A simple and efficient linear dimmer
12V LED strips, over the past few years have become more than just decorative lighting. Although, rather annoyingly, there are still some very poor quality devices still being sold with false specifications, as of 2015, there are strips such as Figure 1 available that are able to provide 2400 lumens per meter (equivalent to 3 60W standard incandescent lamps), attaining a luminous efficacy of over 166 lumens per Watt of input power, while costing £6.60 per meter. These high performance strips can be used for main workshop lighting, and if used correctly, interior lighting as well. Their ease of installation for situations such as the former, where aesthetics are not the main priority, high efficiency despite the internal current limiting resistors (reducing this by up to 30%), and low cost make them a clear winner in the opinion of the author.
Figure 1. A Section of 5063 LED Strip
One drawback, however, is that these devices require an external 12V supply to work correctly. This isn't too much of an inconvenience as there are many high quality switching power supplies out there which will provide a very reliable 12V at fairly high current for a surprisingly low cost, while consuming very little standby power. If you already have installed such a supply in your workshop or home for other purposes, then you're already half way there before you have even got hold of the strip itself. A very useful benefit of their 12V operation is that you can safely do the wiring yourself, without having to worry about the risk of electrocution and other safety concerns involved with mains wiring. The power supply can also be situated away from the LEDs themselves, in a well ventilated area, whereas all-in-one LED lamps designed to replace incandescent fittings that work straight off the mains will have to have their power supply physically close to the LEDs themselves, which produce significant heat. This rise in temperature can quite dramatically shorten the lifespan of the power supply, making LED strips more reliable, long term.
Another advantage of their 12V operation is that there is no mains inrush current when the strip is switched on. Devices with excessive inrush current can prove rather troublesome, especially if there is any significant series resistance in the mains supply, causing RF noise and a momentary drop in mains voltage that can cause problems with some electronic appliances. A good example of this would be audio equipment where a muting circuit will activate, sensing a falling supply voltage, and momentarily disconnect the outputs, much to the irritation of the user.
Such strips could be switched on and off by simply using a switch to connect and disconnect the strip from the 12V supply, but dimmer switches are a much more useful method of control, especially when the strip is used in an situation where different levels of light are needed, such as a living room with a television or other visual entertainment system, or simply for ambient mood lighting.
Unfortunately, many of the dimmers commercially available for 12V strips leave quite a bit to be desired. They often operate by simply switching the strip on and off fast enough so that the persistence of vision gives the illusion of a constant level of light at different levels depending on how long the strip is switched on in comparison to how long it is off for. This can create alarmingly high levels of radio frequency noise as well as some rather unpleasant strobing effects when moving objects are viewed under the light. The vast majority also work only by remote control, which is all very convenient, but as soon as the remote control is lost, broken, or the batteries run down, the entire system becomes useless. A serious disadvantage of this means of control is that if the system is to be used as the main lighting for a single room, then the remote control must be found in the dark every time the lighting needs to be switched on.
What is needed is a dimmer that can be operated from the wall, has a linear voltage output that reaches right up to the top of the 12V supply with minimal loss, and can go from full brightness to complete darkness via a simple control such as a potentiometer. Space is also at a premium, so a simple topology using low cost and readily available components is highly desirable. Luckily, a solution that meets all of the above requirements is fairly straight-forward to achieve with a bit of know-how and common sense.
In order to reach a full 12V output from 12V supply rails, it is important that the voltage dropping element used to regulate the output voltage has a very low on resistance and forward voltage drop. A P channel MOSFET fits this criteria perfectly, and there are many low cost devices that can source very impressive amounts of current while featuring a tiny on resistance. After a brief search I managed to find a very suitable component, the IRF4905, capable of sourcing 74A of drain current while maintaining an on resistance of 20mΩ, meaning that it could supply a current of 10A while only dropping 200mV, which is pretty negligible when it comes to powering LED strips. They are available from RS for roughly £2 a piece as of 2016.
It is also important to consider, not only the maximum on voltage, but how low a voltage is required across an LED strip to turn it fully off. For example, a standard warm or cool white LED strip is comprised of three LEDs in series, each with a forward voltage in the 3-3.5V range and a current limiting resistor. As each of these LEDs requires about 2.5 to 3V across each one before any current flows and light is emitted, it is seemingly reasonable to assume that a voltage of 7.5V or less will not result in any luminous output or current draw. However, occasionally the odd LED on a strip will have small defects that manifest themselves as a parallel resistance of anything from 100kΩ to 100Ω, depending on the severity of the fault, across the junction which may result in the other LEDs connected in series with it lighting at a lower voltage than the combined threshold voltage of all the LEDs in series. For this reason, it is sensible to allow for one of the three LEDs in the series string to have such a fault, and assume an off voltage of 5-6V. There is of course the possibility that two of the three LEDs are defective and have a parallel resistance, but with all high quality strips the likely-hood of this is very low and can be ignored. It is the strong opinion of the author that any strips that have two out of three faulty LEDs at any point are of unacceptably low quality and should be banished to the dustbin, being unworthy of a high quality dimmer circuit.
Figure 2. Circuit Diagram
At its heart, the circuit is a fairly conventional variable low dropout voltage regulator, based around the readily available TL431 precision reference. Using this device instead of just an NPN transistor allows the reference voltage to be made much more accurate while also far more temperature stable. As the reference will be located in fairly close proximity to a source of considerable heat (the series dropping MOSFET), an accurate and temperature stable reference must be used, particularly when the reference will be determining, in conjunction with R5 and R6, the maximum output voltage.
An external linear potentiometer of 10kΩ connected between the 'Control Pot' terminal and ground will provide a means of controlling the dimmer. This can be placed a considerable distance away from the rest of the dimmer circuitry and mounted on a suitable wall plate. Usefully, its placement at the bottom of the voltage feedback network means that a greater range of adjustment is available at lower output voltages, where the LED strip is more sensitive in terms of light output versus voltage, allowing for a smooth and fairly linear control brightness throughout the full range of control.
Regrettably, even the best potentiometers tend to have rather wild tolerances of up to ±20%, so it is necessary that the absolute value of the potentiometer does not affect the maximum on voltage, for this reason the circuit is configured so that the maximum output voltage is reached when the potentiometer is at its lowest resistance, effectively a short circuit. Another limitation of resistive track based potentiometers is that as the device ages with use, the carbon track wears down and the absolute resistance increases. Luckily this topology has some built in immunity to this effect, with the minimum output voltage decreasing as the absolute resistance of the potentiometer increases, ensuring that the LED strip will stay at the potentiometer's minimum setting throughout its life. Like-wise, should the potentiometer's wiper somehow come out of contact with the resistive track, the most common failure mode, then the LED strip will stay off, not being stuck on all of the time.
Capacitors C1 and C3 provide power supply and output decoupling, while C2 affords frequency compensation to ensure stability. Adding any capacitance of more than about 1μF across the output is not recommended as this may cause the circuit to become unstable. If a significantly capacitive load is used (the author cannot think of such a situation in practice), the it would be wise to increase the value of C2 by experimentation. C4 provides AC decoupling across the control input to suppress any noise that may be picked up by the cabling between the main circuit board and the control potentiometer. C4 also contribute to a smooth and pleasant control latency similar to a traditional triac dimmer used with incandescent lamps.
Despite its linear operation, the dimmer is surprisingly efficient, mainly due the low voltage drop across the series pass element (Q2) throughout the range of its operation. For example, when driving 50 Watts of warm white LED strip the highest dissipation at any level is only just above 2 Watts. Not bad at all! Because the LEDs drop roughly 10V of the working voltage during normal use, the internal current limiting resistors in the strip along with the regulating MOSFET are only dropping the remaining 2V. The MOSFET's peak dissipation occurs when the current drawn by the strip is about half of what it would be with the full 12V across it. When the control is set to fully off, with no current flowing through LED strip, the current drawn by the dimmer is 2.7mA at most, which equates to 32mW of idle power or £0.03 worth of power per year in the UK as of 2016.
The supply voltage range from 12V up to about 20V, allowing the circuit to work quite happily off a car battery or any other source with a voltage in this range. However, any voltage above 12V will be wasted as excess dissipation across Q2 and will limit the strip power, requiring a larger heat sink to be used. With a 12V supply the maximum strip power that the author can recommend would be in the region of 250W or so. Although the MOSFET may just about be able to handle up to 500W or so with a big enough heat sink, the author considers such high strip powers to be beyond the realms of what is practical with a linear dimmer.
The main question that the intrepid constructor will be faced with when building this circuit will most probably be, 'How big a heat sink do I need?'. To answer this question and to spare any unnecessary grappling with algebra and data-sheets, I have come up with a couple of quick and easy equations for working out the required thermal resistance of the heat sink (θ) based upon either the maximum strip power or current, while accommodating a healthy safety margin.
Figure 3. Power based equation, where θ is the thermal resistance of the heat sink and P is the maximum power in Watts drawn by the the LED strip.
Figure 4. Current based equation, where θ is the thermal resistance of the heat sink and I is the maximum current in Amps drawn by the the LED strip.
The above equations take the thermal resistance of the junction to case into account, as well as an electrically insulating silicone pad to give a worst-case junction to ambient temperature of no more than 30℃. This may seem very conservative, given that the IRF4905 MOSFET is rated for a maximum junction temperature of 175℃. The author considers that the risk of failure really isn't worth the potential saving given that the circuit board may be located in a hard to reach, and possibly poorly ventilated location, such as under floor-boards, while the heat sink itself will be comparatively small and low cost part.
All resistors should be 250mW types, with R5 and R6 having a tolerance no greater than ±2% to avoid the resulting maximum output voltage from becoming too high or low, which could damage the LED strip in the former case. If you plan on constructing this circuit using strip-board, then remember to solder over the ground track along with the supply rail track and output track to reduce the track resistance so as to allow for the high current. It is also important to keep electrolytic capacitor C4 a reasonable distance (a few centimeters or so will do) away from the heat sink, as increased temperature will substantially reduce its life.
Wire the potentiometer as a variable resistor so that minimum resistance, and therefore maximum brightness, is obtained with the control fully clockwise. For best results connect the wiper to ground along with the other side of the resistive track not in use. This allows for smoother control should the track become dirty or develop significant wear.
So that's pretty much it. It's a fairly straightforward circuit to build, but if this piece generates significant interest then I will consider making PCBs available to buy. As far as installation goes, it's entirely up to the constructor, though as the circuit is very reliable and the potentiometer can be located on the other side of the house if needed, the circuit itself can be located under floorboards or even integrated into a wall control panel without the fear of upheaval should any problems occur. Mine's currently stuck to the ceiling of a fitted wardrobe with self adhesive spacers.
If you enjoyed reading this piece, have any constructive comments or have built this circuit, then please feel free to drop me an e-mail. I always look forward to intelligent discussions!
Michael Fearnley 2016