PWM Dimming Constant Current LED Drivers

[Chet T16]

Hey all,

I'm attempting to dim a standard led light with switch mode constant current driver using a mosfet to dim the output. The mosfet dimming circuit is powered from the DC output of the driver.

The problem i'm seeing on various makes of drivers is that there is a large amount of strobing when the duty cycle gets below 50-80% (depends on the driver). I don't mean mains flicker but more of a low frequency of < 10 Hz. PWM frequency is ~2kHz.

My initial thoughts here is that the CC supply doesn't like being turned on and off and is seeing the non conducting mosfet as an large resistance and when trying to maintain the current the OVP is kicking in.

Any ideas if this is the case or anything I can do to make this work?

[Neilm]

Could you post a circuit - it would be easier to see what is happening. If you are running a constant current supply and are disconnecting the LED with the mosfet you are probably sending it unstable - as you have assumed the driver will just attempt to maintain the current flow. Why could you not use a supply that had a dimming function in? There are many that do, usually by PWMing the enable line.

[wraper]

PWMing output of constant current driver won't work well. It will shoot up voltage every time when circuit becomes open. And when circuit becomes closed again, you'll hit LEDs with high current. Not to say CC driver won't be happy about this as well and probably will fail soon.

[Gibson486]

Usually you want one or the other (either PWM or CC control)....but if you must do it, you better have an inductor in the design.

[Siwastaja]

You can PWM an actual proper (edit: clarification: an ideal) CC supply, but you do it by shorting out the load! With Rds(on)=0, current and power is 0 as well when the FET is on.  With series switch, infinite voltage happens...

CC led drivers, however, might not be designed to be handled that way either (at least they might have extra capacitance outside the CC loop for EMI, which creates a problem when shorted). That's why they almost always have a PWM input.

[wraper]

You can PWM an actual proper CC supply, but you do it by shorting out the load! With Rds(on)=0, current and power is 0 as well when the FET is on.  With series switch, infinite voltage happens...

CC led drivers, however, might not be designed to be handled that way either (at least they might have extra capacitance outside the CC loop for EMI, which creates a problem when shorted). That's why they almost always have a PWM input.

If you PWM short laboratory PSU, IMO many of them won't survive such abuse for prolonged time either.

[Siwastaja]

Indeed, ideal CC supplies in this regard are rare; often custom designed in where pwm-by-shorting is needed.

Simply because most CC supplies have an output capacitor.

My point was, doing the fundamentally wrong and insane thing - equivalent to dead-shorting a CV supply such as a battery! - i.e., opening the LED chain, has no chances of working, not in theory nor in practice. It will just trig the protection, either latching or CV mode (or blow up a really badly designed CC supply if such exists).

Of course, doing the "ideally correct" thing won't work either, but wanted to mention it for reference.

So you are stuck with either using a supply with a PWM control input, or doing some unoptimal kludge (such as partial shunting with a power resistor).