps with current limiting. With this the Power saving factor was 30% compared to the Halogen lights. My Halogen lights are powered by 12V AC.
I use the BD139-10 Transistor for regulating the LED. It should do the job just fine but it gets too hot. I tried to lower the Base resistor down to 10Ohm but it still get's to hot. Maybe it switches too fast while regulating the current.

If i continue on this project i will do some schematics and stuff

[_pike]

If i continue on this project i will do some schematics and stuff

ok.. also datasheet of the leds that you will use would be helpful also...

[kam]

I use the BD139-10 Transistor for regulating the LED. It should do the job just fine but it gets too hot.

you regulate with linear current drop or with PWM?

[cheerio]

on the left VCC i apply the pwm. The circuit allows a fixed max. current that is controlled by R1. With the PWM i can pull the circuit below the max value.

If you do not know this circuit i can tell you how it works.

[kam]

it is the first time that i see this circuit, and it looks good. i think that i will run some experiments with this. Please explain.

[cheerio]

on the left VCC i apply the pwm. The circuit allows a fixed max. current that is controlled by R1. With the PWM i can pull the circuit below the max value.

If you do not know this circuit i can tell you how it works.

If you remove T2 and R1 you got a simple circuit to control eg. a LED with PWM. The PWM signal is applied on the base of T1 and R2 is calculated as a normal Base resistor. The LED would be connected to OUT. This far everyone knows how it works i guess. Now the Current regulating part.

We use R1 to produce a Voltage difference of ~0.7V (UBE value of T2)  between Base and Emitter of T2. And we select the R1 value by -> R1 = (UBE of T2) / Imax

I think an example helps:
We want to limit the current to 350mA. We expect UBE of T2 to be 0.7V .
R1 = 0.7/0.35
R1 = 2Ohm

What happens?

U =R*I
so the Transistor T2 will switch exactly @ 2Ohm * 0.35A
If this happens the Voltage Difference between Base and Emitter of T1 will be 0V because its base is shorted to Ground by T2.
When the current drops below 0.35A T1 starts to conduct again and so on. It is stabilizing itself.
I do not know if my explanation is good enough for you to understand this. Just ask if you don't get it right.

The advantage is that you save energy compared to a LED current limiting resistor that doubles the power consumption. The other advantage i see is that you can apply any Voltage above the minimum Voltage the LED needs. My Power supply is limited @ 24V so i could not test if the LED will still light @100V and above. But as the Current is limited by the circuit it should not be a problem.
For T2 you can use a small signal transistor.
My problem is that T1 get's too hot. Only reason could be that it cannot switch fast enough and is most of the time "half open". Well see


This image my help too

Left y-axis is Voltage in mV. Right y-axis is current in mA. X-axis is time.

[kam]

so i see! I think i understand how it works. What i'm worried about is the switching speed of the transistors. When the current peaks, there is a delay until voltage drop across R1 becomes higher than 0.7, and another delay until T2 saturates and another delay until T1 goes to cut-off.  The two transistors myst be fast switching transistors. I will test it for sure.

Now, the reason why T1 gets hot is that it is forced to dissipate power, something that happens when the transistor opertes between cut-off and saturation. I would start by reducing R2 significantly.

Thanks for the explanation. Now that i do not have a computer to run MPLAB, i will work on this. Maybe this will be my next theory