We now find high-brightness light-emitting diodes (HBLEDs) in a growing array of applications, including automotive displays and exterior lights, backlighting for televisions and video monitors, streetlights, outdoor signs, and interior lighting. Characterizing these devices demands a somewhat different set of test tools and techniques than do ordinary LED devices. In this article, we will examine some of these tools and techniques.
High-brightness LEDs vs. high-power LED modules
What are HBLEDs exactly, and how do they differ from the LEDs with which we are all familiar? Essentially, HBLEDs are LEDs that operate at 1W of power or higher, and usually in the range of from 1 to 3 W. The current draw of typical LEDs ranges from 10 to 30 mA, but HBLEDs generally draw from 300 mA to 1 A. Typical LEDs are packaged in a small 3-mm or 5-mm plastic dome with a pair of leads. In many instances, manufacturers mount HBLEDs on a small, thermally conductive board designed to draw heat away from the device’s junction. The leads themselves must be much thicker to handle the higher levels of current associated with HBLEDs.
As bright as individual HBLEDs are, they are still not bright enough for many lighting applications. Often, multiple HBLEDs combine to create luminaires, such as in an LED light bulb for a retrofit application or a complete lighting fixture. This is a good solution for applications in which it is desirable to have the light fan out in multiple directions and where the luminaire’s size provides sufficient space for multiple devices. However, in applications where space is very limited and/or the light must be directional, this approach is unworkable.
In contrast with HBLEDs, high-power LED modules (HPLEDs) offer far more compact form factors and can deliver lots of directional light. These characteristics have made HPLEDs extremely popular in applications such as automotive headlamps, projectors, and even some high-power flashlights.
HPLED modules consist of one or more large-die LEDs assembled in a single package and can handle higher currents than typical HBLEDs. It is common for a single die to be required to withstand currents of 10A or higher. These high currents produce a significant amount of heat, which must be dissipated and/or monitored during the characterization process. Modules are typically built on a heavy copper-core board or other thermally conductive material in order to pull the heat out of the LED junctions. It is also simple to mate these heavy boards with a heat sink of appropriate size to keep the LEDs cool. Modules usually have a thermistor built into the board to simplify monitoring the junction temperature of the LEDs. Despite manufacturers’ efforts to dissipate the heat produced by the LEDs, operating junction temperatures for these modules commonly go as high as 140ºC.
HPLED module testing requirements
The growing popularity of HPLED modules is creating new requirements for equipment to test them accurately:
Greater power — Today’s HPLED modules are capable of dc power levels close to 100 W; tomorrow’s modules may require even more. Testing these modules properly at their rated operating points requires manufacturers to use test equipment capable of delivering this level of power.
Test hardware capable of pulse width modulation — Pulse width modulation (PWM) is a common method of controlling the brightness of LEDs. In this method, the current through the LED is pulsed at a constant frequency with a constant pulse level while the width of the pulse is varied. By adjusting the width of the pulse, the amount of time the LED is in the ON state is varied, thereby varying the perceived brightness level as well. Although the LED is actually flashing, its flash frequency is so high that the human eye cannot distinguish it from a constant light level.
Technically, one way to limit the brightness of an LED is by simply lowering the level of forward drive current used. However, PWM is a better option for a variety of reasons:
To keep the color of the light consistent as the LED is dimmed. In an LED, the color of the light it emits is related to the forward voltage at which it operates. While the forward voltage of an LED remains relatively constant as forward current is changed, it does vary, however, especially at lower current levels, where the forward voltage can change substantially. This slight variation in forward voltage equates to a slight variation in light color, which is undesirable in lighting applications. In contrast, with PWM, the LED is pulsed using exactly the same current level for each pulse, so the forward voltage is the same and the color of the light emitted does not vary.
To provide linear control over brightness — The amount of light an LED emits is not linearly related to the amount of current used to drive it; in other words, reducing the drive current by 50% won’t reduce the light output by 50% but instead by some other percentage. This non-linear property makes a dimming scheme based on varying the current difficult to apply; each LED would have to have its light output vs. forward current characterized and the drive scheme calibrated to that curve. In contrast, PWM offers a much simpler approach to LED dimming; to make an LED output 50% as much light, one simply reduces the pulse width by 50%. An LED that is ON for half as long produces half as much light.
To ensure greater power efficiency — PWM employs a constant current level for each pulse, so the pulse level to use can be selected such that the LED is operating at its most efficient operating point, i.e., where the lumens produced per watt is the highest. That means the LED is operating at maximum efficiency no matter what dimming level is used. PWM, which is simple to implement and control with inexpensive digital circuitry, also offers an efficiency advantage: LEDs actually output more light for a given drive current when pulsed rather than at dc. Many LED manufacturers’ datasheets include a graph of their products’ forward current vs. luminous flux. If the manufacturer has taken the time to characterize the LED under both pulsed and dc drive currents, such graphs make it obvious that the pulsed characterization curve lies above the dc characterization curve. This reflects the lower level of LED self-heating achieved by applying pulsed current.
Finally, PWM enhances power efficiency by employing highly power-efficient switching circuitry to drive the PWM process itself. When switched off, virtually no current flows and the LED consumes no power. When switched on, the circuit delivers nearly all the power to the LED. In contrast, in a variable current drive scheme, reducing power to the LED involves burning the excess power elsewhere in the circuit.
To emulate actual operating conditions, test LEDs under PWM by running a train of pulses through it and taking an integrated measurement of the light output over the course of many pulses using a spectrometer. During the pulse train, we measure the forward voltage on every pulse to monitor it for changes as the LED heats. Delivering this pulse train to the LED under test requires equipment capable of mimicking a PWM LED driver as much as possible, pulsing at frequencies as high as 10 kHz with duty cycles of as high as 50%. It also must be capable of measuring the forward voltage at every pulse precisely, triggering a spectrometer to start the light measurement, and responding to the spectrometer’s signal to stop.
- To keep the color of the light consistent as the LED is dimmed. In an LED, the color of the light it emits is related to the forward voltage at which it operates. While the forward voltage of an LED remains relatively constant as forward current is changed, it does vary, however, especially at lower current levels, where the forward voltage can change substantially. This slight variation in forward voltage equates to a slight variation in light color, which is undesirable in lighting applications. In contrast, with PWM, the LED is pulsed using exactly the same current level for each pulse, so the forward voltage is the same and the color of the light emitted does not vary.
- Equipment capable of precise timing and consistent pulse widths and shapes — The HPLED test equipment chosen must offer precise timing and consistent pulse shapes in order to ensure the reliability and repeatability of the device efficiency calculations. We calculate an LED’s efficiency by dividing the light power out of the LED by the electrical power into the LED. As a result, consistent pulse widths and shapes are crucial to obtaining consistent measurements of the light power out of the LED spectrometer by ensuring that we are applying the same amount of power to the LED with each pulse.
Fortunately, for OEMs of HPLED modules, the latest generation of source-and-measure instrumentation reflects the requisite capabilities for test of these modules. For example, Keithley’s Model 2651A high-power System SourceMeter instrument provides 200 W of dc power and up to 2000 W of pulsed power output (Fig. 1). Its PWM capability offers a programmable duty cycle from zero to 100% in the standard dc operating areas and a duty cycle up to 35% at 50 A and up to 50% at 30 A in the pulse-only extended operating areas. It has precision timing and synchronization to 500 ns to ensure high pulse-width accuracy; its flexible digital I/O simplifies triggering other instrumentation included in the test system, such as a spectrometer. Finally, it has a 100-nA range with 1-pA resolution, which makes it suitable for handling other types of electrical tests that HPLED modules typically require, such as reverse leakage testing.
Cabling considerations in HPLED module testing
As detailed above, characterizing HPLED modules demands the use of large currents and high power levels. As these levels rise, so too does the importance of proper test system cabling.
Source measurement units (SMUs), generally acknowledged as the best instrumentation option for LED testing, couple source and measure operations that would normally be done by two separate instruments into a single, simultaneous operation. SMUs allow test system integrators two cabling connection options between the instrument and the LED under test: two-wire and four-wire connections.
In a two-wire configuration, the cabling is simple; the same leads used to source the current to the LED are used to measure (or sense) the voltage. However, this approach has a significant disadvantage; the leads through which the voltage is measured are carrying a large amount of current. Due to resistance in the leads, this large current creates a voltage drop across the leads between the instrument and the DUT. For example, if the test current applied to the LED is 20 A and each test lead has 50 mΩ of resistance, the voltage drop across the test leads would be 1 V per lead or a total of 2 V. In two-wire mode, the voltage measurement isn’t actually taken at the LED but instead at the point at which the test leads connect to the instrument, so the voltage measurement returned includes not only the forward voltage of the LED but also the voltage across the test leads. For this 20-A test current example, that would produce a forward voltage measurement that is 2 V higher than the true value.
Fortunately, for LED test system integrators, SMUs accommodate four-wire or Kelvin connections. In four-wire mode, a separate set of test leads (the sense leads) measure the forward voltage at the device. These leads provide input to the instrument’s voltage measurement circuitry, which has very high impedance, so virtually zero current will flow. With no current flowing in these leads, there is no voltage drop across the leads, so the voltage measured across these leads is the same as the voltage across the LED.
Although four-wire mode can eliminate voltage drops in the test leads caused by high currents, excessive lead resistance can still cause measurement problems, including wasted power. Ohm’s Law (V=I*R) tells us that the voltage drop (V) is equal to the current level (I) multiplied by the resistance in the test leads (R). Therefore, a test current of 20 A and test leads with 200 mΩ of resistance would produce a voltage drop of 4 V. Power equals current multiplied by voltage (P=I*V), so 20 A multiplied by 4 V means 80 W of power that the test leads must dissipate as heat.
Another reason to minimize test lead resistance is that there is a limit to how large a voltage drop an SMU can compensate for. Unlike a DMM, in an SMU, the voltage-sense leads are not just for making voltage measurements; they actually feed voltage back to the instrument so it can adjust its output based on this feedback. This voltage-sense function is actually part of the SMU’s source architecture rather than just a measurement circuit and SMUs have a limit to how much drop in the test leads they can handle. Exceeding this limit will cause the SMU to take incorrect readings and may have other adverse effects on the level being sourced. The size of the drop an SMU can handle varies from instrument to instrument, so consult the SMU manufacturer’s datasheet for this specification and select wiring with resistance low enough to stay below this limit.
Finally, excessive lead resistance will actually slow the SMU down. Fast rise and settling times are desirable when pulsing an LED to obtain the narrowest pulse widths to minimize self-heating. Minimizing lead resistances makes faster rise times possible.
However, minimizing lead resistance is not the only cabling challenge to take into consideration when working with large currents; minimizing inductance is equally important. Inductance is essentially a resistance to changes in current resulting from voltage drops created across the test leads during the rising and falling edges of the pulse when the test current is changing (Fig. 2). Calculate the magnitude of this voltage drop using the equation:
V = L × di/dt,
where V is the voltage drop, L is the size of the inductance, and di/dt is the change in current over time. That means any inductance in the test leads will cause an additional voltage drop during the rising and falling edges of the pulse but will have no effect at the top and bottom of the pulse, where the current is constant.
Although inductance will cause a voltage drop across the test leads during the rising and falling edges of a pulse, SMUs compensate for this and will deliver additional voltage to drive through the drop; however, there is a limit to how much additional voltage it can output. Once we reach this limit, the rising edge of the pulse will slow to a level at which the inductance does not create a bigger voltage drop than the SMU can handle. Maintaining the speed of the pulses depends on minimizing the inductance of the cables.
Always check an SMU’s datasheet for the specification on the maximum cable inductance the instrument can handle. Exceeding this limit can cause the SMU’s output to become unstable and may damage the LED under test.
Tips for Optimal Cabling
Fortunately, for test system integrators, minimizing the effects of excessive lead resistance and inductance is relatively uncomplicated. The simplest way to minimize lead resistance is to use the appropriate gauge of wire for the current levels involved. The greater the current is, the thicker the wire you would use.
For example, when testing with 20 A dc, one would use 12-gauge or thicker wire. Another way to minimize lead resistance is to minimize contact resistance; that is, the resistance where two sections of the electrical path come together, such as where the test leads connect to the DUT. Minimizing contact resistance also requires keeping all contacts clean and secure. If solder is used, double-check the quality of all solder joints to prevent problems due to poor contact and high resistance. Finally, minimize the number of connection adapters and/or cable extensions in the path between the instrument and the DUT. Each connector is another contact point that will typically add at least a few milliohms of resistance to the path. Eliminate that resistance by using a longer cable rather than an extension.
Minimizing lead inductance is also surprisingly easy to accomplish; simply reduce the loop area between the high and low test leads by twisting the leads together into a twisted pair, separating them only where absolutely necessary. Coaxial cable is also better for this purpose than individual test leads. In a coaxial cable, wrap the low lead around the high lead with a thin insulating barrier between them, which minimizes the inductance. Coaxial cable is also easier to manage than a twisted pair of leads. The cabling for Keithley’s Model 2651A high-power SourceMeter instrument accounts for these considerations. The coaxial cable supplied with the instrument is rated at just 3 m?/meter and 85 nH/meter, ensuring the lowest possible lead resistance and inductance.
The testing challenges posed by HPLED modules, while significant, are far from insoluble. With the right instrumentation and an understanding of the appropriate connection techniques, manufacturers can develop automated systems that allow them to control their testing costs and remain competitive in the marketplace. To learn more about HPLED module testing, view Keithley’s free online webinar, “Meeting the Electrical Measurement Demands of High Power High Brightness LEDs,” archived at http://www.keithley.com/events/semconfs/webseminars.