Drive time-multiplexed LED arrays at high current (MAGAZINE)

April 9, 2013
Adapting technology developed for display backlighting, you can drive large LED arrays at high current levels in general lighting products using time multiplexing, explains Joel Gehlin.

This article was published in the March 2013 issue of LEDs Magazine.

View the Table of Contents and download the PDF file of the complete March 2013 issue, or view the E-zine version in your browser.


System designers have adopted time-multiplexed architectures for large-scale LED matrices in recent years in order to achieve a large reduction in the number of current sinks/sources required in the circuit. Such an architecture reduces the size and cost of the electronics circuitry in end products with large LED arrays, such as smart commercial lighting and RGB signage. But most implementations are limited in the current at each LED and therefore the lumen output each can produce. You can realize both high current at each LED, and a cost-effective design buy using technology developed for TV backlighting in general solid-state lighting (SSL) applications.

A multiplexed design is more difficult to implement when a high current is required at the LED. The limitation is due to the refresh rate applied by the LED driver in order to spread the current evenly throughout the LED matrix if more than one LED is on at the same time. As a result, designers have found it hard to deliver high current output, high efficiency, low cost and small size using conventional LED driver ICs.

We will explore how the refresh rate impacts designs momentarily. But first let’s look at the limitations of many driver ICs intended for general lighting applications. Conventional LED driver ICs tend to be able to drive a high number of LEDs in a matrix configuration. Close examination of the datasheets, however, reveals the problem. The constant current at each sink/source in matrix configurations is typically in the range of 10-40 mA. Some can deliver as much as 150 mA.

Fig. 1. In fact, for many large-array display applications, a real 150-mA supply would be adequate – but the refresh rate applied in time-multiplexed architectures means that the effective peak current at the LED is often one-half or one-third or less of the chip’s nominal peak.

Time-multiplexed matrices
Time-multiplexing is a technique for driving LEDs in a matrix without requiring a dedicated power source for every LED. Fig. 1 shows the operation of a time-multiplexing scheme. To power on the LED D1, Source1 needs to be supplied with a voltage higher than the maximum forward voltage (VF) of the LED and Sink1 needs to be connected to a resistor or other type of current sink to draw the current through the LED. LED D5 is controlled in the same way via Source2 and Sink2.

But what if D1 and D5 need to be powered on at the same time? If Sink1/Sink2 and Source1/Source2 are all enabled, LEDs D2 and D4 will be powered on as well. To overcome the problem, the concept of time-multiplexing must be used. Instead of turning Source1, Source2, Sink1 and Sink2 on continuously, the driver multiplexes between Source1/Sink1 and Source2/Sink2, refreshing the appropriate LEDs at different points in time.

Provided the flickering of LEDs D1 and D5 is at a frequency of 50 Hz or higher, the light will appear to the human eye to be continuously on. This time-multiplexing technique using an effective refresh rate faster than 50 Hz thus permits D1 and D5 to be lit without lighting D2 or D4.

There is, of course, a drawback—the time-multiplexing with the associated refresh rate reduces the total LED current passing through the LEDs. Let’s say that a given matrix refresh rate for a given set of lit LEDs produces an effective 50% duty cycle applied to the LEDs. At a current set to 100 mA via the current sink, the effective constant current through each LED is 50 mA.

There might appear to be an obvious way to combat this effect. You could double the current at Sink1 and Sink2 to 200 mA to provide a constant current of 100 mA through the LEDs. Unfortunately, a current output of 200 mA is beyond the capability of the conventional LED driver ICs on the market today when operating in a matrix configuration.

Multiplexed refreshed rates
The problem becomes more pronounced when you increase the size of the matrix and therefore further reduce duty cycle. The refresh rate describes the number of times per second that the current through each lit LED in the matrix is refreshed. An example of a matrix control scheme is shown in Fig. 2. Here, D1, D5 and D9 are being lit with a current of 100 mA through each LED.

Fig. 2. If LEDs D1, D5, and D9 need to be lit simultaneously, the control circuitry must perform three sequential cycles each time the LEDs are refreshed essentially a multiplex sequence within a refresh cycle. First Source1/Sink 1 is enabled for D1 and then disabled, followed by similar cycles on Source2/Sink2 and Source3/Sink3 for D5 and D9.

Provided a multiplexing scheme is looped fast enough – between 200 and 1000 times per second, depending on the number of LEDs to be lit simultaneously – the LEDs will appear to the human eye to be continuously on. In Fig. 2, a refresh rate of 200 Hz for the entire matrix means that each LED will be switched at around 67 Hz, which corresponds to a duty cycle at each LED of 33%. This means that each sink needs to handle at least 300 mA in order to produce the constant current equivalent of 100 mA at each LED.

Time-multiplexing also enables the creation of animations. The animation may be created in software code with a pre-defined series of bitmap images that are usually represented as arrays of data bytes, in which each bit represents one LED in the matrix. To realize the picture, the controller must scan through each element of the data array one byte at a time, displaying one column after another.

Ramping LED current
A potential solution to the problem comes from a device type that is far from the most obvious choice – LED driver ICs designed for TV backlighting. The rapid growth in the market for LED TVs has spawned a new generation of sophisticated and highly efficient driver ICs that provide a high current capability. It turns out that the capabilities of backlighting driver ICs can match the requirements of large lighting and signage systems. Because of the requirement for high brightness in TVs, these ICs must be able to control LEDs at high currents, either via external MOSFETs or via FETs embedded in the driver chip.

An example of such a device is the AS3693B from AMS, a 16-channel, high-precision LED controller with built-in pulse-width modulation (PWM) generators for driving external FETs. A sister IC, the AS3693A, features integrated MOSFETs. While the AS3693 family was specially designed to meet the precise current-control requirements of TV manufacturers, the devices may also be used to source/sink and control LEDs in other applications.

A key to the usage of backlighting drivers is the ability to program the current through any channel. In the case of the AS3693, current control can be accomplished via independent digital current control for each channel with a PWM generator; linear current control with an 8-bit ditital-to-analog converter (DAC); or linear current control with an external analog voltage.

While the backlighting driver can clearly work in the intended application, it can also be utilized for:

• Multi-pixel advertising boards,
• Traffic signals,
• Backlit signage,
• General illumination, and
• RGB accent lighting.

LED exit sign
Let’s further examine one application – an LED exit sign. Safety exit signs powered by LEDs are up to 90% more efficient than traditional incandescent signs. Operating 24 hours a day, the cost and energy savings to be realized by switching to LEDs are thus very considerable.

LED exit signs also offer savings in maintenance and repair, since their re-lamping cycle is typically 10 years. In addition, LED signs can offer better optical performance.

Fig. 3. Fig. 3 shows how a single AS3693B can control a time-multiplexed matrix of 60 white LEDs. Alternatively, you could use a very similar circuit to control 3 banks of 20 RGB LEDs. The use of such a device in a time-multiplexed architecture offers the system designer considerable bill-of-materials and space savings compared to a conventional design, which would require four 16-channel driver ICs of the conventional type. The conventional design would also occupy a much larger circuit-board area and incur a much higher bill-of-materials cost.

The constant current through each LED is provided by NMOS transistors. Selection of an appropriate NMOS device will provide for high efficiency and high brightness. The maximum current can be limited by the RSET resistor connected to the source of the external MOSFET in each current sink.

Unlike the case of a conventional LED driver IC, the IC in Fig. 3 does not have to source the required LED current. Instead, current comes directly from another supply source that we will discuss in more detail momentarily.

To reduce power consumption even more, the AS3693’s current settings can also adapt to ambient light levels. During daytime, the sign could be dimmed.

As light levels drop the system could boost the brightness to produce higher visibility and contrast.

Fig. 4.Supplying matrix current Back to the current source, the LED matrix could be connected directly to, for example, a 12V constant-voltage power supply. Doing so, however, would reduce the design’s efficiency as power is dissipated as waste heat in the NMOS transistors when the full-scale voltage isn’t required. To optimize efficiency, for instance in battery-powered equipment, the LEDs can be powered from an external DC-DC converter that can dynamically change its output voltage to match the VF requirements of the LEDs powered on at any moment.

The operation of such a circuit is shown in Fig. 4. The AS1341 is an efficient step-down converter with adjustable output voltages ranging from 1.25V up to a maximum 20V. The IC senses the output voltage requirement via a resistive voltage divider. This voltage divider can be modified to set the output voltage between a minimum output voltage and a maximum output voltage which is the basis of the device’s dynamic feedback control.

The AS3693B offers three different paths for feedback to control an adjustable power source and each LED can be assigned to a specific LED supply. In the case of Fig. 4, the three paths are tied together –pins FBR, FBG and FBB – and used to control the voltage divider input to the AS1341. In other designs the AS3693 could control multiple power sources. And each PWM generator in the AS3693 can be independently paired with any of the three feedback paths.