Most small-size color LCD displays today use white LEDs for backlighting. These
systems usually involve handheld devices with an LED drive circuit powered by
a battery whose output voltage varies over time. Therefore, an optimum LED driver
design requires a system approach covering:
- Battery type
- LCD characteristics
- System power requirements and efficiency
- LED driver IC and its external components
- PCB layout and component placement
- Possible noise generated by the LED driver
- RF immunity in cell-phone applications
The most widely used battery nowadays is Li-ion. These batteries start at 4.2
V when fully charged, but drop down to 3.2 V when discharged. Therefore, the
driver circuit must operate properly over this input voltage range.
Power requirements affect LED brightness and efficiency. An LED’s light
output is proportional to its current, so uniform brightness requires dedicated
drive circuitry that controls a constant current for every member of the LED
array. This must remain true for exposure to cold temperature or when the battery
is low.
In most backlight designs, white LEDs are spaced evenly along one side of
the LCD. The number of LEDs is proportional to the dimensions of the LCD. Some
LCDs have integrated LEDs already connected either in series or in parallel.
Typically, larger-size LCDs require serial topologies, sometimes with multiple
strings in parallel. A serial configuration has the benefit of ensuring that
all LEDs in a string have the same current; thus, they will exhibit similar
brightness throughout the panel. Serial topologies have the advantage of minimizing
the number of connections to the LCD. This is usually accomplished with a flex
PCB that results in smaller size and lower cost.
LED Configuration and Driver IC
Small, white LEDs for backlighting typically exhibit forward voltages of 3.4
V at 20 mA. These LEDs may require a higher voltage than that available from
the battery, so their drive voltage must get a boost. There are two ways to
increase LED voltage in LCD designs: use a capacitive charge-pump topology or
an inductive boost converter. Series topologies often employ inductive boost
LED drivers, while parallel LED topologies typically go with fractional charge-pump
ICs. The LCD configuration and overall system requirements usually dictate the
choice of charge pump or inductive boost converter approach. A charge pump is
usually easier to implement and guarantees lower noise performance, whereas
the inductive boost converter generally exhibits higher efficiency.
Figure 1 shows a typical circuit for backlighting
a two-inch cell-phone LCD incorporating a fractional charge pump. The charge
pump is attractive because it requires a small number of external components
and minimal pc-board real estate. Meeting this goal today are 3-mm by 3-mm driver
ICs and 0402-size capacitors.
The 1X/1.5X fractional charge pump supports two modes of operation with automatic
mode selection, depending on the battery input voltage as well as the LED forward
voltage. Typically, when the battery voltage exceeds 3.6 V, the driver operates
in "1X mode" where the supply is directly connected to the output
via a pass transistor. This linear mode exhibits the highest efficiency and
lowest noise.
When the battery voltage drops below 3.6 V, the driver transitions from 1X
mode to 1.5X mode, boosting the output to 1.5 times the battery input voltage.
It uses a switching scheme with two flying capacitors (C1 and C2 in Figure 1)
that transfer energy to the load. During the transition, the input current steps
up 1.5 times, causing the battery to discharge faster. For example, with three
LEDs at 20 mA, the supply current is about 60 mA in 1X mode, and goes up to
about 90 mA in 1.5X mode.
Charge-Pump Efficiency
Charge-pump efficiency is the ratio of the output-to-input power:
Efficiency = PLED/ PIN = PLED / (VIN x IIN)
where PLED= LED power and PIN = input power. Therefore,
for a given LED current and VIN, an input current increase reduces
circuit efficiency.
Under normal conditions, most white LEDs have a forward voltage (VF)
between 3 and 3.6 V. Lower VF LEDs allow the driver to operate longer
in 1X mode, which has better efficiency. Ambient temperature also affects the
forward voltage with a coefficient of about -7 mV/°C, causing VF
to increase at a lower temperature.
The 1.5X mode guarantees a regulated and accurate current in the LEDs. Any
switching noise reflects back onto the supply, but is mostly filtered by the
input capacitor. Charge-pump solutions are desirable when low-noise performance
is critical, such as in RF communication devices.
There are two methods for adjusting LED current. One uses an external resistor
RSET, as shown in Figure 1. The
other approach is to control the driver with pulse-width modulation (PWM) that
controls LED current and brightness.
Component placement is critical. Therefore, all four capacitors (Fig.
1, again) should be placed close to the LED driver IC. When the driver operates
in 1.5X mode, switching current flows through the flying capacitors (C1 and
C2). Therefore, the loop areas should be as small as possible to avoid any interference.
Because of their low equivalent series resistances (ESRs), use ceramic capacitors
(X5R or X7R dielectric material) rather than tantalum types. You can locate
these LEDs remotely with no degradation to the LED current, which is dc (making
EMI a non-issue).
The driver will amplify noise in the RSET resistor, resulting in
LED current noise that flows back into the supply, which leads to potential
RF interference in cell-phone applications. The cell phone’s RF section
should not be placed close to the RSET resistor. The RSET
resistor should be tied directly to the ground plane with a minimum loop area
and avoid any current path.