I have this idea to dim LEDs and I'd like to get some opinions if it's feasible.

enter image description here

Q1 and M1 form a constant current source, Q1 ensures a 0.7V drop across R2 by pulling M1 gate to ground. Q2 gets a PWM signal on it's base turning M1 off completely during positive pulse.

I'm a bit worried about M1 slamming LED hard until Q1 turns on, my guess is that a capacitor from gate to ground would slow it down, am I right?

Does anyone see any other problems in this design?


Looks workable.

I doubt that you'd have major issues with the FET turning on substantially faster thahn the bipolar unless Q1 was a seriously low Ft part. If the LED current gets even say 20% above nominal then Vbe will be seriously high compared to normal (if 0.7V usually then = 0.84 V at 120%) and at +50% current Vbe = 1V+ and the transistor is trying very very hard indeed.

1 kHz is OK for many things but depending on duty cycle you MAY get effects that some people can see and if you are moving the LED, and depending on what it is illuminating, you may get motion artefacts. eg if you move the LED at 1m/s then in 1 mS it move 1mm so you have the lighting effects of 1 PWM cycle spread over 1mm. Fairly fine - you may see surface patternin. At 5m/s and 5mm/frame you can probably see the PWM patterns as dark / light sequences on an illuminated surface.

A base capacitor is probably a bad idea. If time constant is of the order of a PWM frame period or longer you start to get mean DC at the gate and the FET may run in partially on semi linear mode. A bit smaller Tc you round the pWM corners and slow the transitions and add heat to the FET. Exact effects depend on how hard Q2 is driven. Turning a transistor on across a charged capacitor tends to create undefined results and high current peaks unless specifically designed.


The circuit shown is not PWM nor does the square wave input do anything but cut off the current 50% of the time. Q1 does control current but in a linear fashion resulting massive heating of the FET at its operating point. IF you remove Q1, the circuit will control current and provided the current controlled is small, the FET probably won't over heat. To actual have a switched current control, one that can handle large currents and stay cool, you need a triangle wave generator, a comparator, and an measurement of the current using op-amps.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.