Maxim
Currently, portable products widely use color LCD displays with white LEDs as backlights. Powering a white LED requires a special converter that requires high-voltage and constant-current driving of the LED forward conduction to reduce the brightness variation caused by the battery voltage change and the brightness mismatch between the different LEDs. To achieve this, there are two main types of converters: inductor-based boost converters and capacitor-based charge pump converters. Both converters have their own advantages and disadvantages, and it is necessary to decide which architecture to use according to the specific requirements of the system.
This article compares the two conversion architectures with the MAX1561 boost converter and the MAX1573 charge pump as an example. The advantages of each converter are evaluated in this paper, and the conclusions obtained help the system designer to choose the right solution. The MAX1561 and MAX1573 are designed at the same time, in the same factory, using the same process, with a switching frequency of 1MHz, which is suitable for comparison.
Circuit complexity: charge pump slightly dominates
Figure 1 shows the circuit diagram of two schemes. Both circuits have only a few simple external components, but the boost converter requires an inductor and a Schottky diode (some boost converters have Schottky diodes integrated inside, but usually Will reduce efficiency).
Figure 1. The MAX1561 boost converter (a) and the MAX1573 charge pump (b) are two LED powering schemes. The circuit complexity is basically the same, but the charge pump does not require an inductor.
Efficiency: The efficiency of the charge pump is slightly superior
Figure 2 shows the efficiency of the two schemes, measured in the case where a standard lithium battery is discharged at a C/5 rate to power the LED. The efficiency curve of 18mA/LED represents the efficiency under normal display brightness. The average efficiency of boost converter and charge pump is 83%. The efficiency curve of 2mA/LED in the figure represents the efficiency of LED when it is in dark state. The charge pump can achieve an average efficiency of 76%, which is significantly better than 59% of the boost converter.
Figure 2. The MAX1561 boost converter (a) and the MAX1573 charge pump (b) have an average efficiency of 83% over the entire battery operating time under 18mA/LED test conditions. When the LED is dark, 2mA/LED, the charge pump efficiency is higher than the boost converter.
The above results are unexpected because the efficiency of most charge pumps does not reach such efficiency. The MAX1573 delivers industry-leading efficiency because it includes a 1x voltage bypass and 1.5x boost boost charge pump mode with adaptive switching, and a low dropout linear current regulator can be used when the battery voltage drops. Maintain the 1x pressure mode as much as possible to achieve high efficiency. The traditional charge pump solution does not have a 1x mode and can only achieve 50% to 67% efficiency. Some competing products also include the 1x mode, but the time spent working in this mode is shorter, so the average efficiency is generally less than 83%.
For boost converters, the MAX1561 is an industry-efficient product. Higher efficiencies can also be achieved with some trade-offs, such as the MAX1599, which is 87% efficient at 18mA/LED and 71% at 2mA/LED. The MAX1599 is very similar to the MAX1561 except that the switching frequency is reduced from 1MHz to 500kHz, reducing switching losses at the switching frequency. However, the reduction in frequency increases the size of the external inductor by a factor of two.
Physical size; charge pump dominates
Figure 3 shows the PCB layout for both options, including external components. The boost converter has a small pin count, allowing for a small, 3mm x 3mm package, but the inductance makes the overall size larger and higher. An inductor of about 1 mm in height even takes up more board space than Figure 3. Although the charge pump itself is large in size, 4mm x 4mm, it requires only a small 1μF ceramic capacitor. Figure 3 (b) shows the 0603 package capacitor, at least three manufacturers can provide the 0402 capacitor shown in Figure 3 (c). In the case of particularly demanding size requirements, the MAX1573 is available in a 2mm x 2mm package and the entire charge pump solution is only 11mm2.
Figure 3. The boost converter (a) takes up more board space and height than the charge pump (b) because of the inductor. If you use a 1μF capacitor in the MAX1573 and 0402 packages in a wafer-level package, the overall charge pump solution (c) will be very small.
System flexibility: boost converters predominate
An important advantage of the boost converter is that it supports series LEDs, and the charge pump can only drive parallel LEDs. As can be seen from Figure 4(a), the LEDs arranged in series require only two wires between the boost converter and the LED. This advantage is important if the boost converter or charge pump is placed on the system board and the LED module is placed on the display panel. In this case, the boost converter requires very few contacts. In addition, the boost converter can support more LED modules, each of which can be connected in series with a different number of LEDs. Moreover, in practical applications, it may not be necessary to change the boost conversion circuit to replace the display module; or the boost converter may be changed without changing the display module. It can be seen that the series LED architecture greatly reduces the design risk.
In order to improve the efficiency of the charge pump, in the battery direct drive mode, each LED requires a separate current regulator, as shown in Figure 4 (b). If you change the number of LEDs, the LED connection must also change. Moreover, in order to turn off the current source that is not used, it is sometimes necessary to change the circuit (for example, connect the current regulator that is not used by the MAX1573 to IN). Some competing solutions can create many problems in this situation: unused current regulators need to be turned off in different ways (for example, connected to OUT or floating); even worse, the newly designed charge pump may use a common cathode LED. Instead of a common anode configuration, this requires more changes to the display module.
Figure 4. The boost converter (a) has only two wires to the LED; the charge pump (b) requires more wires. Therefore, the use of a boost converter is more flexible, the LED configuration can be changed without changing the boost circuit, or the boost converter can be changed without affecting the LED configuration. When using a charge pump, the LED must be matched to the IC.
Ripple and noise: charge pump dominates
Because charge pumps and boost converters are switching converters, they generate voltage and current ripple at the input and output, and EMI at the inductor and switching nodes. Sometimes, these ripples and noise can couple into system circuits, such as the RF receiver of a cell phone, affecting performance.
Input ripple is obviously important because the battery input is common to many circuits in the system. As shown in Figure 5, the input ripple generated by the charge pump and boost converter is of the same order if the same input capacitance is used at the same switching frequency and driving the same load. It should be noted that only a 1μF ceramic capacitor is required at the MAX1573 input. To compare this with the MAX1561, we increase this capacitance to 2.2μF. Increasing the input capacitance to 4.7μF or 10μF further reduces input ripple, but increases cost and increases physical size to some extent.
Figure 5. If the switching frequency is 1MHz, driving the same number of LEDs, using the same input capacitance, the input ripple of the charge pump (b) and boost converter (a) is essentially the same. However, because there are more leads between the charge pump and the LED, it is recommended to use a shorter wire (antenna). In addition, the pump capacitor produces less EMI than the boost converter.
Output ripple is also a problem, especially if the output line is long, antenna effects may be generated or coupled to adjacent circuits. In order to solve this problem, it may be more inclined to choose a boost converter, but only because it requires fewer output leads, it can be placed far away from the LED. Charge pumps require ICs and LEDs to be as close as possible because of the large number of output connections.
The boost converter stores energy in the electromagnetic field of the inductor and produces more EMI than the charge pump capacitor. Therefore, it is recommended to use a shielded inductor or shield the system. In addition, the boost converter has fast high-voltage fluctuations at the junction of the inductor and the Schottky diode. A small capacitor can be added to the switching node to slow the EMI radiation generated by the switching signal, but this will sacrifice efficiency.
Other features: as needed
The following issues are not characteristics of the boost converter or the charge pump itself, but these characteristics are important when choosing any specific backlight IC.
Both the MAX1561 and MAX1573 include output overvoltage protection. This feature prevents the IC from damaging the IC when the diode (or any output) is open. If you do not have this feature, you need to add a Zener diode externally.
Brightness control reduces the LED current (display brightness) when the LED is not operating to extend battery life. Users can also adjust the brightness of the display according to their personal preferences. There are many ways to adjust the brightness, including analog DAC, logic input, on/off PWM control, PWM filtering, single bus pulse interface and SPI? or I2C serial port. The MAX1561 and MAX1573 use a variety of brightness control methods.
The MAX1561 uses a CTRL input to control the brightness. This signal can be a simple on/off logic level or an analog signal from the DAC output, or a PWM signal with a frequency between 200Hz and 200kHz. Because the MAX1561 integrates a feedback loop internally, the PWM signal is internally filtered to a DC LED current with lower input/output ripple and noise compared to traditional on/off PWM brightness control.
The MAX1573 uses two logic inputs: EN1 and EN2 to control LED turn-off and current ratings of 10%, 30%, and 100%. In addition, when EN2 is driven high, a 200Hz to 20kHz signal can be added to EN1 to regulate the LED current, and the PWM signal can be used to adjust the current from 10% to 100%. In addition, the external resistor Rset of the MAX1573 is used to set the current maximum of 100%, so the brightness can be controlled by applying a different resistor or by applying an analog or logic signal to the SET pin.
The soft start is used to suppress the inrush current during startup and minimize the drop in battery voltage to avoid affecting other circuits in the system. As shown in Figure 6, both the MAX1561 and MAX1573 include a soft-start circuit. A reasonable soft-start mechanism prevents any input overshoot current, and some soft-start circuits only prevent the overshoot current from exceeding a certain limit.
Figure 6. The soft-start and shutdown waveforms of the MAX1561 boost converter (a) and the MAX1573 charge pump (b) indicate that there is no input overshoot current (IIN) to minimize battery drop to avoid affecting other circuitry in the system.
The fast, fixed switching frequency allows the use of small external components to maintain low input/output ripple. However, if the switching frequency is too high, the switching loss will increase and the efficiency will decrease. According to current semiconductor processes, the optimum operating frequency range is 600kHz to 1.5MHz. Some backlight driver ICs use different frequency PFM architectures or gate oscillator control mechanisms, which may generate large input and output ripples. The ripples have a large number of harmonic components, which may interfere with the normal operation of other circuits. If you use the PFM architecture, it is recommended to make a careful assessment before using it.
Higher current accuracy and matching result in optimal display brightness and power loss, minimizing the difference in brightness between different LEDs. Designers may be very concerned about this issue, but not as rigorous as they might think. Even if the current accuracy is reached to the extreme, the LED itself will have a brightness deviation of ±20%. Moreover, the human eye is not sensitive to 40% of the overall brightness error and ±30% deviation between LEDs.
Old-fashioned regulated charge pumps use large resistors, and the accuracy and matching that can be achieved are unacceptable. Multiple current regulators are integrated into the new charge pump to provide active control for each LED. Even so, maintaining good matching at low currents is still a challenge for some IC designs. The boost converter uses a series LED architecture to fundamentally maintain excellent matching at any current, but the boost IC also needs to ensure reasonable accuracy over the entire brightness range.
When the charge pump is switched between 1x voltage mode and 1.5x voltage mode, the mode switching hysteresis function prevents the LED from flickering. A better adaptive mode conversion mechanism is to monitor the current regulator and switch the operating mode just before the voltage drops to the lowest threshold to maintain an efficient 1x mode at the lowest possible battery voltage. It is critical to monitor each current regulator. Otherwise, some LEDs may flicker before the mode transition, causing significant LED brightness to change when the 1.5x voltage mode is turned on. Once operating in 1.5x mode, the hysteresis function avoids repeated switching between modes, resulting in large input/output ripple and significant LED flicker. If the hysteresis voltage is set too large, the charge pump will be placed in an inefficient 1.5x mode when a very small battery voltage drop occurs, and the charge pump will still be prevented from returning to the 1x mode when the battery voltage returns to normal. Therefore, the hysteresis needs to be optimized. For example, the MAX1573 not only monitors each current regulator, but also uses a patented technology to actively modify the hysteresis threshold to optimize efficiency and avoid flicker (of course, boost converter For example, the MAX1561 does not require mode conversion).
Conclusion: The boost converter gets 1 point and the charge pump gets 4 points.
The above comparison shows that the charge pump has a greater advantage, of course, depending on the specific situation and the characteristics of each driver IC. To date, most boost converters offer higher efficiency and are more common. However, since the new generation of 1x/1.5x charge-charge pumps bridges this gap, the charge pump solution will be favored in most new designs.