PWM the output of a LED driver for dimming

[aiq25]

I'm planning on purchasing few Mean Well LPC-35-700 Constant Current LED Power Supply because I found at a great price. However these are not dimmable. So I'm thinking of using an N-Channel MOSFET with low on resistance to control the output of these drivers. I'm new to LED drivers and switching power supplies, so I don't really know if this will work. To me it seems like it should work.

I think I should be able to connect the output of the driver to an N-channel MOSFET and just drive a PWM into the gate to control the brightness of the LED's. I think one issue will be noise and voltage spikes. I'm thinking of using capacitors to help mediate that. Would this work or am I just asking for trouble? Feedback would be appreciated. Thank You.

[T3sl4co1l]

As in, drain to +V, source to -V, gate and source to PWM...?

If it's truly a "constant current source", this is the proper way to do it.  However, I'd be surprised if it actually is, because CV supplies are so much easier to build -- usually there's feedback (with a low bandwidth) controlling the average current level, but the actual output is bypassed with capacitors and stuff so that it's not also an AC current source.  Which means the frequencies in the PWM signal won't be treated too kindly, and you'll have problems.

It also means using a series switch is problematic.  Yes, you can turn off the LEDs and be absolutely sure that they'll be off... but for how long?  Ideally, a current source climbs to infinite voltage when open circuit; the power supply likely has an overvoltage protection in it to limit this, but the flipside is, when the PWM turns on again, how long does it take before LED current is normal?  You'll get a huge inrush as the voltage comes down from limit to nominal.  Or the supply shuts off and goes into an intermittent state, and takes time to recover.

A better option is to simply get a supply that has dimming control available.  If this is more expensive, so be it.  You could also try hacking one to make it dimmable, but without the circuit, obviously we can't give direct advice.

Usually there's a summing node, where the current feedback signal goes into a controller.  You can add current from a resistor divider -- coming from direct or filtered PWM -- to fake it into thinking there's more LED current than there actually is, and therefore throttling down.  Beware, you may have to do this from the AC line side -- and the controller being self-powered, means if you throttle down too low, it loses its own power and it starts blinking.  That's one [accidental] operating mode of an LED light I made for myself, when turned down too far.

Tim

[aiq25]

Thanks T3sl4co1l. I think I will take one apart and see what's inside. I don't really need to dim these modules, they are mostly for testing anyways. I'm planning making my own dimmable drivers using the LM3414.

[dannyf]

I don't really know if this will work.

Without knowing what it is, it is tough to say if it will work - generally, smps doesn't play well with dynamic loads.

[David_AVD]

Without knowing what it is, it is tough to say if it will work - generally, smps doesn't play well with dynamic loads.

I'm assuming you're talking about CC power supplies?  CV SMPS power supplies are used all the time with hugely dynamic PWM'd loads.

[Kjelt]

Thanks T3sl4co1l. I think I will take one apart and see what's inside. I don't really need to dim these modules, they are mostly for testing anyways. I'm planning making my own dimmable drivers using the LM3414.

Is it for livingroom?
There are some things to consider:

• Constant Switching Frequency Adjustable 256-step dimming resolution requirement for general
from 250 kHz to 1000 kHz

I designed a circuit few years back that had same sort of device (max 16832) and was very happy with it, good ground plane and all components esp coil close together. Then for the fun I had a friend that has much more power electronics knowledge than I have check the final PCB and he found with his spectrum analyzer out that I was jamming the entire FM radio band 
He fixed it for me with a couple of beads in series somewhere in a critical path. 
Just to warn you upfront it looks like childs play but there can be some ugly catches.

[aiq25]

Is it for livingroom?
There are some things to consider:

• Constant Switching Frequency Adjustable 256-step dimming resolution requirement for general
from 250 kHz to 1000 kHz

I designed a circuit few years back that had same sort of device (max 16832) and was very happy with it, good ground plane and all components esp coil close together. Then for the fun I had a friend that has much more power electronics knowledge than I have check the final PCB and he found with his spectrum analyzer out that I was jamming the entire FM radio band 
He fixed it for me with a couple of beads in series somewhere in a critical path. 
Just to warn you upfront it looks like childs play but there can be some ugly catches.

No it's for a reef aquarium LED lighting actually but I might use it for other things. I'm mostly in the electronics learning phase right now and I'm really interested in power electronics, so I thought this might be an interesting project. And the Meanwell drivers I bought was to test some LED's I have (Chinese ones) and test them.

Oh wow jamming entire FM band? That's sounds fun. lol

[Kjelt]

Ok then another warning, keep the wires between the leds and driver as short as possible due to the emission. Normally not a problem but in an aquarium you don,t want the drivers in the hot humid zone. For an aquarium you want IP66 (outdoor) stuff, potted drivers. Personally I would buy them for this application.

[bookaboo]

You may have issues doing PWM on the output of a constant current unit. The p/s will attempt to regulate the current and on your "off" cycle it may do strange things, on one we tried the p/s got very warm when we tried this. Luckily it had adjustable settings so we just set it to CV mode and all was well.