Electronic Circuits and Projects Forum

Corey,

Here is a hack at making a linear constant-current regulator for a string of three of the W724C0 LEDs using my favorite LM431. I configured it to deliver ~1.6A. Note that it keeps the LED current constant while the battery voltage sweeps from 12.2 to 15V. Below 12.2V, the NFET is saturated, and the current decreases. R1 sets the current.

In the third plot pane, see the power dissipation in one of the LEDs, the FET, and the resistor R1. At 14.5V (motor running alternator voltage), the dissipation in the FET and R1 are about 4W each; quite manageable.

Let me know if this seems to be heading in the right direction. If it is, when I get some more time, I will simulate the effects of transients.

I made it adjustable by adding a pot.

My 0.02: I used "10 W" LEDs very similar to this on my car's reversing lights.

I strung 3 in series, and utilised the car's exisiting wiring as a 'current regulator'. They lasted a suprisingly long time, some weeks, and well long enough to evaluate them.

My conclusion was the coloUr rendering was a bit poor, so I didn't persue it any further.

A plus with these LEDs is the efficiacy, 70 lumens / Watt, a lot better than incandesant at 5 to 10 lm/W

Originally Posted by Hero999

See page 34.
http://www.electro-tech-online.com/c...7/41215a-1.pdf

Obviously you can use a discrete comparator rather than a MCO and add a pull-up resistor if it has an open collector output.

This circuit is wrong. It won't oscillate as drawn. You would have to reverse the inverting and non-inverting inputs to the comparator to make it oscillate. It also has an awful start-up transient until C1 charges which causes the current through the load to be very high initially.

I think SoftStart was ommitted for clarity of illustration. I'd say the LED would tolerate the high starting current anyway, going on the longievity of my "10 W LEDs setup" in my earlier posting.

The 'Comparator' doesn't have to be a comparator, it can be a transistor. It doesn't need to compare accurately, just fast.

Could the posted circuit have its flaws fixed enough to become a practical circuit?