Insensitivity to vibration, long lifetime, high-energy efficiency and the possibility for superior control of the light source are key factors for automotive LED applications. Compared to incandescent light bulbs, LEDs are insensitive to mechanical vibration but need driver circuitry. Universally, automotive electrical supply systems are based on a lead-acid battery, which is charged by an alternator/regulator that is mechanically driven by the engine. Such a system suits older incandescent light bulbs, but not LEDs. A well-regulated current supply is necessary for optimal LED performance.
To drive an LED properly requires control of the current that is independent of the supply voltage. The light output is dependent mostly on the current, not the voltage. Theoretically, every electron transforms into a photon and a fixed percentage of the photons escape the LED chip as visible light.
For low-quality solutions, a resistor is enough to limit the current if the supply voltage is constant. It is worth noting that LEDs are somewhat self-regulating when used in a simple series resistor configuration. If the temperature rises, LEDs lose efficiency and light output, simultaneously diminishing the forward voltage. This lower forward voltage results in an increased current and slightly compensates for the reduced light output due to the higher temperature. As long as the battery voltage is constant, a series resistor performs adequately for computer and instrumentation applications. But automotive industry requirements mandate accounting for battery variations between 8 V and 18 V and up to 80 V peaks. In addition, high-intensity LEDs generat substantial amounts of heat from the resistor. This makes thermal designs more difficult.
A better, but not optimal alternative is to use a dc-dc voltage converter to create a suitable stable voltage and then combine this with a resistor. This works if you already have a dc-dc converter for supplying a computer or other electronics, and is probably the most widely used method of driving LEDs.
However, a better solution for driving LEDs is through current regulator circuits that work independent of the supply voltage. Two fundamental types of current regulators include energy wasting or energy converting.
An example of an energy wasting regulator is a linear voltage drop regulator, which burns away the energy the voltage difference represents for a given current. Alternatively, an energy-converting regulator tries to save the energy difference between the different voltage levels.
The equation describing this energy conservation is one of the fundamental laws of thermodynamics:
Power in [W] = Power out [W].
Substituting W = V • I gives:
Vin • Iin = Vout • Iout + (100 - X%efficiency)Wheat
If you put the forward voltage of the LED as Vout and the desired current as Iout you have the general equation that describes your LED driver.
ENERGY-WASTING LED DRIVERS
It is relatively simple to create a linear regulator with discrete components. Figure 1 describes a regulator of discrete components. D1 should be a zener diode or a voltage reference. The equation: ILED = VD1/RSET defines the current. D2 gives simple temperature compensation in relation to the transistor base diode.
Although this is a simple circuit, as with all energy-wasting LED drivers, there remains the problem of energy loss and dissipating the heat generated by the resistor. This heat dissipation will be more severe with increased LED brightness. The brighter the LED is driven, the more energy is wasted.
For lower currents and especially if the sum of the LED forward voltages in series are slightly below the supply voltage, this type of regulation can be suitable.
Several manufacturers of LED driver ICs use this type of current regulation. However, this is not recommended method for driving high-intensity LEDs.
ENERGY-CONVERTING LED DRIVERS
In most cases, a switching regulator provides a better electrical solution. Switching regulators switch a series load on and off, hence their name. During one period an RLC (tank) circuit is charged. In the following period this energy is used to drive the load or add to a voltage level in a storage source driving the load. This reorganization of the energy typically yields better than 80% efficiency and in most cases better than 90% is achievable. So switching regulators can be used to increase voltage, decrease voltage and even invert the voltage. These capabilities are not possible with linear regulators.
The simplest switch regulator is the buck regulator as shown in Figure 2.
Operation: The voltage difference between input voltage and the LED voltage charges the coil L. When energy is stored up in the coil, voltage over it will sink, and the current increase. When the current reaches a defined level, the control circuit will switch off the transistor in series. Then during the defined off time, the coil discharges some of its energy as a current through the LED. The result is an alternating current over the LED. The switch regulator circuit controls the peak value of the current. This value might be set by programming the regulator IC or by external components. The current is also defined by the selection of a sense resistor on the drain side of the NFET switch.
The buck regulator current over the LED is continuous but alternating. The current consumption to the complete circuit is in contrast discontinuous (Figure 2). This can cause problems on the supply side and easily creates noise through the supply line.
If the supply voltage is lower than the sum of forward voltages for all LEDs in series, a boost regulator is the solution. A boost regulator is more complicated as it also needs to control the boosted voltage in addition to controlling current.
This type of boost regulator cannot handle the case when the supply voltage is higher than the sum of the forward voltages of all the LEDs in series. The current will then rush uncontrolled as seen in Figure 3.
This type of LED driver also creates a pulsating output current to the LED. This is difficult to filter due to the relatively large current in LED applications. In principle, a simple boost regulator creates relatively high noise levels on the output to the LEDs. This makes it necessary to use relatively short wires and PC runs to the LEDs.
A SEPIC regulator is a single-ended primary inductance converter. This regulator can work as a boost regulator and as a buck regulator. One weakness, though, is the capacitance between the coils. That capacitor must handle all the energy that is converted to suitable current and voltage for the LED.
This type of regulator is useful when overvoltage can exist on the supply line and you primarily need a boost regulator.
The most stable and safest solution for a good boost LED driver is a combination of a boost regulator with a buck regulator by connecting them in cascade. This minimizes optimization issues. One boost regulator can also preferably deliver the voltage for several buck regulators in parallel.
This is also a good solution with respect to noise. It combines good behavior on the supply from a boost regulator with the similar good current output from the buck regulator.
DRIVING LEDS FROM HIGH-SUPPLY VOLTAGES
With a very high supply voltage and a forward voltage drop (Vf) over the LED 10-20 times lower, a problem arises because of a very fast charge time for the coil in series with the LED. Fast charge (and discharge) events result in poor efficiency.
The charge and discharge cycle runs according to the Figure 4 and it is simple to see that the rise time is equal to 10 to 20 times higher frequency than the base frequency for the regulation.
A good solution for high efficiency and low levels of emitted electrical noise is to select a switch frequency that gives a rise time equal to the frequency of the coil specification. In the case of 10 to 20 times voltage difference, you should choose a switching frequency that is 10 to 20 times lower than the coil's optimal efficiency frequency.
Howver, when the the supply voltage is double the sum of the LEDs forward voltages in series, it results in an optimal solution. The regulated current then appears as shown in Figure 5, where the waveform has reasonable symmetry.
It is also possible to use the coil as a transformer with the advantage of having independence with respect to the supply voltage, and high isolation between supply voltages and LED current. This can replace both boost and buck solutions, but is less efficient. A strong magnetic coupling between primary and secondary winding of inductor will improve efficiency. This has the advantage that any one of the wires to the LED can be shortened to the ground or the supply without any hazardous current.
ELECTRICAL NOISE FROM SWITCHING REGULATORS
All switching regulators generate electrical noise. Common dc-dc voltage regulators working with a voltage-level control often can deliver a well-filtered supply voltage. This is achieved by a large capacitor on the output and gain improvement by high switching frequency. LED regulators should use current regulation, not voltage regulation.
The earlier mentioned buck regulator is a simple and cost efficient current regulator, but poorly dimensioned it can generate strong electrical noise in LED applications. Circuit board layout and cables are critical to controlling the noise levels.
Rules of thumb for low noise:
- Low switching frequency.
- Short cables and a small current loop to the LEDs.
- Noise filter if longer cables needed on the LED.
- Fast feedback diode.
- Switching transistor on the center of the circuit board.
- Careful with cable and noise filter on the supply line.
In addition to these rules of thumb Melexis has implemented schemes to help control driver IC noise. In the MLX10801 and MLX10803 LED drivers, a pseudo-random generator on the switching frequency minimizes electrical noise.
In the automotive electrical environment there are several tests and test setups to rate the relative noise performance of electronic modules. One commonly used standard and test setup is defined by an International Electrotechnical Committee (IEC) subcommittee, the CISPR (an acronym from the French name for the committee known in English as the International Special Committee on Radio Interference).
Figure 6 is a sample schematic of a low noise application that can meet CISPR 25 level 5. The Coil L1 must be dimensioned in relation to the switching frequency and the LED current. To do this, we use software programs and Excel sheets available at the Melexis web page, www.melexis.com. The switching frequency should stay below 150 kHz to avoid the lowest frequency band regulated by CISPR 25. The program also provides values for ROSC, RSET and RSENSE. L2 is part of a filter that is needed to pass the highest level of noise suppression. For CISPR25 level 1-3 the coil L2 can be omitted.
When using this schematic in Figure 6, a good starting point is to set L1 and L2 to 100 µH for a typical LED current of 0.5 -1 ampere. The noise is broad banded so the capacitors need to handle low and high frequencies. This is why we employ the double set of capacitors on both sides of the filter coil L2. The main source of high-frequency noise is the feedback diode D1. Select this diode carefully, and try it in the application. For input voltages below 100 V, Schottky diodes are the best choice.
Melexis web site forum “Knowledge Base” provides more information and support to design your optimal solution.
TEMPERATURE-COMPENSATION OF LED
Red and amber LEDs of gallium arsenide (GaAs) and gallium arsenide phosphide (GaAsP) types have strong variations in light output depending on the junction temperature. Typically, an LED with a 100% light output at 25 °C will have only 40% output at 80 °C. This variation of light output easily can be compensated. In the modified schematic of a low noise design, only a PTC and a NTC resistors have been added (Figure 7).
As can be seen in Figure 8, using temperature-compensating PTC and NTC resistors, the relative light output at 80 °C is the maximum. If the junction temperature is lower than 80 °C for any reason, the PTC resistor will have a corresponding lower value and decrease the current proportionally. It is then necessary to balance the PTC coefficient to the light output of the LED.
For protection, if the temperature goes above 80 °C an NTC resistor on the other reference input of the MLX10803 will lower the current in a similar way.
STOP, TURN AND TAIL LAMP LED APPLICATION
Nearly all cars today employ red GaAs LED taillights. Consequently, most LED taillights shine too brightly in cold and dark surroundings and too weakly in hot and bright surroundings. The legal standards controlling automotive lights were established many years ago based on incandescent light bulbs. Incandescent light bulbs operate in a temperature range of several thousand degrees temperature of the heated filament light source. So the effect of an ambient temperature difference of 60 degrees (between 20 degrees and 80 degrees) on light output is hardly noticeable. Today, in cold weather it is easy to see the difference between taillights with LEDs vs. incandescent bulbs. It is questionable whether the LEDs are too bright in cold weather. A temperature-compensation scheme as previously described will give the car a more professional, sophisticated appearance and look much better when used in combination with light bulb applications. This situation may also become a subject for the standard-setting bodies to address, establishing reference specifications for lights at different operating temperatures.
In applications where the stop and taillight function is combined, pulse width modulation (PWM) is used to create two levels of brightness; one for tail lamps and one for brake lamps. This is done because LED manufacturers have typically tested and sorted their LEDs at only one defined current, i.e., the current drive level needed for the brake light level. In the past, there has been no testing or sorting for the lower light output level suitable for taillight levels. Fortunately, this is changing and LUMILEDs now offer LEDs tested at two current levels.
This is good because with LEDs matched at two current levels, the driver can drive two specific current levels instead, without PWM. Then PWM not needed. The light emitted from these LED PWM tail lights with a PWM ration from 1:10 to 1:20 and frequencies from 80 Hz to 100 Hz, can seem unpleasant and harsh to view compared to traditional incandescent lamps. This is due to the human eye's sensitivity to red light and these frequencies. It is especially acute with cold temperatures and poor current compensation from the cold temperatures.
The drive toward more sophisticated automotive LED lighting applications has resulted in the development of several targeted ICs and associated applications circuits. Many of the challenges have been addressed in the preceding examples. Many more details beyond the scope of the space here are required to successfully deliver fully specified lighting modules. This general review should forge a better understanding of the engineering trade-offs to achieving reliable, pleasing lighting designs with current high bright LEDs.
ABOUT THE AUTHOR
Roger Alm is a product marketing manager at Melexis. He started with Melexis in 2004 as a product marketing engineer in the Business Unit Door Systems and ECUs. Alm holds a masters degree of science in applied physics and electrical engineering from the Linköpings University in Sweden.