|Elliott Sound Products||Lighting Dimmers - Part 2|
Rod Elliott (Elliott Sound Products)
Page Created and Copyright © March 2015
The vast majority of all household dimmers are 2-wire. They are wired in series with the load and have no neutral connection. With incandescent lamps, this isn't a problem, simply because the lamp's filament provides the connection to neutral when the dimmer's internal switch (usually a TRIAC) is turned off.
This has worked well for many, many years. However, most modern energy efficient light sources are either compact fluorescent (CFL) or light-emitting diodes (LED) and these use an electronic power supply. No electronic power supply is capable of drawing significant current until the voltage differential across it is enough to cause the circuit to start switching, and the dimmer has no reference voltage until the load starts to conduct.
Herein lies the problem - the dimmer needs a reference (the neutral), but it doesn't get its reference for some time after each zero crossing. In the worst case, it might take 4ms or more before the dimmer has a reference because the load won't pass any current, and it can't start its own internal timer until that reference is provided. This creates a situation where the dimmer depends on the load and the load depends on the dimmer. That there are serious incompatibilities between dimmers and dimmable CFL and LED lamps should not be a surprise. That some actually work is the only real surprise!
There are many articles on the Net that describe the problems, but almost without fail, the author(s) blame the lamps. The lamp is and never was the problem - the real problem is the continued use of 2-wire dimmers where 3-wire dimmers only should be used. It doesn't help that there are few 'wall-plate' dimmers that are 3-wire, and no-one seems to think that they need to become the standard sooner rather than later.
Unless otherwise stated, the voltage used for all examples is the Australian/ European standard of 230V at 50Hz. A full cycle takes 20ms, and the peak voltage is nominally 325V. For 120V 60Hz mains, the period of one cycle is 16.67ms and the peak voltage is 170V. Readers in the US will need to make the necessary conversions to suit the lower voltage and higher frequency. The general principles are unchanged though.
|NOTE CAREFULLY: As described in Part 1, it is extremely important that the reader
understands that dimmable electronic lamps (both CFL and LED) are commonly claimed to be compatible with leading and trailing edge dimmers. With very
few exceptions, this is not true! Almost all electronic lamps draw a very high peak current when connected to TRIAC (leading edge) dimmers, because the
rise-time of the mains input is incredibly fast.|
This places enormous stress on the dimmer itself, and more importantly on the lamp's electronics. Despite makers' claims, the lamp will almost certainly not survive the abuse for very long, so the lamp life is reduced - possibly dramatically. A trailing edge (or universal) dimmer does not subject the lamp to a fast rising waveform, so doesn't cause excessively high peak current to be drawn.
Based on the above overview, it is important to understand that all standard 2-wire dimmers were designed to be used with incandescent lamps. Despite anything you will read elsewhere, operation will be unpredictable with ANY load that is not an incandescent lamp! For reliable performance with electronic loads (dimmable LED or CFL lamps), the dimmer should be 3-wire (active, neutral and load). Unfortunately, these are uncommon and will usually be difficult to install as a retro-fit because most lighting switch boxes don't have the neutral available. 2-wire dimmers were designed for incandescent (resistive) lamps, and were never intended for use with electronic loads.
A very small number of manufacturers are now supplying 3-wire dimmers. These (especially if trailing-edge) are a much better option for electronic lighting loads. However, many CFL and LED lamps are not designed to be dimmed by any means, so it's essential that only lamps/ luminaires specifically designed for use with phase-cut dimmers are used or the life of the lamp may be reduced considerably. An interesting fact is that even some 'non-dimmable' lamps can be dimmed successfully using a 3-wire trailing edge dimmer, but this isn't something you should rely on. Do not attempt to dim any non-dimmable lamp using a 2-wire or TRIAC dimmer.
ESP Has two designs for true 3-wire dimmers - see Project 157 - 3-Wire Trailing-Edge Dimmer and Project 159 - 3-Wire Leading-Edge Dimmer for the details. The leading-edge dimmer is not suitable for most LED or CFL lamps, but is designed to be used with 'conventional' (i.e. not 'electronic') transformers or other inductive loads.
It's important to understand that a traditional/ 'legacy' dimmer and lamp are simply wired in series, and the dimmer may well find itself with a neutral but no supply voltage. There's actually no difference whatsoever - 'no active' and 'no neutral' are functionally identical, because in both cases the dimmer has no reference. Whether the reference is the neutral or the active is neither here nor there - it's working with an AC supply and the polarity reverses 50 (or 60) times every second.
Having made the bold claims in the introduction, we need to understand why there is no neutral reference with electronic loads, and exactly how the situation comes about. It's actually quite simple for the most basic power supplies (aka 'ballasts') used for both CFL and LED lamps and/or fittings (luminaires). There are some minor complications when power supplies have an active power factor correction (PFC) circuit, but these may be less of a problem. There is still a vicious circle though - the lamp can't draw any power until the dimmer conducts, and the dimmer can't start to conduct if there's effectively no load.
Although you will see countless references to electronic power supplies presenting a 'capacitive load', this is nonsense. Yes, there is a capacitor, but it's isolated from the mains by a diode bridge and the capacitance is not reflected back through the diodes. The mains waveform is non-linear, not capacitive. There's a big difference between the two, and to claim that a standard non-linear load is 'capacitive' is wrong in all respects.
Because these non-linear loads are the most common - at least for the time being - we'll look at them first. This also helps because it's far easier to understand why they cause so much grief when used with dimmers. The so-called 'steady state' waveform is the only thing of interest, and the steady state conditions are set up very quickly, usually after no more than two cycles of the mains waveform.
Figure 1 - Non-Linear Load Schematic
Figure 1 shows a typical (but simplified) electronic load. The capacitor charges, and only loses a portion of the total charge during each half cycle of the mains. Because the diodes can't conduct until the mains voltage is greater than the voltage across the capacitor, most of the time no current flows and the power supply presents close to an open-circuit to the incoming mains. For the example shown (a 65W load), there is zero mains current until the instantaneous input voltage reaches 246V, and current is drawn until the voltage has passed the peak (325V) and fallen to 320V. After that, the circuit is open-circuit again.
Figure 1 - Non-Linear Load Voltage & Current Waveforms
Figure 2 shows the voltage and current waveforms. Current (shown in green) is drawn only when the incoming voltage is higher than the capacitor's charged voltage, and this only happens for about 2.8ms during each 10ms half cycle. For the remaining time (7.2ms) the load is open circuit - no current is drawn other than a tiny leakage current which is not enough to make a dimmer 'happy'. If you recall from Part 1, a normal 2-wire dimmer relies on the circuit through the load, because it has no other path to the neutral.
It doesn't matter if the dimmer is leading-edge or trailing edge. If it's a 2-wire type, the only time it can get a reference at all is during the short period that the electronic load is able to conduct. That means that instead of having a usable reference from the zero-crossing point of the mains waveform until the TRIAC triggers, there is nothing for the first 2.75ms, the reference is than available for 2.8ms, and then it's gone again.
Dimmers cannot be expected to work properly without a reference, so it's not at all surprising that they don't. For the sake of simplicity I have shown a leading-edge (TRIAC) phase cut dimmer below, not because it can be used with the load shown but because it's far easier to explain than a trailing-edge type. The circuitry for these is far more complex and difficult to understand, but they are really the only type of dimmer that can be used with an electronic load.
Figure 3 - TRIAC Based 2-Wire Phase Cut Dimmer Example
The 'lamp' is now the circuit shown in Figure 1, and it should be apparent that the dimmer can only 'see' the mains voltage via the lamp. If the lamp fails to provide a continuous path the dimmer must be compromised. With an electronic load such as that shown in Figure 1, the dimmer circuit only has a mains connection during the short period where the circuit is able to conduct. The remaining 7.2ms leaves the dimmer circuit floating, with no useful AC path. In reality there will be a low-current AC path due to the interference suppression capacitors that are nearly always used with switchmode electronic loads and TRIAC dimmers (C1 in the dimmer schematic). These create their own problems, and don't help the dimmer a great deal.
In order for the dimmer's timing circuit to function, it needs the supply to be present. This is provided automatically with an incandescent lamp because it has a resistive filament that always presents voltage to the dimmer - even when the dimmer is not conducting any power. This is exactly where it needs the supply voltage, because without it the timing circuit (VR1, C2, R1 and C3) can't function at all. A 60W (230V) incandescent lamp has a nominal filament resistance of 880 ohms, and that resistance can provide ample current for the dimmer's timer to function normally. When an electronic load is used, this essential connection is either lost or greatly compromised. Trailing edge dimmers are no better off, because they also need the supply as a reference. These 'legacy' dimmers were designed to operate with resistive (incandescent) loads, not with any form of electronic load.
There is an underlying problem with all TRIAC dimmers that may not be immediately apparent. The DIAC (DB1) is a bidirectional breakdown diode - a switching device that shows a negative impedance characteristic. When it discharges C3 into the gate of the TRIAC, that's the one and only chance the TRIAC gets to turn on during that half-cycle. This can't happen at a time when the load isn't drawing current, because there's not a complete circuit, but the TRIAC can only turn on at or near the peak of the AC waveform with most electronic loads of the type shown above. A TRIAC will turn off when the current through it falls below the 'holding current' - typically 30-50mA - or if the polarity reverses. Turn-off is very fast, and can cause ringing in filter inductors for EMI suppression.
TRIAC based dimmers (which are all leading-edge) have an issue with the holding current of the TRIAC. Once the current drops below the minimum for the TRIAC, it switches off. This is a problem for LED based lamps in particular, because they draw comparatively little current, so a point is reached in the dimming cycle where the load doesn't draw enough current for the TRIAC to remain on. It may turn off almost immediately after being triggered, and the lamp will either flicker or go out altogether. This performance is unexpected, and is seen by the consumer as a fault. Again, the lamp will get the blame - not the dimmer ... "The dimmer works just fine with an incandescent lamp, therefore this LED (or CFL) lamp is faulty!"
The DIAC + TRIAC circuit creates another insidious problem too. Most DIAC/ TRIAC combinations are slightly asymmetrical, and at very low settings you might only get half-wave operation. If this happens, the lamp may flicker at low settings making the combination unusable. The flicker is likely to be very obvious and is very disconcerting. Trailing-edge dimmers don't use a DIAC so this is less likely to be a problem, especially if the dimmer is a 3-wire type.
Half-wave operation is one of the more serious sides of using conventional (legacy) 2-wire dimmers with electronic loads. Because there's no neutral and therefore no reference signal for the dimmer's electronics, the circuitry can become 'confused'. One of the effects of this confusion is flicker, which can be an indication that there's something more sinister happening.
I have seen dimmers of several different makes and models with various LED lighting loads, and I always monitor the current waveform when I'm running any tests. It's not at all uncommon to get the dimmer into such a state where it only triggers on one polarity - half-wave operation. The result is an effective DC component in the mains waveform, which is particularly bad for conventional iron-cored transformers and motors that are connected to the affected mains circuit. Even remote equipment on a different mains branch circuit can be affected, and the supply authorities everywhere frown on any equipment that creates a DC component because it causes transformer overheating and if bad enough, failure.
While it's unlikely that a few LED lamps will cause general mayhem on the mains, it's definitely not a good thing to do, and you will nearly always see flicker from the lamps when it happens. Three-wire dimmers by their very nature are far less likely to suffer from half-wave operation, and with IC fabrication techniques at the point where they are now, it would be easy to include half-wave detection to create a dimmer that would then be able to correct the problem if it ever happened.
It's almost impossible to include any circuitry to a 2-wire dimmer to prevent half-wave faults. The reason is simple - without a neutral connection, there's no permanent and reliable reference. As noted already, the reference is essential so the dimmer can function normally, and by far the easiest way to ensure that it's always present is to use only 3-wire dimmers.
The simple answer (for the majority of manufacturers of lamps and dimmers) is, unfortunately "Bugger All". Yes, there are steps being taken, but most involve changes within the lamp circuitry that attempt to 'trick' simple 2-wire dimmers to make them work properly. Mostly, these attempts have been inadequate, but there are a few that work reasonably well if the user isn't too fussy.
Many lighting manufacturers and power supply IC makers have recognised that there are endless problems created by legacy dimmers [ 1 ]. As a result, there are new approaches to the problems that reduce the interactions between lamp and dimmer, in the hope of true compatibility. These additional circuits usually work well with some dimmers, but less well (if at all) with others. Mostly this has resulted in more marketplace confusion because some dimmers work with some lamps, and other combinations either don't work or are erratic.
There are several ways that have been used to provide some level of compatibility, including the use of 'bleeder' circuits that are intended to give the dimmer a reference in the hope that it will work normally. Another technique is to include a 'damper' circuit that limits the peak current so TRIAC dimmers don't create very high current spikes, and damps parasitic oscillation that causes erratic behaviour. The disadvantage of these techniques is that they all add to the overall heat load of the lamp's power supply and reduce efficiency because there are inevitable losses.
For any dimmer to function normally, it needs its reference. A 2-wire circuit can only 'see' the required reference through the load, and most electronic loads are effectively close to being open-circuit unless the incoming mains voltage is high enough to cause conduction. Dimmers need to know when the mains voltage crosses zero volts because that resets their internal timer(s). If the load doesn't conduct at all until the instantaneous voltage across it is perhaps 90-200V or so, the dimmer will malfunction.
It's not reasonable to expect lamp makers to incorporate perhaps quite a bit of additional circuitry in the hope that the lamp will then work with some of the most common dimmers. Users have to understand that they are dealing with an entirely new form of lighting, and stop expecting it to work just like the lamps they had before. Despite this, a lot of the lamp makers are trying very hard to make power supplies that will work with legacy dimmers. It would be a lot simpler all round if 3-wire dimmers were more readily available.
Figure 4 - 3-Wire MOSFET Based Phase Cut Dimmer
A conceptual trailing edge (or leading edge with additional electronics) 3-wire dimmer is shown in Figure 4. The power supply, zero crossing detector and timing circuit are wired between the active and neutral so it no longer relies on the load to provide a reference. This means that it will operate normally regardless of the nature of the connected load. The MOSFETs will turn on when commanded, regardless of the current drawn by the load. Please note that Figure 4 is purely conceptual - I do not have a complete circuit for a 3-wire MOSFET based dimmer.
With modern electronics it's fairly straightforward to make a 3-wire trailing edge dimmer that will address all the problems. Being 3-wire, there is no longer any issue with the dimmer not having a reference, and trailing-edge eliminates the high current pulses that can damage the dimmer or electronic load. Such an arrangement will dim any dimmable lamp, including incandescent, CFL and LED. As with all trailing-edge dimmers, such an arrangement is intended for lamp loads only - trailing-edge dimmers are not suitable and must never be used for motor loads or iron-cored transformers!
Figure 5 - 3-Wire TRIAC Based Phase Cut Dimmer
Figure 5 shows a TRIAC based 3-wire dimmer. Again, it is conceptual and has NOT been built or tested, and I discourage readers from attempting to build it. Since the use of TRIAC dimmers is unwise for electronic loads such as the power supplies for LED lamps, there's no good reason to build one anyway.
With leading edge dimmers, if the load (lamp with electronic power supply) has an EMI suppression capacitor wired directly across the mains terminals or has a capacitor input filter as shown in Figure 1, the current spikes into the load are limited only by the total series resistance within the circuits and can easily exceed 10A for a few microseconds, even with a low powered load. A trailing edge dimmer has no such problem, because the waveform increases slowly (but is turned off quickly).
In general, TRIAC based dimmers are a poor choice for electronic loads. There's usually only one chance to trigger the TRIAC, and if the external load can't draw enough current (for the TRIAC to conduct) at the moment the breakdown diode (DB1) sends a trigger pulse, the TRIAC may not turn on at all for that half-cycle. This will cause problems.
The part that I find puzzling is that so few manufacturers are looking towards fixing the dimmers. Most of the efforts have been focused on trying to make electronic loads work with 2-wire dimmers, and there is a dearth of ICs designed to make a decent 3-wire dimmer. There are a few, but given the scale of the problem I would have expected more effort to be concentrated where it can really help. A simple IC that can drive a pair of power MOSFETs in a straightforward 3-wire trailing-edge dimmer would be a real boon, but no-one seems to be interested. One of the few is the Atmel U2102B (a schematic is shown in Part 1), but I don't know of any manufacturer who is making dimmers based on that IC.
The industry consensus is that LED lighting must work with legacy dimming technologies [ 4 ], but this means that every lamp or luminaire has to have additional circuitry that won't work with all dimmers anyway. It's a far better solution to provide a new class of dimmer that is designed to work with non-linear electronic loads. This approach still has problems though, because the switch box where the dimmer is installed usually won't have the neutral conductor available. This means that consumers will have to have some rewiring done, something that most will avoid if at all possible.
It should be mandatory for all new buildings that every switch box or wall-plate must have a neutral wire available so that when sensible 3-wire dimmers do become readily available they can be retro-fitted with the minimum of expense. Unfortunately, I fear that this probably won't happen in a timely manner. In Australia, some enlightened electricians do provide a neutral to lighting switch boxes, but these may be in the minority.
The problems with 2-wire dimmers have been recognised in some areas. In the US For example, it is now (apparently) a requirement that a neutral be made available in lighting switch boxes, but the section of the wiring code (NEC (National Electrical Code) 2011 404.2(C)) allows for exceptions that are subject to interpretation in some cases. I'm not in the US and don't know the code, so I'm relying on the little I can glean from the Net. Other countries may have similar provisions, but I don't have any details [ 5 ].
Another option that's being explored by some lighting manufacturers is to use digital codes transmitted along the mains wiring [ 6 ]. This eliminates the direct issue of dimmer incompatibilities with the lamps/ luminaires, but it's almost guaranteed that a neutral wire will be needed to power the transmitter circuit. Because this is a new method, it's inevitable that the dimmer codes from one company will be incompatible with the power supplies, lamps and luminaires from others, so installers will be faced with having to use power supplies and dimmers from the same supplier. If the style of lamp or luminaire you want to use in a location is not compatible with the controlled driver, then you're out of luck. You may also be faced with having to purchase lighting from a supplier whose pricing may not be sensible from the customers' perspective.
It's potentially a good solution, but requires standardisation before it's actually useful for installations. Because the protocols will almost certainly be proprietary this could lead to a situation that's even worse than we have at present. It's probable that any existing methods are covered by patents, and other suppliers will be unable to use the protocols developed so far without licensing agreements which may be quite costly.
There are already several different control systems in use. Australia uses mainly C-Bus (which was developed here), and most other countries use either DALI or 0-10V. The latter is an old analogue system that's particularly easy to use and can be implemented very cheaply. These control systems are not compatible, although some 'protocol converters' can be found. Use of any of the major automation systems makes an installation very complex compared to a 'normal' lighting system, and most require specialised installers and maintenance personnel.
Wireless (either as part of the IoT, via smartphone or stand-alone) is making inroads. Personally I find the idea of controlling lighting from a smartphone or similar to be (at best) slightly ludicrous, but you can expect to see more lighting systems being controlled this way. Security is important - you don't want random strangers to be able to turn your lights on and off at will. This isn't an area that I've investigated in any depth, but a web search will find that the IoT is touted as the greatest thing since sliced bread. Whether or not the IoT ever reaches critical mass remains to be seen.
Dimming is challenging, and until 3-wire dimmers become the standard there will be continued problems. The idea that the consumer can simply change from incandescent to CFL or LED lighting with no other changes is not working. This is despite the efforts of many lighting, power supply and IC manufacturers to provide work-around solutions so that users can continue to use legacy dimmers that are clearly unsuited to lighting that relies on electronic power supplies.
There's no doubt that some of the latest lighting products often work surprisingly well, but this may only be the case with a limited number of dimmers. If the consumer happens to have dimmers that don't work properly with their new LED lamps it's usually the maker of the lamp who's held to blame. The real problem is not the lamp, it's the dimmer!
There are certainly ways that can be employed that can make lamps perform better than they do at present. However, this can probably be done only within luminaires where there's usually plenty of room for extra circuitry. Retro-fit lamps are more difficult because of limited space and very limited thermal capacity, which means that any additional heat is far harder to disperse into the surrounding air.
Traditional 2-wire (legacy) dimmers can often be made to function very well with dimmable CFL or LED lamps by including one incandescent lamp in the circuit. Naturally this can only work if there are two or more light sockets being controlled by the same dimmer. The incandescent lamp is enough to allow the dimmer to work normally, because it always conducts so the dimmer circuitry is never without its essential reference. While this definitely works (proven in workshop tests and in my lounge room) it's not always convenient or even possible.
The solution is inescapable - 3-wire dimmers must become the standard, and 2-wire dimmers should be phased out of production. It's quite obvious that they are unsuited for use with any lamp that has an electronic power supply, and the continued attempts to try to make new lamps compatible with legacy 2-wire dimmers are clearly not working very well.
|Copyright Notice. This material, including but not limited to all text and diagrams, is the intellectual property of Rod Elliott, and is Copyright © 2015. Reproduction or re-publication by any means whatsoever, whether electronic, mechanical or electro-mechanical, is strictly prohibited under International Copyright laws. The author / editor (Rod Elliott) grants the reader the right to use this information for personal use only, and further allows that one (1) copy may be made for reference. Commercial use in whole or in part is prohibited without express written authorisation from Rod Elliott.|