Why do LED lamp dimmers in Australia use trailing edge control?

[Konkedout]

I live in USA where the mains voltage in 120VAC 60 Hz.  In my house we have various combinations of (incandescent and LED lamps) and 2-3 styles of dimmers including digitally controlled and old fashioned knob rheostat type.  I believe that all have triac output and control the leading edge (or turn-on) timing of the applied voltage.

In fact I have done a quick search of vendors: Home Depot, Lowes, and Amazon (all for USA) and could not find any mention of trailing edge dimmers.  I think that it is not a "thing" here in USA with 120 VAC.

I am a power supply designer and it is not obvious to me why trailing edge control might be advantageous.  Certainly the higher mains voltage in Australia is somewhat less convenient for driving LED lamps, because LEDs are low voltage devices.

I can imagine that if there is significant inductance in the wiring between the dimmer and the lamp, there may be large inductive voltage spikes when the current flow is suddenly interrupted.  That would seem to be a disadvantage of trailing edge control.  I assume that a snubber could address this issue, but that is more or bigger parts and a bit of added power dissipation.

Does anyone understand the reason for trailing edge dimmers in Australia?

[IanB]

It is my understanding that dimmers sold in the USA marked "suitable for LED lamps" are trailing edge control. This apparently works better with the circuitry inside dimmable LED lamps (less flickering at low light levels, more stable control). It would apply to nearly all dimmers currently on the market. They don't say "trailing edge" on the packaging or literature because they assume the general public would not have a clue what that means.

However, I have found that high quality LED lamps don't have any real problem with the older leading edge dimmers. So I'm not sure how much it really matters.

[Doctorandus_P]

They are also used in the EU.

Leading edge dimmers are traditionally built with triac's, and those switch very abruptly, and create a lot of electrical noise while doing so. When you use a FET, you can control the switching speed and lower EMI.

But that does not really explain the leading versus trailing edge.
One reason could be that with trailing edge, you can give some power to the load, and then turn it off when it has had enough for the current (half) period. With leading edge dimming, you just set a phase angle, and then just hope that other parameters do not change (too) much.

With trailing edge, you also have better current inrush control. Because the current to the load gradually rises with the input voltage, it is feasible to detect fault conditions such as a short, and turn of the FET before the current gets so high that the fet gets damaged.

[Konkedout]

It is my understanding that dimmers sold in the USA marked "suitable for LED lamps" are trailing edge control. This apparently works better with the circuitry inside dimmable LED lamps (less flickering at low light levels, more stable control). It would apply to nearly all dimmers currently on the market. They don't say "trailing edge" on the packaging or literature because they assume the general public would not have a clue what that means.

However, I have found that high quality LED lamps don't have any real problem with the older leading edge dimmers. So I'm not sure how much it really matters.

Thanks a lot.  Yes all of this makes sense.  Don't confuse us Americans with technical jargon.  Beyond that, I am getting a drift/hunch about this:

1) I know that triacs and other thyristors have a minimum "holding" current.  This is a minimum current level beneath which the device may turn off.  There is probably some noisy cycle-cycle variation in the current level at which an operating triac actually turns off.

2) LEDs are much more efficient so draw much less current.  The Triac holding current would be a higher percentage of the lower LED operating current than of the higher operating current for an incandescent or halogen lamp.

3) If the power level is the same, a 240 VAC LED lamp ought to draw about half as much current as a 120 VAC LED lamp.  So the triac holding current (and the noisy cycle-cycle variation) would be an even larger percentage of the operating current when the mains voltage is 240 VAC.

4) It seems possible that the higher voltage triac required for the 240 VAC leading edge controlled dimmer may have a higher holding current than a lower voltage rated one for 120 VAC.

5) LEDs respond much faster than incandescent or halogen lamps.

So all of this suggests a reason why noise and flickering might be more of a problem with triac-based leading edge modulation and with LED lamps and operating from 240 VAC.

Ordinary triacs must rely on leading edge control, so trailing edge control must use some other sort of switch, such as a MOSFET.  This would not have the issue of holding current which could possibly be at the root of flickering.

This is all a theory which I am putting together.  (??)

[johansen]

1) I know that triacs and other thyristors have a minimum "holding" current.  This is a minimum current level beneath which the device may turn off.  There is probably some noisy cycle-cycle variation in the current level at which an operating triac actually turns off.

2) LEDs are much more efficient so draw much less current.  The Triac holding current would be a higher percentage of the lower LED operating current than of the higher operating current for an incandescent or halogen lamp.

3) If the power level is the same, a 240 VAC LED lamp ought to draw about half as much current as a 120 VAC LED lamp.  So the triac holding current (and the noisy cycle-cycle variation) would be an even larger percentage of the operating current when the mains voltage is 240 VAC.

4) It seems possible that the higher voltage triac required for the 240 VAC leading edge controlled dimmer may have a higher holding current than a lower voltage rated one for 120 VAC.

5) LEDs respond much faster than incandescent or halogen lamps.

the problem is the LED driver and the fact that it has to convert the ac into dc first and store a minimum amount of energy, while holding up the led output somehow with a minimum of energy storage to reduce cost. 

its pretty typical for 600 volt rated triacs to be used on both 120vac and 240vac rated dimmers, so they will have similar holding current.

keep in mind that the flicker isn't from the turn off time, which happens as the voltage is falling and has already charged up the cap in the LED lamp, its from the variation in turn on voltage.


because the dimmers are operated in series with the load and very few of them have a neutral wire connection to facilitate intelegent turn on time control, the turn on point depends on a minimum amount of current flowing through the load to activate the triac. 

[Konkedout]

As to energy storage; I am not sure about that.  I have opened up the rheostat triac dimmers and I do not see much in the way of (capacitance or inductance) for storing energy.  I do think that the LEDs blink at 120 Hz in North America or 100 Hz in most of the rest of the world.  If the lamps contain energy storage, I sort of doubt that it will be enough to ride through several milliseconds.

But comparing an incandescent to an LED:  The incandescent will act like a resistor (if its temperature is constant over an AC cycle) and will have about 0 volts across it until the triac turns on.  With the LEDs the voltage across it may dwell at some significant level before triac turns on.  So you may have a point that the turn-on timing of the triac may be noisier when driving LEDs.

My house was built around 1990.  The wall mounted light switches do have neutral wires, and the newer (digitally controlled) dimmers which I have installed in place of the switches require a connection to that neutral wire.  The old rheostat operated dimmers did not use the neutral connection but at least in my situation they work satisfactorily, even at low levels.  Actually the one which exhibits a bit of noise is one of the digitally controlled ones which as a neutral connection.

It happens that my ~20 year old oscilloscope died a week ago and I am waiting for delivery of a new one.  I would not connect a probe tip directly to the mains, but by holding it near the dimmer wiring I may be able to get a crude view of the waveform....   Even with that I hope to be able to discern leading edge versus trailing edge control.

[tooki]

As to energy storage; I am not sure about that.  I have opened up the rheostat triac dimmers and I do not see much in the way of (capacitance or inductance) for storing energy.  I do think that the LEDs blink at 120 Hz in North America or 100 Hz in most of the rest of the world.  If the lamps contain energy storage, I sort of doubt that it will be enough to ride through several milliseconds.

1. He didn’t say energy storage in the dimmer. He said energy storage in the LED driver.

2. You need to go learn how LED replacement lamps work, because right now you don’t have any clue. 100Hz blinking LEDs would be insufferable.

What we basically never do is just directly apply mains AC to a string of LEDs. LEDs don’t like reverse voltage very much, but even if you use antiparallel LED pairs, it’s flickery enough to be unpleasant as lighting.

Some very cheap LED lamps use capacitive droppers.

A typical LED driver, though, is fundamentally a switching power supply. Yes, it has energy storage; it has to. A single cycle of 50Hz mains is 20ms, so to smooth that out you’d need at least enough energy storage to make it until the next cycle.

[D Straney]

I worked in consumer lighting for a while, and spent way too much time looking at the system dynamics of lamp & dimmer interactions for an induction lamp we were developing, so let's clear up some misconceptions and lay down some truths.
Caveat 1: my knowledge is about 6-8 years out of date, but I doubt anything has changed drastically since then.
Caveat 2: this was all in the US, so I don't know much about the situation in other countries.

Leading-edge vs. trailing-edge dimmers: Leading-edge ones are the simplest (hence why they're the most common) because they use triacs, which do require a continuous small hold current to stay on.  Trailing-edge ones necessarily use either MOSFETs or IGBTs, and are more complicated as a result.

Leading-edge dimmers create a large dV/dt at turn-on, which creates EMI as well as transient-response ringing issues with the lamp's EMI filter, if you aren't careful to damp it properly and add RC snubbers in the right places.  Trailing-edge dimmers do create a dI/dt transient at turn-off which means the stored energy in the wiring needs somewhere to go (if you don't do anything, it'll result in a large voltage spike), but my best guess is that in practice this is easier to deal with and redirect into a snubber cap, because of the small current drawn by each lamp (and therefore lower energy), and the fact that the inductance of the wiring is probably swamped out by each lamp's EMI filter inductance.  I don't have any direct experience with trailing-edge dimmers.

"LED-compatible" does not mean "trailing-edge" in the USA: At one point, I bought a wide range of dimmers, from old to new, simple to fancy, to experiment with and provide a test platform for the company's lighting products.  Not a single one of these used trailing-edge control, even the newest, fanciest ones that were advertised as being compatible with every LED lamp out there.  Every single one used a triac, and a timing circuit made out of a combination of passives & sometimes some diacs.

What "LED-compatible" meant for the dimmer, is that (1) the timing circuit was relatively insensitive to load impedance, and (2) they used modern extra-sensitive triacs with low hold currents (a few mA, which was easy for the lamp's capacitance to sink @ 60 Hz).

Energy storage and dimmer compatibility in lamps: Good-quality non-dimmable LED lamps would use a bulk capacitor for energy storage, and then "peak-charge" it from the AC line, to give a fixed voltage directly across an LED string (sometimes with a simple current regulator).  This bulk capacitor was for, as tooki said, storing energy across the 120 Hz cycle to provide a relatively steady light level.

Dimmable LED lamps would use a simple self-oscillating (2-transistor) boost or buck-boost converter to run open-loop at a constant(-ish) input current throughout the dimmer's on-time (therefore drawing well over the triac hold current the entire time) and dump that charge into an aluminum bulk capacitor connected to the LED string.  Smaller conduction angle → less power delivered → lower voltage on bulk cap, less current through LEDs.  The bulk cap is of course then responsible for filtering out 120 Hz ripple from the chopped input voltage, as well as the lower-frequency (1 Hz - 120 Hz) content created by the timing jitter in the dimmer, to prevent visible flicker.
The only other requirement for making a good dimmable LED lamp, besides the always-on converter and having enough bulk energy storage to smooth out the dimmer jitter, is having its EMI filter damped well enough (as mentioned above) against the leading-edge turn-on.  In my experience, any LED lamp that does these 3 things will work well with pretty much any normal dimmer, whether listed as "LED-compatible" or not.

I have seen & taken apart plenty of bad-quality non-dimmable LED lamps though which save cost & complexity by using no bulk capacitor & no energy storage, just putting the rectified AC line voltage directly across an LED string (as tooki mentioned).  They exist, they're out there, and they're AWFUL - the massive 120 Hz flicker is a bit too high-frequency to perceive directly, but any movement within their light creates all kinds of bizarre, distracting strobe-light effects.  I hate these with a passion and any architect or landlord who buys these should have their license taken away.


Anyways, I've got a whole bunch of reverse-engineered dimmer schematics and simulations if anyone's curious.

[coppice]

When I was working in the same place as a team developing LED lamp solutions I was amazed at the number of different types of dimmer they tested their dimmable designs against. They found a lot of variation in dimmer behaviour beyond the basic leading edge, trailing edge, and leading plus trailing edge categories, and had test setups with a sample of pretty much every commercial dimmer they could get their hands on.

[johansen]

When I was working in the same place as a team developing LED lamp solutions I was amazed at the number of different types of dimmer they tested their dimmable designs against. They found a lot of variation in dimmer behaviour beyond the basic leading edge, trailing edge, and leading plus trailing edge categories, and had test setups with a sample of pretty much every commercial dimmer they could get their hands on.

I, like D above. Have never found anything other than triac dimmers.

[coppice]

When I was working in the same place as a team developing LED lamp solutions I was amazed at the number of different types of dimmer they tested their dimmable designs against. They found a lot of variation in dimmer behaviour beyond the basic leading edge, trailing edge, and leading plus trailing edge categories, and had test setups with a sample of pretty much every commercial dimmer they could get their hands on.

I, like D above. Have never found anything other than triac dimmers.

It depends where you live, and what wiring practices are used. In a place like the UK there is usually no neutral connection at the lighting switches. So, people there use almost exclusively triac leading edge dimmers for consumer applications. Where both live and neutral are available at the switches there is a lot more flexibility, and a variety of dimmers designs are used.

[Marco]

The non remote controlled trailing edge dimmers here (Netherlands) seem to mostly steal a bit of power during switch on and every cycle, no neutral needed.

There's also a standard icon which depicts what it supports with a combination of R, L, C (with C being a capacitive load, so trailing edge control). Not sure if that's just a custom, some industry standard or something official.

[SteveThackery]

The non remote controlled trailing edge dimmers here (Netherlands) seem to mostly steal a bit of power during switch on and every cycle, no neutral needed.

Wouldn't it need to be during switch-off (live to neutral via the light fitting)?  There is no potential difference during switch-on, is there?

[coppice]

The non remote controlled trailing edge dimmers here (Netherlands) seem to mostly steal a bit of power during switch on and every cycle, no neutral needed.

Wouldn't it need to be during switch-off (live to neutral via the light fitting)?  There is no potential difference during switch-on, is there?

Some of them bleed a little current through the lamp when its switched off, so they can keep the remote control powered.

[Marco]

There is no potential difference during switch-on, is there?

Just put a high value resistor in series with the LED bulb at switch on, it will have potential across it.

I don't know what they actually do though. Surprises me Big Clive never did a tear down of one.

[Konkedout]

https://www.eevblog.com/forum/projects/why-do-led-lamp-dimmers-in-australia-use-trailing-edge-control/?topicseen#:~:text=What%20we%20basically%20never%20do%20is%20just%20directly%20apply%20mains%20AC%20to%20a%20string%20of%20LEDs.%20LEDs%20don%E2%80%99t%20like%20reverse%20voltage%20very%20much%2C%20but%20even%20if%20you%20use%20antiparallel%20LED%20pairs%2C%20it%E2%80%99s%20flickery%20enough%20to%20be%20unpleasant%20as%20lighting. 

Yes of course.  Something like a bridge rectifier (or two opposite phase half wave rectified strings but this seems less likely) and some sort of current limiting impedance seems necessary.  It seems possible that the LED's themselves might be made with enough current limiting impedance.

When you discuss a driver (at least in USA) there cannot be much there.  It would need to fit inside the envelope of an ordinary light bulb.

I believe we have been living with 120 Hz blinking fluorescent lamps for many years.  Yeah OK I would not have wanted that in a bedroom, but it is not so objectionable by most of us.

One other thing: To my understanding, white LEDs use a blue (or violet?) LED and a ~white phosphor.  Don't some phosphors have a slow decay time?

[D Straney]

One other thing: To my understanding, white LEDs use a blue (or violet?) LED and a ~white phosphor.  Don't some phosphors have a slow decay time?

Depends what you mean by "slow".  Measurable, definitely.  Nowhere near slow enough to filter out 120 Hz in any sort of meaningful way (the common ones I've worked with at least).
Interestingly, the energy storage in incandescent bulbs happens in the thermal mass of the filament: the vacuum within the bulb prevents convection, and the large length-to-area ratio of the filament makes the conductive thermal resistance high, which means that the temperature (and the light) decays on a 100s-of-ms sort of timescale - this is responsible for filtering the dimmer waveform.  The measure of "flicker sensitivity" of a lamp (light output variation vs. input voltage variation) is usually normalized by a normal incandescent's response.

[D Straney]

When you discuss a driver (at least in USA) there cannot be much there.  It would need to fit inside the envelope of an ordinary light bulb.

Also you'd be surprised how an entire boost converter can be fit within the screw base!    Some impressive mechanical tetris-ing going on there.

[coppice]

One other thing: To my understanding, white LEDs use a blue (or violet?) LED and a ~white phosphor.  Don't some phosphors have a slow decay time?

LEDs use a phosphor. However, they are very small, and use a very small amount of phosphor. There is too little to store much energy. The nice thing is there are no ions smashing into the phosphor, as in a fluorescent light tube, so you can push the phosphor hard and it still has a long life.

[Andy Chee]

[Konkedout]

Amazing that there is so much stuff in there.  He seems to have found an inverting buck boost converter in there. I admire his patience.

Think about it.....all that for less than $3 US?  60W equivalent 120V in USA for < $1.70 US (I saw that recently in Costco.  I do not remember the exact package size or price but it was something like 6 for $10.)

The other thing is that it is hard to imagine how that sort of assembly would be automated.