Replacing Traditional Light Bulbs with LEDs

The only way it will be 'lossless' is to have a huge number of leds in series, so there is no aditional circuitry. This would mean they weren't dimable...unless maybe you dimmed them with a triac.
Of course the leds themselves generate heat...I expect many labs around the world are looking for a cold light...
Del

Del

Using a triac would mean feeding them with a regulated mean-power and LEDs use a constant current. I was looking for the different LED driver IC' but the all have a max voltage of +/- 40v, so I have 180 to spare and I would like to do it without heating up a resistor. I can live with the heat generated by the LED's but al supplementary power loss would destroy my energy savings

JP

Looks like you'll have to design your own from discrete components then .
Good luck...nasty high votages...
Del

Del, even with a triac LED's on main voltage still need some incandescent resistance, or they go crazy, like dimming cfl or ccfl without resistance..

Donald

Del, stupid question but if you put two series strings of LED's in parallel with a single blocking diode (The diode is to make sure the PIV of the LED string is high enough to keep from burning out. LED's aren't optimized to be rectifiers after all and a couple 1N4007's are cheap insurance. You also might want to put an MOV across the line as well. It might also pay to put a small to medium electrolytic cap across each LED string between the first and last LED in the string but beyond the rectifier diode,to cut down on 50hz flicker, say 10-100uF maybe.) in each string, each connected with opposite polarity to the other so that one string is on when the other is off. (assuming 220V 50 Hz AC for Belgium, that would be about 73 LED's in each string assuming 3V drop for each LED)

Why wouldn't that set up be dimmable? And besides, aren't ALL AC dimmers triac based anyway?

I dunno...I'm an empirical Cat.
If I wanted to find out I'd try it, probably at a lower voltage for safety initially.
I'm a great believer in questioning the 'received wisdom' as just occaisionally we learn something new.
Del

If the issue is that there needs to be a resistive load across the output of the dimmer for it to function, you could always put a 480K 1/4 watt resistor across the line in paralell with the LED's. It would only dissipate abut 1/8 of a watt and draw half a milliamp at 240V but that should be sufficient load to stabilize the triac I would think.

It was Cambell Lighting (#10) who said we need some R.
I was willing to assume he was right 'cos of his name...mind I don't trust his opinion on cat food.
It would be interesting to try this stuff if I had the time, the parts, the inclination and now't else on my plate.
Del

I've been thinking about doing that myself to be honest.

Del/Hairlessimian/Campbell Lighting

Isn't the circuit I described not terribly unlike the LED based christmas light strings that you can buy for 10 bucks? Couldn't you test your dimmer circuit with one of those?

Not Sure about that circuit... But I do know that LED's are Much easier to dim on the low voltage side..

When you dim them with a mains dimmers, (My experience is with Triacs), they halfway work, and obviously that is NOT acceptable.. So, the easy solution is to add some incandescent resistance.

We have sold several hundred thousand electronic, dimmable, bulbs to our Poultry market, some cfl's some ccfl's and some LED bulbs, all working on 120v 60hz..

If the customer trys to operate them alone, they turn the Chicken/Turkey house into a disco..

Obviously not the best for content, fat, Kentucky Fried Chickens...

So, they simply add a 60 watt bulb to a whole 20-30 amp circuit, and everything works like it is supposed to...

By the way folks, for you that were interested in that CedarShield product, we are running a controlled test in a Darkling Beetle infested Poultry House, and we will report the results..

Donald

.....and the reason this research cannot be done at the local home improvement supermarket is what, please?

As I say, I'm automating my home with ZigBee (home build) so I would like to solve it custom build and in a 'eco' friendly way. (I have to power my PIC, my PWM-Current-driver and my LEDs without using large objects like a tranfo)

jp

Just to make sure I understand what you are trying to do...

In US, standard bulb base is E26 (~26mm, 120Vac@60Hz).
In Europe, standard bulb base is E27 (~27mm, 220Vac@50Hz).

Goal is to replace common E27 bulb with an efficient solid state LED equivalent.

Two questions...

1) How many LED's per "classic bulb" were you planning to use?

2) Dimmable from a standard wall plate (phase controlled Triac/SCR) dimmer control?

Hello,

I live in Europe so 220-240V is the main voltage and most light armatures are foreseen on a E27 socket.

1) - number of leds depend on light-flux and has to be the same as existing glow lamps.

2) Dimmable: the electronics have to be made, they are ZigBee controlled and (now) I use triacs with phase steering but I think going to LEDs, it would be better using PWM (all controllers are PIC and software is developed by myself so a switch to another technology is not a big issue)

Until now I have found the MAX16801 who can be fed with 265V DC but when I rectify 220v I have more than 300V so it is not directly usable.

JP

I'm going to offer some suggestions without going into too much detail.

1) Standard wall plate dimmers (triac/SCR) have a large installed base and are is still commonly used here in the US. I'd try to use a circuit design that worked with these dimmers.

2) I'd maximize efficiency, within practical limits, while making the circuit as SIMPLE as possible. Matching the efficiency of a CF bulb would be a good starting point.

3) I'd ignore cost for now since LED equivalent bulbs cannot currently compete anywhere near 1:1 with compact FL bulbs.

For a 15 Watt LED bulb (E27,220Vac,50Hz) that puts out ~870 lumens, my part list would be:

(1) fuse
(1) film cap
(4) diodes
(51) LED's
(3) resistors

Sorry, that's the limit on my info. Best wishes for your project!

This topic has already been covered.

http://cr4.globalspec.com/thread/42890

And the low loss capacitor voltage divider.

http://cr4.globalspec.com/thread/13464/Capacitors-as-AC-Voltage-Drop-Components#comment133640

Alternatively you could just buy an LED light bulb from your local electrical or lighting supplier, which are now fairly common (as mentioned in post #2). Much safer if you have little or no experience with 230V, and less likely to accidentally burn down your house.

Dimming circuitry can be implemented externally for strings of LED lights if necessary and there are circuits and designs on the major microchip supplier websites and the web (although I don't know how suitable they would be with a purchased LED light bulb). I believe you can also buy external dimmers that do work with LED light bulbs also (again try local electrical or lighting suppliers, or the good old web).

The topic has a main difference with all previous topics, the load varies. A capacitive voltage divider works only if the load is more or less constant. Here I have to dim the light so 'constant load' gone…

Using a standard plate dimmer is not an issue neither, it has to be remote controlled by RF-ZigBee protocol. As far as I know there is no supplier out there in the big mean world who can deliver such a device at an acceptable price for home use.

The only solution I see is a kind-off simple switching power device who can reduce the main voltage to about 30 tot 50Volts (it has not to be stabilized) independent to the load.

What the experience with 230V concerns, there was a time in my live I was developing and maintaining mercury controlled rectifiers for speed regulations on 1100V engines from steel mills. Just like the SCR's now but much bigger. J

I thought solving this would be a challenge to all clever brains posting here, so let your cogwheel run at full speed.

JP

I don't know if this would work but, instead of placing the led driver at the point of use (i.e. at the light socket), would it be possible to replace the light switch (or connect it in series) with a led driver (mains power to the driver and hook the driver outputs to the existing lighting circuit). I'm not sure if the resistance in the wires from the driver to the light socket would be problematic, but it might function. It would be interesting to see if it functions and what the efficiency is if it does work.

There are several LED drivers that accept <100~>230V AC input and output constant current low voltage DC (if you don't know where to look, check out www.ledsupply.com and look under the led drivers section). The Xitanium model they have is also capable of dimming.

Hi on7ami,

Would this idea be worth developing. The psu for the electronics is a simple cap/resistor dropper, this should work from the ripple of the 325Vdc, if not then feed it from the L230V. This is the way many washing machines etc power their electronics.

Then drive a mosfet with your PWM. You may need some resistance in the diode string to limit peak currents. fairly simple and not too much to lose if it goes POP!

Let us know your final design.

By the way, I've found an easy way to put diagrams up in CR4. Draw them any way you want then use "Snipping tool" to screen capture the piece you want. save to desktop and insert via the "camera" icon.

regards

Chas

Chas,

It's a good idee to separate the powering of the driverelectronics and the LED's. So indeed it is possible to separate the 'constant load' portion and the 'variable load' portion.

I'll work on this idée, I keep in touch with the results.

Tanks a lot for the idée.

Jean Paul

- Et en la casa del carpintero?

Regards.

A GA for good circuit.

I think that a smple circuit using a Bridge rectifier & a filter capacitor across the output is all what is needed.

Wouldn't do this myself except in protected space but since you use crap cirquits too here it is: Why does the voltage drop have to be resistive? (Meaning all losses) Drop it by a calculated value HV capacitor at AC in series and then full bridge (full is essential) rectify it and directly drive your LEDs. Actualy this topology is more like a current source and whether you drive one or five leds in series won't have much impuct on their current. Just calculate (or mesure) the capacitors AC current at mains (asuume you understand you need an non polarized >600VAC capacitor) to match your led sustaining current. Remember power ON/OFF must be controlled ONLY at AC side.specially if you use another capacitor for filtering the LED voltage or else you 'll hear a BIG and I meen BIG bang and lots of smelly smoke. Think diagramm not necesary pretty simple idea.

To give you a clue Iac=2*Vac/p*f*C (Farads)

Hey Minotaur Simplemind,

And if he does this how will his dimming work? I think the HF present in the waveform of the dimmer output will cause some interesting results. To the extent that the start of dimming may make the leds brighter to begin with.

I know from bitter experience that PWM inverters will burn out the capacitor/resistor dropper power supply circuits on washing machines etc.

Regards

Chas

As easy as any effort to dim about 30 ma. Not easy but not because of the capasitive current (triacs are driven by resistive V drop, in-phase with current so V phase is irelevant and will still stop conducting at Zero current) but because of very LOW current. Ever tried to dim a 30 ma load? (Close to your triac gate peak). Your results?
Just use more lines of leds in parallel and you're OK (change capacitor of course.)
And another: HF from a dimmer driving 30 ma?. Think I'm left little behind here .

Still crap cirquit though

This may be of use.......

TRIAC LED Driver Enables Full-Range And Flicker-Free Dimming

http://electronicdesign.com/Articles/ArticleID/21818/21818.html

http://www.national.com/pf/LM/LM3445.html

Designed for use with standard triac dimmers and powered directly off residential single phase voltage. There are probably other similar controllers also.

Hey,

Certainly worth trying, I ordered some samples to do some tests.

I let a message when finished (will take some weeks).

Thanks a lot and Greets,

Jean Paul

I have a slightly different requirement, but the solution should encompass this thread topic: I have an existing modern luminaire which has 5x 240V 40W Halogen lamps, and multiple white LEDs forming a pattern on the ceiling plate. These LEDs are all in series and fed from a bridge-rectified capacitive dropper with a 72V zener (36+36 1W) across the chain. The capacitive dropper is 0.33uF 400V in series with 150R 0.5W

I want to be able to dim this luminaire with a standard switchplate triac dimmer; the exact response of the LEDs is not a primary concern as most of the light comes from the halogens, BUT using the dimmer results in excess dissipation in the resistor, which then burns out due to the fast transients generated. (I've also had the zeners go S/C.) I can't find a realistic way of calculating the new resistor wattage required. Can anyone help or give any alternative thoughts?

Like your earlier enquirer, there is no room for a transformer anywhere or a more complex SMPS etc. - I just need to make the existing design reliable and long-lived with a triac dimmer. The basic design is inherently so simple, but the solution to dimming is not so obvious!

BTW, I can't emphasise enough that the dropper capacitor be at least X2 rated for safety - several of the posts seem to dismiss this aspect. An underrated capacitor can at least cause a nasty mess, and at worst destroy not only the load, but the environment around it!

Thanks, Sally Jelfs

Hello Sally

I am Not exactly sure about what you need, but if you would like to respond to me privately, I will describe what I have done, using led VERY bright 21 pcs sticks.

Donald

e-mail: lightmeup5@aol.com

Hi Sally,

If I've understood your description your circuit looks like this:-

For your application the only sensible answer is to go to a purely resistive dropper. Where this would normally be a nuisance, with a high power resistor dissipating loads of heat, it need not be for you. In place of a normal resistor use a suitably rated bulb. This is how I would suggest to do it:-

I would guess from your voltage that you are UK based. This probably means you would have a ceiling space to hide the resistor or bulb. Be careful about letting anything get over-hot.

Guessing at 30mA current for the LEDs , 72V string and 240V supply you will need 168V x .03A = 5K6 5W.

If the system is still working at this moment then measure the actual LED current when on full sine wave so that you can do the calcs correctly.

Regards

Chas

Your resistor dies above 8.6 V or 3.3 ma over it. Too much power to handle Current most low power leds can handle continusly is about 20ma . With 5W resistor you should be fine but check the actual current your cirquit draws at full power

Dear All,

A huge thank you for your prompt responses. I've put a bit more effort into the (re)development of this, so I'll reply to each in turn and add a few empirical comments at the end.

Capblanc: The circuit you provide is near enough exactly what I have (for anyone wanting to assemble something similar, you will want to add a 470K 0.25W bleeder across the X2 rated capacitor, as well as a feed fuse in case the capacitor fails.) You are correct that I'm in the UK, so it's 240VAC 50Hz operation ... 110V 60Hz systems will need a different capacitor. Unfortunately, in this application there isn't room in the floorspace above for a lamp (quite apart from any reliability issues and choice+mounting of a suitable lamp.) BUT there is room to upgrade the existing resistor to a 150R 50W (instead of 0.5W.) Why 50W? Because I have one (?!) It's metal cased for heatsinking and nestles nicely between the LED strips in the shallow fitting tray. It is 0.5" x 0.5" x 2" (+ tags) and tests show that this doesn't get vaguely warm even in free air, so I think I'll go with this solution. (The resistor is an Arcol HS50, available from RS Components - 'rswww.com' - part no 161-010)

Guest:The 150R 0.5W resistor is fine on a full sinewave as the capacitor provides the major impedance. But at lesser phase angles (i.e. dimmed), the resistor has to take over the work. It's more complex as the waveform in now non-sinusoidal, the effective voltage is now reduced, and the capacitor is providing only some part of the voltage dropping chain. Back of an envelope calculations suggest that 4W ought to be the dissipation, so your 5W ought to do the trick, but using 50W gives an enormous Factor Of Safety as well as very little temperature rise. (Don't forget those 5x 40W lamps hanging just below!)

Comments: For added protection against transients, I've added a 40V Transzorb across each 36V zener, as I've already lost one set S/C. The chain of 21 white LEDs has a terminal voltage of only 65V (although the zeners total 72V), so there's some protection without continual zener dissipation. It may be thought that adding a capacitor of -say- 10 or 20uF/100V would further add protection, and it does, but there is then no chance of any sort of dimming action - indeed, they slowly dim over about 30 seconds after being switched off! So that's out.

Just running the LEDs on the rectified AC waveform does result in a small amount of dimming at the very bottom of the range, so the apparent variable mark-space ratio of the supply doesn't do a good enough job for LEDs - although the Halogens are fine! So, more (design) work is required to convert the incoming waveform into something that will dim the LEDs with any sort of control range - they run on current, not voltage. (And don't forget that the supply for this system also has to come from this same varying RMS non-sinusoidal voltage.)

Anyway, thank you all again: It's not as easy as it seems! Regarding the original thread, I imagine that your E27 can be made with LEDs reasonably easily. (They're on sale in the shops over here). but the dimming issue is another matter. I'd like to see it resolved elegantly, but I'm not holding my breath just yet.

Hugs, Sally Jelfs.

A late thought about dimming white LEDs is that the probable reason for them not proportionally dimming with the change in phase angle is that they are not simple LEDs. They are actually UV LEDs with a phosphor to generate the visible white light - as in a fluorescent lamp - so it is most likely the phosphor persistence (afterglow)which sabotages the dimming attempt. I've not tried dimming simple LEDs (e.g red or green) to try to prove this theory, but if this IS the case, the solution may require some form of control with optical feedback to control the duty cycle?

Sally Jelfs

Hi Sally,

I think what you have may suit better than equal dimming. This way the halogens dim out allowing progressivly higher proportion of light from the LEDs then these too start to dim.

If you add a high power zener in series with your new resistor you will find that the dimming will come more in line with the halogens.

We made a ship's mimic and dstribution panel with led back lighting for the silouette and breaker labels and LEDs for the nav lights. They were all with different seris string lengths and thus resistances. All were dimmed from a common 24Vdc linear dropper which gave the very pleasing effect of dimming out progessivly the silouette first, then the labels and lastly the nav light mimics.

Chas

what is better, a blackberry bold 9700, or an iphone 3gs.... --- haha just kidding

hello jean paul,

we have exactly what you want, you can either call or e-mail my son, daniel campbell, and he will quote you.

don campbell campbell lighting inc 800-359-7265 fax 501-767-1234

e-mail: dan@campbell-lighting.com or campbelllightingco.com

don campbell