How do I dim a LED using AC power?

[krazyplace]

I just bought an AquaController and now I want to figure out how to get it to dim a LED moonlight

Since a LED runs off of DC power (0-10v) converted from a 120v AC supply, there must be a modification to allow a variable AC to vary the DC power to an LED lamp.

Then I can control an LED moonlight via a standard X-10 lamp module.

Does anyone have any ideas on how to dim an LED by adjusting the AC power source? Any EEs out there that have an idea how to build this circuit?

[bbragg]

LEDs are usually dimmed via Pulse Width Modulation. At 100%, the LED is fed DC at a voltage somewhat greater than it's forward drop to produce light. I'm thinking the curve of current to light output is abrupt and doesn't lend itself to dimming via voltage variation as is common with an incandescant lamp.

You would need a circuit that could sense the dropping DC from the power supply, regulate to a minimum fixed DC value, and then chop the DC into a PWM square wave based on the drop from the basic supply.

For example, if the supply outputs an unregulated 10vdc from a nominal 120vac line, and the LED needed to see 4.5v @ 30ma, you would need a circuit that would regulate the 10vdc to 4.5vdc AND would do PWM based on the unregulated voltage of say 0% at unregulated 5vdc and 100% at unregulated 10vdc.

In that case, the LED would see its required Vf and the varrying duty cycle would take it's queues from the lowering of the AC line voltage as sensed on the unregulated output of the 10v supply.

Good luck with that. There are a myriad of cheap semi's that do PWM.

By the way, I don't think the AquaController can send any of the X-10 dimming commands. It's pretty much on/off stuff. I have an AC2 and I'm in the middle of moving to the AC3 myself. Perhaps this is something in the new AC3?

[krazyplace]

Next time I'll be more careful what I ask for!

I built a LED moonlight using a AC/DC transformer and some resisters, so I'm not too ignorant on this stuff. In my limited exposure to circuits in school, I learned that anything beyond resistors and capacitors should be left up to EEs. I think this might be beyond my knowlledge base. To keep this thread from getting out of control, I need to put some boundaries on my abilities. I can build minor mods, but I can't design/build any complex circuits. I was hoping for something that is more of a retrofit.

I bought an AquaController Jr which has a moonlight function that can turn on a lamp module's on/off signal as well as send the dimmer values to it. I think the AquaController II does the same thing. You might just need a firmware update if you can't do it now.

[old salty]

Though it's possible to do what you propose, it's much much easier to dim LEDs with a rheostat or a potentiometer. When an LED lights, it has X amount of a voltage drop across it (it is a diode by nature.) It requires this voltage or it won't light at all. A current limiting resistor allows you to maintain this voltage drop across the LED while lowering the current (dimming the LED). Hope that kind of sums it up in laymans terms.

[krazyplace]

Ya, I'm with you. I still need to figure out how to "automatically" adjust the resistance when the AC supply is adjusted.

Basically I need either a PWM that will automatically adjust based on the adjusting AC input and the affect DC output from a AC/DC transformer.

As you just said, I could also use a adjustable resistor too.

Doesn't something do this already? You know, like a dimmable LED light bulb?

[bbragg]

I've looked up the mcd / I curves for a few LEDs (white) and they appear to be fairly linear until you get down to 20% of recommended operating current. The "mcd" is milli-candellas which is the typical measure of light delivered in the LED world.

This linearity means that what "old salty" has proposed should give you enough variability in brightness for your purposes. By varrying the current, you can go from 20% to 100% brightness. Of course you can probably exceed the recommendations to get past 100% but I'm assuming you're not after that.

A PWM circuit would let you go from 0% to 100% percieved brightness ... but I think 20% to 100% would be fine for your application. Of course, you still get 0% as well by turning the device completely off.

Your power supply to the LED must be unregulated though. If there's a regulator in the path, it will severly limit your tunability because it seeks a constant voltage ... where, in your case, you would like to vary the voltage.

An unregulated supply would look like a small transformer, a rectifier, and a small smoothing capacitor to suppress the 60Hz ripple. If that's what you're driving your LED with and the unit puts out 10volts, then you need the Voltage Drop for your LED and it's optimal operating current.

If the Vf of the LED is 4v and it's optimal current is 20ma, then the math would be ...

(10 - 4) / 0.020 = 300ohms.

You'd need 300ohms in series with the LED. And it will dissipate 0.12 watts so you need to probably use a 1/4 watt resistor and not a smaller 1/8th watt version. Then the TRIAC in the X-10 controller can chop down the AC, which will result in lower voltage on the output of the transformer, which then get's rectified to DC and smoothed a bit by the filter cap. TRIACs often put out some level of DC offset in getting their job done so there will be some strain on the transformer (heat) but as long as it's well within it's capacity ... it shouldn't be an issue.

At about 1/2 level on the AC line (x-10 50% dim), the unregulated power supply should be putting out about 5v. Given the same LED we have ...

(5 - 4) / 300 = 0.003ma

So that's about the minimum you can achieve ... around 15% brightness. In fact, at that current level the LED may not produce any light. Minimum MAY be at 5.3v or somewhere thereabouts.

So what that means is that varying the dimmer via X-10 from 50% to 100% will result in varying the LED output from 15% to 100% of it's mcd. Below 50% on the dimmer TRIAC, the LED will likely just go out.

Seems like a workable solution to me. Until you start to put your LEDs in series or parallel. Then you need to recalc for the resistor and will have a new dimming range.

[krazyplace]

I think I really need to go from 0% to 100% so that I don't have a "no moon" situation 20% of the time. Do you think I can put a PWM circuit in and allow the "unregulated transformer" to adjust the frequency?

[bbragg]

I see what the issue is. I've read the AquaController manual and I see that it varies the dimmer from 0% to 100% for it's simulated moon cycle. But below about 50%, using the method I previously described, your LED will simply go out. All of it's range will be packed into the 50% to 100% AC line dimming area.

You want the full range of dimming on the LED to match the moon cycle that the AquaController is sending to the X-10 dimmer to handle what would normally be an incandescent bulb.

I will look into it further and let you know what I come up with. You can certainly form a circuit that matches the dynamic range of the input (AC line voltage) with the output (LED operating current) without going PWM. I will look into it further.

What are you using for your LED moonlight? Is it something you put together or something you bought?

[krazyplace]

I have a DIY that I put together using Radio Shack parts (12v transformer & resistors) and some LEDs that I bought off the Internet. The look is great! Right now my tank is torn down for a move. I was think about buying a "real" one if I could find one that will work with the AquaController.

[szwab]

there is a diyt plan out there for dimmable leds that do not use a transformer.
that would work with your AC