HELP : Single Transistor Constant Current Driver

UK Owl,

I presume that you are looking for a simple, cheap linear design. The Buck circuits that Streak mentions are the way to go if you want an _efficient_ circuit, but if you want simple, then linear is better.

There are some very simple _two_ transistor circuits that you can use. One transistor is in series with your LED to regulate the current. Base current is supplied by a simple resistor to across your supply. The second transistor is part of your feedback circuit, and acts to 'steal' some of the base current if the load current gets too high. See, for example, something I put up at http://borealis.com/~winnie/LED_linear/index.htm

-Jon

Hi guys,

I am a long term lurker and first time poster who would like a little help.

What I would like to do is create a Constant Current driver circuit to drive two series white (or blue) LED's from a single PP3 battery. I would like to be able to alternate between NiCd and Alkaline cells and have a constant level of light output throughout the battery life, so I will need something more sophisticated that a simple series resistor.

I was thinking about connecting two LED's in series with a Transistor (I have some BC548c Transistors left over from when I previously had to make some multivibrator circuits) to control current.

The problem is how can I provide the Base current needed to produce a stable collector-emitter current of 28 milliamps through the LED's with a supply voltage changing from 9.6 volts down to 8.2 volts or lower.

Anyone tried this type of thing or have any ideas that could help me ? any circuits would be most appreciated.

I am no expert but there are a lot of Buck circuits that should do the job. Take a look at:-

http://www.maxim-ic.com/appnotes.cfm/appnote_number/358

http://www.intersil.com/engineeringtools/tools/buckandboostcircuit.pdf

Use the LM317 in constant current mode
R=1.25/Current

Jonathan's circuit is pretty nifty, but I have an even simpler one...



I talk about it more here:
http://www.candlepowerforums.com/cgi-bin/ultimatebb.cgi?ubb=get_topic;f=15;t=000090

My final implementation used a 1.2V voltage reference diode instead of the 1.7V LED to set the base current. You just have to choose the right resistor then to get the current you want.

It should work fine for your 8.2-9.6V input range, driving two LEDs at around 7.2V. For 28mA you'll need about a 17.9ohm resistor instead of the 10ohm in the diagram.

Hi guys,

Thanks for your input, what i was looking for was more like the one from 'Highlandsum', yet with this i think you would have a leakage of about 10mA through your voltage reference LED.

I have modified the circuit a little using the idea of a second led as a voltage reference :-

A 22k resistor connected to +vcc then through an LED to oV.

The junction of the LED and resistor is connected by another 22k resistor to the base of a bc548c NPN BiPolar transistor. The Emitter of the transistor is connected to oV. The a white LED is connected between Vcc and the Collector of the transistor.

result :

about 0.3 mA leakage (waste) through the Voltage reference LED.

for a supply of 8.5 V (half dead alkaline PP3) 25mA taked.

for a supply of 5.5 V (stack of 4 half dead AA alkaline cells) 23mA taken.

note : current readings taken with a Digital MultiMeter.

I think the circuit is far from perfect as it is not completely linear as an LM317 one would be, but it is by far better than a simple series resistor. Also for an LM317 circuit the supply voltage must be at least :

LED_Voltage + 2.25_Volts + Minimum_Voltage _Across_LM317

yet for this the minimum voltage needed should be LED_Voltage + Minimum_Collector_Emitter_Voltage.

Thanks again guys and to all posters for your inspiration.

The LM1117 has a much lower dropout voltage of 1.5 V and should work fine in this situation (as long as your current needs are <= 800mA).

The circuit isn't completely linear, but with minimum junction voltages around .3-.6V it's hard to do better. I've seen temperature-compensated components but frankly I don't see the need to fret over it. Granted, my application is on a car, where the electrical supply is nearly infinite compared to the LED needs.

The NPN layout you describe is pretty much where I started from, although I didn't have the 2nd resistor between the voltage reference and the base. I would think that will further throw off your linearity since the resistor adds more variation to the circuit. 22Kohm sounds good, I don't think the choice of resistor on my part was very critical.

The theory I took was to use the extra LED to provide a constant voltage source. With this constant voltage I could provied a constant current to the Base of the transistor.

such that :

LED/Resistor junction voltage = 1.59 Volts (measured with DMM)

Transistor Base threshold voltage = 0.6 volts (typical for Silicon transistor)

so constant voltage considered is 1.59 - 0.6 = 0.99 volts

Given the hfe of a bc547c is typically 600 (from data sheet.

To get 28 mA with a hfe of 600 you need a base current of 28 / 600 = 0.04666 mA

Given Ohms Law :

i.e. Current is directly proportional to the voltage and indirectly proportional to resistance

i.e. I = V / R

so to calculate value of resistance we use

R = V / I

given :

V = 0.99 Volts
I = 0.00004666 Amps

then :

R = 21217 Ohms

i.e. nearest prefered value = 22k

I think the first 22k resistor used to provide the voltage reference might be a little high so for a reduction in supply to 5.5 volts this can not supply enough current through the second resistor that is connected to the base. Either that or the difference in supply voltage is changing the Voltage Reference LED Current, and thus the voltage across it is changing.

When I make a second one I think I shall try reducing the 22K resistor connected to Vcc to on a size or two lower.

Given this, on criticism of my circuit is there is no form of feedback to compensate for thermal drift, whereas the one by highlandsun does, as the current flowing through the 10 ohm resistor will act as a form of feedback control.

You could add a small "ballast" resistor on your Emitter lead. Common amplifier technique to reduce dependency on transistor gain.

If it isn't too late for suggestions, I modified a two-transistor current mirror circuit to work as a constant current source. It is currently designed to drive three 3.7V LEDs from a 5.25 - 4.5V source at 20mA each, but you can probably modify it to handle your 9V source.

You can find it on my home page.

HTH

Tennessee