LED Dimming

The correct way to control any LED bulb is with a square wave chopper circuit. This is the same technique used to control electric drill motors which are also running on DC. Essentially, the square waves remain at the same amplitude, but get shorter to dim the LED. You can purchase a little goody to do it from CED company for about $85.00 and Phil, the manager at the Fairfield, California branch was very knowledgable and helpful. (707-425-0137) I am working on changing all my lighting specifications for custom wine cellars over to LED technology and there is some exiting new technology evolving. As an example wireless individual switching for large arrays... Paul Wyatt www.Customcellar.com

Thanks Paul... I am somewhat aware of that wireless technolgy, it works really well with the stage and studio industry I am told.

Actually, we have some of the latest LED Soft Neon technology where you can use DMX512 and interface with your computer. That way you have total control of your system.

Thanks for the tip...

Don Campbell

In theory it shouldn't matter. All a standard dimmer is is a crude duty cycle modulator which should work fine. However, the power lines can be hostile and LED's are not as forgiving as incandescents. What may be required is better transient suppression. Also, if the led's are not wired in two strings that conduct on both half cycles then they are even more susceptible to transients on the "off" phase since LED's typically don't have a very high reverse voltage rating.

Hi. I don't see how a standard dimmer can modulate frequency, it is designed to modulate amplitude and the LED will, at the very least, switch off at its operating voltage threshhold. It must have a power level above this threshhold and dimming occurs when the input power is reduced by making the square waves shorter. I agree with you that a smoother supply would obtain from using a full rectifier because these little goodies are a bit fragile. A stabilized DC supply for a twelve volt LED must not exceed 14 VDC. A lot of low priced transformers have quite a large start up voltage spike before the core gets stabilized and once an LED goes OOPS, it is gone.

Perhaps the operation of a common dimmer for incadescent lamp is unclear to you. A triac turns power full on at a variable point in time in each AC half cycle providing only a portion of that half cycle power to the lamp. Now in an incadescent lamp the filament responds proportunately to the amount of power applied. This is no different in effect than any other way of switching the power on for a variable portion of 100%. Frequency modulation is not inherently duty cycle modulation unless you have a fixed on time that occurs more frequently in proportion to a frequency increase.

I do understand a little about how AC dimmers work, however we are talking about the power interface with an LED which would be DC and my comment about operating voltage threshold should still be valid. Can we attach wave drawings here? (And how about some circuit diagrams...) If there is an easier way of doing this, I would like to know, but the bit about bulb failure is not acceptable for my product line...

I was addressing the original comment: "We have had reports of premature failures of LED bulbs used on 120 volt dimming circuits." I'm not sure of the correlation to your comments. Yes all LED's run on DC so in an LED bulb we assume there is a rectifier to create DC for the LED's. The original comment was really the only issue I intended to address. I am an electronic designer with a great deal of experience using LED's in hundreds of circuits so my apologies if I misunderstood the nature of this discussion. May I point out that there is no such thing as a "12 volt LED", all LED's are current mode devices and you must use a means of current limiting so for an LED on a 12 volt circuit that might be operated at 20ma and exhibit a 2v forward voltage you would drop 10 volts at 0.02 amps (10/0.02=500) and so use a 500 ohm resistor that would dissipate 0.2 watts of heat. If you raised the voltage to 14 you would only increase the operating current 24ma which would do nothing more than make it a little brighter. There may in fact be a problem with multiple LED lamps on 120 VAC due to the fact that the ratio of voltage dropped on an limiting resistor to the voltage of the string is very low which would cause voltage transients on the AC to produce large current spikes in the LED's but in fact most LED's can absorb quite large current spikes of short duration.

Don, can you make measurements on a similar setup at your own plant? Alternatively, can you send a field engineer out to the customer site? Some hard data would be invaluable at this point.

I'm assuming your customer is using a standard triac-type dimmer in his circuit, together with the assumption that your product is designed (and intended) as a drop-in replacement for a standard incandescent bulb, yes?

Given these assumptions, there is no reason this type of dimmer should be killing your bulbs. In terms of inrush currents, LED bulbs are far tamer than incandescent bulbs. Incandescent bulbs have a very low filament resistance when they're cold (off). The filament resistance increases significantly as the filament heats up (time-scale here in milliseconds). As a triac-style dimmer can apply the voltage to the incandescent bulb at the peak of the AC cycle, the bulb is hit with the full peak voltage, which is around 170 volts. As the filament is cold, the resulting current spike will be quite large - basically 170 volts/(filament cold resistance). It only gets better after that, as the filament resistance increases as the filament heats up, plus the fact that the voltage in the AC cycle at this point can only go lower until it crosses the zero point, at which the triac shuts off. Not only is this the worst-case condition for incandescent bulbs, but it happens 60 times per second until the dimmer setting is moved elsewhere or shut off. With LED bulbs in this type of dimmer circuit, the situation is a somewhat different.

Taking a look at the LED forward voltage vs forward current, you see that the forward voltage is highest at the lowest forward current - the opposite case, as it were, than the case with incandescent bulbs. There is no inrush current spike. If there is any "inrush current," it is due to the charging of capacitances in the circuit which, in the case of an LED bulb, is insignificant to this type of circuit. Nor does it matter that your bulb is seeing a 170-volt peak AC voltage, as it would see this voltage anyway, whether or not there is a dimmer in the circuit! So what is causing the failures? Any number of things.

If the circuit contains an inductance, you might get a voltage spike if the current flowing through the inductance is suddenly interrupted. As triacs shut off at the end of the AC cycle, I'm not quite sure what mechanism would be suddenly interrupting a current flow, much less understand why there is an inductance in the circuit to begin with - unless there are other lamps in the circuit which might be, perhaps, some kind of flourescent fixture with a ballast (compact flourescents don't usually use ballasts). Ballasts are nicely inductive, and are known triac-killers. But as I don't have any knowledge of the specific case, here, I can only speculate.

There may be other sources of voltage spikes nearby (electrically speaking) whose spikes are making it into your LED bulb's circuit.

One LED killer I'd be on the lookout for is the case where the LEDs in your bulb are being subject, for some reason, to high reverse voltages.

Back to inrush currents: the fact that LEDs are commonly dimmed by means of PWM techniques, which use square waves, is an argument against the failure being due to inrush currents. These square waves often have leading and trailing edges spanning the full voltage range in a matter of tens or hundreds of nanoseconds, and sometimes less than 10 ns or more than a few microseconds. This is a strong argument against your bulbs being killed by fast turn-on transients. And so there's simply no mechanism intrinsic to your triac-style dimmer that could explain the early demise of your bulbs as being due to fast turn-on currents.

So, in summary, I'd look for:

1) High applied reverse voltages. There are fixes for this, BTW.

2) Stray (or deliberate) inductances which really have no business in a triac circuit anyway.

3) Voltage spikes.

To reiterate, I'd get some hard data just about now.

Take care,
--Europium

BTW, high reverse voltages are not a problem intrinsic to the bulb's presence in a dimmer circuit. If reverse voltages were killing your bulb, the bulbs would be dying whether or not they were in a dimmer circuit. The fact that this is true can form the basis for an experiment:
Install one or more bulbs from the same lot in the dimmer circuit. Also install one or more bulbs from that lot in an adjacent (electrically speaking) circuit which is not controlled by the dimmer. If either circuit has other loads connected which differ significantly in some way - especially if either one contains inductive loads - might tend to invalidate the results. Try get as close as possible to matching conditions between the two circuits without altering the dimmed circuit in any way, as might be done by removing fluorescent loads for example. Once you're fairly certain these condtions are met, see if your bulbs have comparable lifetimes (patience may be required). If the bulbs die prematurely in both cases, the problem is not the dimmer.

--Europium

I would check to see if there is a power supply or similar across the line that may allow a small amount of leakage current while off which may forward bias the LED's and then make them more susceptible to line transients.

Another thought might be to use a Reverse Phase Control Dimmer. This is a dimmer for ELV's that uses FET's instead of Triacs and turns on at the zero cross and off at some point during the line cycle. This would be "gentler" on the LED's as there is no sudden application of large (170V Peak) voltages. The problem with this may be finding this type of dimmer in your power range.

Shawn

If the circuit contains an inductance, the act of interrupting the current flow at some point in the AC cycle will generate a voltage spike. Guaranteed. And depending on the amount of inductance present in the circuit, and depending on the "internal resistance" of this inductance - however realized in practice - and depending on whether the effects of this inductance are mitigated by some sort of "snubber" or flyback diode (the latter being unlikely, considering this is an AC circuit), this voltage spike may be quite large. Large enough, perhaps, to turn the more tender semiconductors present in the circuit into foul-smelling globs of incinerated silicon/epoxy toast?

Moreover, I suspect Don's concerns are specifically directed toward making his bulbs truly drop-in replacements for incandescent bulbs. If so, it wouldn't especially be the most cost-effective approach for his company, as some kind of roving Good Samaritan, Inc., to go around reengineering every customer's application that doesn't happen to play nice with his product. On the contrary, I suspect Don's concerns are primarily with making his bulbs accommodate the customer's application - whatever it is and without modification - so long as his bulbs are not subject to conditions beyond what might be expected for any garden-variety incandescent light bulb - not the other way around.

--Europium

If I read well You want to use a standard triac dimmer to dim the light that comes out of a LED bulb.

In short: you are killing the LED bulb as fast as possible.

Did you ever try to read the output of a dimmer with an ossciloscope? A triac generates spikes when the current is swiched, theses spikes can be more than 3 times the nominal voltage. Only good TrueRMS multimeters can handle the signal to give you a reading.

I can write you a full story on how a phaze cutting triac fucks up your complete setup.

Are the internal electronics designed to work with this signal?

Do it correct and use a PWM system as described above.

Thank you Gwen.

Here are some screen shots of an actual waveform from an actual AC dimmer running a 100 watt incadescent lamp. I can't see anything scary here. This is off of a 100 mHz sampling scope so I think it is fast enough to see anything of concern.

Sorry, I din't see the pictures but I have had serious trouble with Triac driving a 2kW heating system. The spikes made voltage measurement and shunt type current measurement impossible.

We didn't see this in our lab when we first tried the setup, the dimmer we used at that moment, was a high class type from a renowed Belgian make (Niko) as it was available.

when we build the panel we had to go for another type from another supplier. only little difference: the Niko dimmer had a loww pass filter integrated so that your HiFi doesn't notice him. The finally used type didn't had this filter.

What I try to explain is the fact that LED is a new technology that incorporates a lot of driving electronics (White LEDs) to get the exact colour. A filament bulb has nothing of this. These driving electronics will see the steps as walls. but as 120 of these walls come by each second (over there at least) you need one failure to kill something in your system.

If you want to dimm LEDs you will have to incorporate this in the driving electronics capabilities. How to tell these electronics the level? another discussion

But here we might have a task for us all: write a paper that combines the knowlegde so that a first step of a standard is set. Or propose a standard and explain and help Cambell to implement it. I'm willing to help and test it.

White LED's are blue LED's with a phosphor that converts the blue light to white. There are no fancy electronics to the LED it's self. Your problems with measuring may be due to the fast rise when it turns on but this should not cause large transients certainly not 3x the line voltage, yet it does cause RF noise since when you do an FFT on a fast rising edge you will see that it comprises many high frequency components and it is these that the low pass filter supresses.

I came across this post when I was looking for methods of dimming LED Circuits. Granted the Triac or SCR based designs are no way to directly control an LED chain in common current bar "resistor based" topologies. With "current source" Buck type drive topologies running off a rectified 120VAC having the ability to shutdown or enable the circuit via a PWM channel. Is there a way to to take a Phase dimmed AC signal and convert it to PWM with a duty cycle congruent to hacked wave? I am trying to find resources for obtaining any information on Discreet means for this converter. I have very little experience with this method of conversion yet have seen it used in various circuits for PFC buck systems (power supplies) or dimmable florescent ballasts. it seams that there is a major lack information regarding this type of circuit out there on the WEB!!

See here: www.elecosn.com

Here are dimmable LED Bulb, and LED incandescent bulb

See here: www.elecosn.com

Here are dimmable LED Bulb, and LED incandescent bulb

Hey Guys, I just wanted you all to know that all comments, and expertise is GREATLY appreciated!

We are not a huge company, but we do see a huge opportunity for providing the correct products that will indeed help solve some of our more pressing problems, energy charges, pollution of our land and water ways, (Mercury, and other heavy metals), as well as saving people labor just to change light bulbs!

I am passing this valuable information along to the proper people, and hopefully our industry will improve as a result..

Sincerely,

Don Campbell

Campbell Lighting

By the way, Paul, the answer to your question about cutting the 2 wire, Mini "Soft Neon", 80 LED per 40 inches, IS: you can cut it at approx. every 20 inches.

This product comes in Red, Yellow, and Orange with power consumption of 8 watts per 40 inches, and Blue, Green, and White, with power consumption of 5 watts per 40 inches!...

We have LED bulbs and LED dimmer.

We make them work well.

Our LED bulbs can work at 120V, 220V, with white,red, green, RGB colors.

I think LED lighting will be the future of lighting definitely

LED dimmers are available at digital lighting systems, inc , www.digitallighting.com Manufacturer located in miami florida

•The users can use inncandescent lamp dimming controller and halogen lamp electronic transformer to do the dimming, not need to have the specific LED dimming controller
•The dimming functions are integrated into the LED driver, not need additional Signal Pins (PWM Dimming Controller) for dimming. The bulb can keep only two pins that is exactly the same as traditional halogen bulbs.
•What the users should do is just to connect the bulb to traditional electronic transformer, which is used by halogen lamp, and incandescent bulb controller
•NO lower lifecycle Liquid Electrolyte Capacitors applied in the drivers, so the lifecycle of the LED Driver is the same with the LEDs, that guaranteed the extremely long use life of the bulb
•Our 4.2W EverBright is the only LED Lamp that can replace 50W Halogen MR-16 Lamps with similar brightness, and the payback from the saving power bill is only 6 months in condition that the bulb are 24hrs turned on. Most of other LED Lamps from other providers are not able to replace the 50W halogen lamps because of the insufficient brightness

http://www.elecosn.com/pro/bulb/spot_3.html