LED Lighting. I need some design and theory help.

Thread Starter

Tom Kay

Joined Feb 10, 2009
35
Hello All;

This is my first post, and I want to jump right in with a problem regarding LED lighting to replace all the 110VAC lights in my house. I'm starting with the home theater, and I need the help because in my theater, I want all the lights to be dimmable so I can see the movie in partial darkness.

Background: I have 5 separate circuits in my theater now. On each circuit there are 6 pot lights in the ceiling, currently running 50 watt halogen GU-10 type bulbs. So each circuit is consuming around 300 watts.

Halogens dim nicely, but they give off a lot of waste heat, and they tend to blow more often than I would have guessed. I want to replace all the lights in my house with dimmable LEDs, starting with the theater. The goal is to save money, since my little monsters leave lights on all over the place, and also to never have to replace another light bulb again.

I bought 2 LED bulbs from Ebay, made in China. They are the GU-10 type, rated at 3 watts of consumption, and each have 72 LEDs soldered in place. They were advertised as dimmable with a "normal" dimmer, which nowadays means a Triac-based PWM dimmer from Home Depot.

The problem is that they don't dim as expected. I put one LED bulb in place of a halogen, left 3 other halogens in the circuit and switched it on. The LED did nothing until I slowly turned up the dim setting to about half way, then POP, on it came, almost full intensity. I was able to back off a bit and dim it more evenly, but the dimming knob was very sensitive. Almost no movement of the knob, and a big change in light output from the LED bulb.

I just popped one bulb open today to have a look at the internal circuit. I discovered 4 diodes in a bridge (no surprise) 2 capacitors, 3 resistors, and the fact that the 72 LEDs are arranged in two parallel banks of 36. Each bank of LEDs is made up of all 36 LEDS connected in series (one long chain). This surprised me a bit, although I wasn't sure what I'd find.

I guess the LEDs are all sharing 110 volts divided by 36, or about 3 VAC each. I could easily be wrong about this, but it looks that way.

So, here it is: Why isn't it dimming as advertised? How can I fix this, assuming that I want nice even dimming of the LED bulbs?

I also have read about LED drivers. These apparently, are the items to control LED dimming, but they are also PWM circuits. I know all PWM circuits are not created equal, but why wouldn't this work with a cheapie Home Depot PWM dimmer?

Thanks, and any help from this forum would be greatly appreciated.
Cheers, Tom Kay, Ottawa Canada.
 

Thread Starter

Tom Kay

Joined Feb 10, 2009
35
Hi Bill;

Thank you for the links and all the info. There's a lot to read there, so I'll try and sift through it asap.

It sounds like I've found the right forum, too, and that's always helpful.

My knowledge of electronics is limited, so I won't be reading at warp speed. However, the goal is still the same, and that is to replace all the lights in my home with LEDs, dimmable where desired.

Bill, do you see a major hurtle to that goal, given the modern higher output LEDS that you've mentioned in your posts?

Thanks, Tom.
 

italo

Joined Nov 20, 2005
205
As you find out LEDS are not lamps but rather diodes that follow an exponential curve. So what can you ? do instal red leds for your room they are great for see enough but not enough to be bright. Submarines use red lights for adjusting the eyes for the dark. I actualy installed them in my bedroom in some recesses to give off the glow. can i adjust them not realy but you may add a plastic cover to dimm them. If do install a dimmer you will never get a glow it will just shut off.
 

Wendy

Joined Mar 24, 2008
23,415
Actually it is very doable, but kinda expensive. Luxeon's Rebel series would do it nicely, but be careful, at $30 or so a pop you want to get it right the first time.

My personal opinion (and this has been specifically discussed at AAC before) is it is a little too soon. When it happens the costs will fall dramatically, that and the other tech/science sites have listed several technical breakthroughs that have to filter down.

What I would do is pick the most expensive light in the house, and make it the Guinea Pig. I don't think at the technical expertise you've claimed I would go for a dimmer type, though that is also doable.

The problem is dimmers are designed to slice off parts of the AC wave form with SCRs or Triacs (specific electronics devices used for AC), which works on RMS gadgets like light bulbs, but not so well on LED's. The methods used for LEDs tend to be DC circuits (PWM). The technologies are basically not very compatible, at least not yet.

One other thing, the color of white can be important. I learned this with fluorescent bulbs, and it applies with LEDs too.
 
Last edited:

Thread Starter

Tom Kay

Joined Feb 10, 2009
35
Hi Again Guys;

Italo, thanks for the thoughts. I should point out that any room I use LEDs in will have to be "family suitable" which usually means multi-purpose. While it might look cool to my 2 children to have a submarine ambience to each room, I suspect that my wife would get pretty tired of it in about 30 milliseconds.

And really, taking my home theater as an example, it will be a theater, a games room and a bar, so again, at times I'll need brighter lighting. But hey, I'm willing to listen to all ideas, so thanks.

Bertus, thanks for the link. Bill has just given me a ton of home work, so I'll read your link when I can. Cheers.

Bill, I have continued to read your info, right up to the point of my incompetence. It didn't take me long to get there, but one method I have used in the past, to understand things, is the repeated-reading method. The Chinese water torture of learning, so I'll shake my head and go back at it soon.

Question, these Luxeon Rebels, are they designed to control my existing GU-10 type, 110VAC LED bulbs do you think? Again, I'd be looking at 5 circuits, with 4 to 6 bulbs on each circuit. If I read you correctly, I'd need 5 x $30 for a total of $150. Not bad, unless you mean one Rebel per bulb. Then it gets ugly. I'll try to find this on the web, or even call Luxeon if I can.

Cheers again, Tom.
 

Thread Starter

Tom Kay

Joined Feb 10, 2009
35
So, let's step back a tad and ask the question;

Using the GU-10 bulbs as described (in my first post) can we buy an off-the-shelf driver/dimmer for circuits that have 4 to 6 of these bulbs right now?

Or, failing that, is the design and fabrication of this dimmer feasible using hobby-type parts and skills?

In other words, if you were any self-respecting 3 Volt LED, hooked in series with 35 of your closest friends, powered by 110VAC rectified to the right voltage and current, what would you need to dim smoothly?

I already know that a PWM dimmer acts by switching on the current at some delayed point past the "zero" as a sine wave starts to form, and switches off at the moment the wave decays back down to zero. Then it repeats the action on the opposite part of the sinewave form. (see the picture).

I tested a basic Home Depot rotary dimmer. This dimmer is cheap, but it's at least got electronics in it, presumably a triac that helps the whole thing act as a PWM circuit. Again, I'm no ace, so I could be wrong.

We measured the voltage out of it, and at ANY point on the dimming knob, the voltage was a constant 115VAC, which further leads me to believe that it is a PWM device, not just a variable resistor. (I knew that anyway, because it builds up little heat during its operation, and is labelled as an electronic dimmer).

Another possible concern that my local electronics guru came up with is this: This is such a low power (3 watts) that the dimmer may have trouble functioning when it's really designed for 600 watts. Perhaps adding more LED bulbs in parallel will give it a more appropriate load, thereby allowing it to do its job. Sounds a bit wishy-washy, but it makes some sense. The LED bulb did dim a little better when I put it in the circuit with 3 other halogens. It was still sensitive, but didn't dim down from full to zero in an instant. The knob was still too touchy, though, but it did dim far more smoothly than it does when it's the only bulb in the circuit.

So again, if YOU are the LED bulb, what voltage or current form would YOU like to be fed to successfully dim from 100 to 0 percent? As silly as it sounds, if you're willing, please describe it literally in laymen's words. I typically find that's the first part of design, whether I'm machining an aluminum part on the mill, or hoping to create a circuit. Then we can get more technical and look at what circuit devices might create pulses, or increase frequency beyond 60 Hz, etc.

Thanks, and I ain't giving up yet. Tom.
 

Attachments

Thread Starter

Tom Kay

Joined Feb 10, 2009
35
Hi Alberto;

Thank you for the reply. Any input is welcome on this topic, of course, but as always, my electronics knowledge is somewhat limited. I just can't think like an electron. Or a LED for that matter.

So, I'll go through your response. First, I wonder if the capacitor (the large one on the VAC input wire) has some other duty, such as smoothing the voltage coming into the light if you get momentary fluctuations elsewhere in the house (ie, a fridge starts up or some other start-up load). This cap (I am assuming that it's a capacitor) is much larger than the polarized electrolytic cap that's mounted with the 4 diodes on the mini circuit board. I'm sure the small electrolytic cap does its standard job of smoothing the humpy sinewave form AFTER full wave rectification. But the much physically bigger cap is still a mystery to me.

Second, I wonder if the correct voltage is supplied to each LED by virtue of the fact that they are in series. 110 Volts divided by 36 LEDs would be 3 Volts per LED, which seems conceivably perfect for each LED, assuming a 3 Volt drop on each LED. There is also a small resistor on the positive wire that supplies power to each of the left and right bank of 36 LEDs. Would those be needed to create a "load" to then supply the correct current to the LEDs? Again, typically, I'm a tad out of my league on this.

I'll draw out the circuit so I can share it with the masses, at some point soon. I have drawn the company logo that's on the mystery part. The part itself is 7/8" wide x 3/4" high x 3/8" thick. It's rectangular, redish brown coating, and has 2 wires. There's a small resistor attached across its wires. I'd bet it's a capacitor and the resistor drains the charge at required times, like when the light's are switched off. For the markings, see below.

I know what the 250V means, but I'm a bit stymied by the 205K. Is this a part number, and not a value???

Back to yoiur reply, I odn't know what you mean about non-linearity of the LEDs. I have read that thye dim in quite a linear fashion, meaning that 50% input power will give you roughly 50% of the max light from the LED. So I'm not sure what you mean.

However, staying with the PWM dimming method for a moment, despite any possible inefficiency, what might the right approach be? Staying with a simply verbal explanation for a moment, how many pulses per second would be good for a LED? I have read that LED response times are tremendously fast, and this is why these devices can be used for fiber-optic communication, because they modulate voice and other info well. Would you stay with 60 hertz or try to multiply this and feed the LEDs with more pulses per second, like 600 or 6000? Would some simple timing device like a 555 timer do this? Would triacs be involved? Keeping the pulse length the same, how would we shorten the pulse width that we feed to the LEDs? Hopefully that would be done with a trim pot, much like a dimmer control.

As you can tell, I'm approaching the point where I'd like to discuss circuit design. Not sure I'm quite ready before I understand some of the theory.

Thanks again, Tom.
 

Attachments

Last edited:

Thread Starter

Tom Kay

Joined Feb 10, 2009
35
Hi Again;

I want to hopefully get some feedback from somebody who can tell me what sort of wave form will properly dim a LED, or a bank of LEDs in series.

The first sketch is a squarewave version of basic 110VAC voltage. The second is a rectified squarewave version of AC, now turned into DC.

The third is showing the effects of Pulse Width Modulation, or at least I assume that's what it would look like. The "Time Delay" amount depends on how much dimming you need. Or in other words, the power pulses become narrower as you need less light from the LEDs.

The question is, will this dim a LED smoothly? All 110 Volts wouldn't be fed to each LED, but rather to 36 in series, so around 3 Volts each, and assume constant current for the moment.

Is this how LED dimmers or drivers work? I'd like to establish that first, then work on a way to create that wave form.

Thanks, Tom.
 

Attachments

bertus

Joined Apr 5, 2008
22,270
Hello,

When you are rectifying the 110 V AC you will get a higher voltage with a capacitor afther the rectifier.
See the attachment for more info.

You can use PWM to dim the leds.
This can be done with a much higher frequency than the 60 hz from the net.

Greetings,
Bertus
 

Attachments

Thread Starter

Tom Kay

Joined Feb 10, 2009
35
Hi AudioGuru;

I can't argue about your evaluation of cheap Chinese junk. I have felt the same for years, especially with regards to tools that I'd like to keep working.

I will say though, with regrets, that the person who informed me that the LED bulbs were dimmable with a simple Home Depot dimmer, wasn't Chinese. Nor was his address. I'll give you 3 guesses as to where he resides.

Because of the low cost, I have no regrets about buying the 2 bulbs. It's a learning experience, and I'll keep digging and learning. This will either be do-able or it won't, not without substantial changes to the method. $15 for 2 LED bulbs was worth the risk to me, and of course, I'm glad it wasn't more.

Cheers, Tom.
 

fish4fun

Joined Feb 24, 2009
8
Tom,

I am working on a similar project, that is line-driven dimmable LED lighting. I noticed that you asked about the series capacitor, and no one properly explained it's function. The Series connected capacitor acts to limit the current to the LEDs which can be calculated as follows:

I = 2 * Pi * F * Vrms * C

Where:
2 * Pi = 6.282
F = 60hz
Vrms = 120V
C = 2uF (according to your image)

SO:

2 * Pi * 120 * 60 * (2*10^-6) = 90.48mA

The problem with attempting to "Dim" this type circuit is that your triac based dimmer is chopping the AC waveform. If the firing angle is prior to the "peak" of the line voltage then essentially the LEDs will not be "dimmed". Dimming will occur if the firing angle > the peak voltage, but only until the firing angle falls below the LED threshold voltage. So, as you have discovered, the dimming action will only occur in a very narrow region of the dimmer's range.

A rheostat type dimmer should produce the effect you are looking for. Another approach would be to build a circuit to alter the frequency of the AC voltage. Lower frequency = Dimmer, higher frequency = Brighter. This type of circuit is more complex, but is potentially more efficient.

Fish
 

Thread Starter

Tom Kay

Joined Feb 10, 2009
35
Hi Fish;

I hate when this happens. I just wrote a detailed reply, and the website coughed and wouldn't post it, despite the fact that I logged in again. So this won't be as long or comprehensive as the message I just lost.

I largely understood what you mentioned. Not entirely, but somewhat.

I am looking at 2 main approaches to LED dimming;

1. Lowering the voltage and using a PWM.
2. Leaving the voltage alone and using a triac if possible.

I found a method by National Semiconductor that might work. I found a chip, LM3445, plus their demo board that uses a triac in combo with their circuit to dim LEDs in series and parallel strings. They even have a circuit designer on the website, although I couldn't really get it to work yet. I'll call their tech guy, Bill McCulley (1-800-272-9959) to help me work it through.

Maybe you could have a look at it and see what you think. Google the company and enter LM3445 and also look at their demo circuit board. It's for sale, all soldered up with parts and ready to install.

I'm not opposed to buying an off-the-shelf solution if it exists affordably. My goal is dimmable LED lighting in the entire house, starting with my home theater as my "lab". If their solution works, then it should also work with Smarthome's IR remote controlled dimmers, assuming they're triac based, such as the Insteon series. That would be the best if it worked.

I have not played with a rheostat for dimming experimentation, but wouldn't that just suck up energy and dissipate it as waste heat? Also, if you go below a threshold voltage or current, won't your LED just turn off? Again, I haven't tried it, but this would be my concern.

Well, this is shorter than my last attempt. Probably better, huh?

Keep us posted on your success, OK?

Cheers, Tom.
 

fish4fun

Joined Feb 24, 2009
8
Tom,

PWM will work for a constant current source (read: "power supply"). PWM can be employed in one of two ways: You can use PWM to create a variable constant current source, or you can use a constant current source and use PWM to adjust the period of time the LED "sees" that current source.

The TRIAC, in the case of your LED project is likely going to fail without rather drastic circuit modification. Let me briefly outline why. A "TRIAC Dimmer" Chops the AC waveform to limit the amount of power to a primarily resistive load (like an incandescent light bulb). Your LED light bulb uses a capacitor as a reactive current limiter. In order for the capacitor to function properly it needs an AC waveform. The Chopped up AC waveform does little to limit the power delivered through your capacitor to your LEDs until the firing angle of the TRIAC reaches a narrow region in the 2nd and 4th quadrants of the AC signal. Essentially you are attempting to use two incompatible methods of current limitting.

Using a rheostat will, in theory, "waste more power" in the form of heat, but let's look at exactly what we are talking about:

A 10W LED Light Bulb operating @ 120Vac is drawing 83.33mA and has an impedance of 1440 ohms. If our Rheostat is to dim our LEDs then it needs to be able to reduce their power consumption from full rating to about 1/10th of full rating. So, 1W to our LEDs = I^2 * R ==> I = .02635A. P = I * E ==> .02635 * 120 => 3.162W. Our "Wasted Power" is then 3.162 - 1 = 2.162W. This would be our maximum power loss for 1/10th power to the LEDs. If we compare this to an incandescent bulb, 2.162W "wasted" is a HUGE savings. Can it be better? Absolutely! In a properly designed system we might waste as little as 0.3W dimming our LED fixture. The cost of electricity is ~$0.10/kWh. If we used a 100W light bulb for 10 hours this would be $0.10. If we dimmed this same light bulb using a TRIAC dimmer to 10% of it's output then the same 10 hours would cost us ~$0.01. If we used a 10W LED buld @ full output then 10 hours would cost us the same $0.01 as the "dimmed" incandescent. If we dimmed the LED to 10% and "wasted" 2.162W our 10 hour cost would be $0.003162, so we could run the same "dimmed light" for ~3 times as long for the same $0.01. Compaaring the savings at full power we could run the LED 10 times as long for the same money it would cost to run the uncandescent. So, using a Rehostat we are haggling over saving 10x vs saving 3x. But @ 3X we are only talking about 31.62Whr for 10 hours.

Is there a "better way?" Absolutely, but not using "off the shelf" LED lamps with "TRIAC Dimmers". Could you design the current through an LED array to achieve both "dimming" and Maximum efficiency? Sure, and that's what I am working on, but you can't do it with a combination of off the shelf components thrown together.


I am going to start a new thread about the design I am working on so as not to hijack your thread


Fish
 

Thread Starter

Tom Kay

Joined Feb 10, 2009
35
Hi Again Fish;

Thanks for doing some math for me to illustrate your point. What you say makes sense, and I guess "wasting" heat at the rheostat is the lesser of two evils, as compared to powering, say, halogens.

What you have pointed out about triacs and LEDs not mixing well is both true and confirmed in my mind, based on the 2 72LED bulbs I bought from Ebay. So something more is needed than a standard off-the-shelf solution.

However, I am getting excited about the prospect that at some point soon, some company will come up with dimmable LEDs using basic dimmers, or proprietary dimmers that use 110VAC. And in fact, that is what I'm looking into now, this dimming circuit from National. I can't say yet if it will work, but I intend to peer deeper into it.

I'm a bit odd when it comes to "saving money" in that I don't mind investing a bit up front (even more than the cost of the LED bulbs, if you include the added hardware) as long as I can stick it to the utility company from that point onward.

I'll keep an eye on all the LED threads here, post what I find out, and please do the same. I did read your new thread about the panel, but is your ultimate goal to dim household LEDs, or something else? Just curious.

Cheers, Tom.
 

italo

Joined Nov 20, 2005
205
Lexon devices are bright as the sun. But the design will have to be as close to the belly button as you can get. They are great and uses much more power then leds. temperature dissipation on the devices is a problem 1mm can dissipate 7-10 watts so means to remove the heat is of concern.SAMS sell ready made lamps of LED 3.5w comparable to 45w. They are also dimmable i tried one got good results. Somebody claim that dimmers only works on passive resistance that is not true. I f the circuit is designed correctly with kickback protection and inrush protection it works forever at least mine does 20 years and counting
 

Thread Starter

Tom Kay

Joined Feb 10, 2009
35
Hi Italo;

Thanks for the response. Not quite sure what you mean by belly button, but then I don't get out much. Plus I'm Canadian, so I only see belly buttons 3 months out of the year. The other 9 months it's like a cryogenics lab.

So, I looked up Lexon, thinking that you mean Luxeon, and BOTH are companies that deal in LEDs. How's that for unlikely?? Lexon seems to make small battery powered LEDs and I think Luxeon makes the really bright high-flux LEDs.

Anyway, so you've gotten long service out of the one you have. And it's dimmable. Can you give me some details, such as what model, voltage and where you can buy them now?

Thanks, Tom.
 

Bernard

Joined Aug 7, 2008
5,784
Lots of interest in dimming LED's. look at thread of malaybiswas on Projects.He found a dimming unit, 120V AC in, driving 3 strings of 6 ea bright 350 mA, 3.3V LED modules.Dimming o-10V DC ,pot, or switch.
 

Thread Starter

Tom Kay

Joined Feb 10, 2009
35
Hi Bernard;

Thanks for the example of malaybiswas's design. I see some similarities, as you pointed out, but some differences of course too. So it was worth having a look at his application.

I think I'll post the circuit I was mentioning (National Semiconductor's) in this forum and ask questions about modifying it to add more LEDs if possible. They make their circuit diagram and detailed design info completely available to the public, and my approach would be to make the dimmer "be all it can be" by getting pretty close to its max current and/or voltage output. This will quite possibly involve some component changes on their demo circuit board. I'm a cheapskate, so I'd like to have a reasonable likelihood, through educated forum guesses, that their hardware wouldn't go POOF as soon as I plug it in, with lots of LEDs attached. Of course I'd start small, then work up to some sort of max acceptable output (total load on the board).

Even though I wouldn't have invented this circuit, it would be quite exciting to me if this off-the-shelf solution to dimming LED bulbs, with a basic Triac, without changing the house wiring or voltage, would work as claimed by National. Hopefully we'll see soon. It then opens up all kinds of LED lighting options to the homeowner, not to mention the lighting industry if they buy National's parts.

So thanks again, and stay tuned for more!
Cheers, Tom.
 
Top