Adding a multi-cell battery system to single-cell designs

Technology News | March 21, 2013

By eeNews Europe

Early tablet designs were often dominated by their predecessors: the smartphone. This meant that design teams continued to use a single-cell Lithium-Ion (Li-Ion) or Lithium-Ion Polymer (Li-Polymer) battery stack and simply added cells in parallel to achieve longer run times.

The drawback, also common to smartphone designs, was that backlight efficiency suffered. The tablet form factor also suffered due to the higher current draw of more white light-emitting diodes (WLEDs). This one-series multiple-parallel battery configuration requires a longer charge-time, thicker board traces, and higher-current connectors. Hence, a number of designers now are considering multi-cell stacks to improve backlight efficiency and reduce current levels. This creates another problem, however, how to use a smartphone design validated with components compatible with single-cell?

This article takes a different approach: using a front-end power management unit (PMU) that stands between the multi-cell battery stack of future systems with the single-cell-based designs of past systems. The PMU enables system designers to continue using their single-cell-based designs, while increasing efficiency of subsystems needing higher input voltage.

Improving efficiency
With the emergence of tablets, design teams must overcome the challenge of improving run-time with the burden of the larger screen, while minimizing time-to-market and keeping charge time reasonable.

Common smartphone systems use a single Li-Ion cell that operates at a voltage between 2.5 volts (V) and 4.35 V with 3.6 V to 3.7 V nominal. This battery needs to drive approximately eight WLEDs, depending on screen size, configured with two strings of four WLEDs. With a typical WLED forward voltage drop of 3.0 V to 4.0 V, the eight WLEDs require a forward voltage of 12 V to 16 V with additional voltage to allow for headroom drop across the current source or sink and other losses. This means boosting the Li-Ion battery voltage to meet the forward voltage drop of the WLED string.

A good rule of thumb is that efficiency drops as the ratio of output to input voltage increases in a boost converter and as more parallel WLED strings are needed. For a tablet, with a screen size two or three times that of a smartphone, the number of WLEDs increases respectively; typically at least six WLEDs in series are used and often with multiple strings in parallel. The efficiency then drops since the output voltage must now increase to 18 V to 24 V, plus the extra voltage to account current drive or sink voltage headroom, the number of strings increase, and the output to input voltage ratio increase. To overcome this toll on battery life due to the increased number of WLEDs, batteries often are added in parallel. This increases the current output capability and capacity, but does not improve efficiency.

The smartphone single-cell design has been through a number of revisions. It has been thoroughly tested and is in mass-market production. Preserving this known-good design reduces risk involved in redesign that normally would delay time-to-market. In a two or three cell Li-Ion stacked system, the voltage is nominally 7.2 V or 10.8 V. The higher voltage means a smaller difference between the battery stack output voltage and the voltage required to drive the forward voltage of the WLEDs, thus reducing losses in the boost converter.

This multi-cell battery stack configuration has trade-offs. Analogous to the way boosts are less efficient at high output-to-input voltage ratios, buck converters are less efficient at higher input-to-output voltage ratios and efficiency of converters connected to the processor, memory and IO may all suffer. Additionally, this higher voltage cannot be connected directly to most of the other circuitry in the system.

For example, the inter-integrated circuit (I²C) bus used to communicate from integrated circuit (IC) to IC inside the cell phone typically operates at a maximum of 5 V. The single-cell power management unit likely has a 6 V input voltage limit The low dropout (LDO) regulators are most efficient when the input voltage is equal to the output plus the dropout voltage. Efficiency drastically decreases, if connected directly to a multi-cell battery stack input. While overall system efficiency with the multi-series combination may be better, modifying each system component to tolerate the higher input voltage may require substantial redesign. This leads to another approach in shifting from a single-cell to a multi-cell battery stack.

Shifting to a multicell battery stack and maintaining validated design
A solution that allows for improved efficiency of the backlight with the upgrade to a multi-cell battery stack and the reuse of the validated single-cell design is to use a front-end PMU. An integrated device such as the TPS65090, which has a high input voltage range (for example, 6 V to 17 V), a charger with power-path management, three step-down converters, two always-on LDOs, seven load switches and an I²C interface, can stand between the multi-cell battery and the single-cell power system. This approach allows for the shift to a multi-cell battery stack solution by adding the front-end PMU and using an LED driver optimized for the higher voltage IC.

The front-end PMU can connect directly to the battery stack with the power-path and PMU charger functionality. This protects the system and controls charging the multi-cell battery stack using an AC adapter. The power-path, or system voltage, connects to the WLED driver and the frontend PMU switching converters input and LDOs, with outputs of 5 V, 3.3 V and a converter with adjustable output from 1.0 V to 3.3 V. Since these voltages are typical for single-cell battery systems, and the switching converters are capable of delivering a single-cell battery-like 4A continuous current, these front-end PMU outputs drive the downstream ICs in the existing single-cell design. Moreover, the TPS65090 has seven load switches, capable of reducing power consumption of disabled peripheral ICs down to 1 uA.

A simplified block diagram of a tablet using the front-end PMU is depicted in Figure 1. The two- or three-cell series Li-Ion or Li-Polymer batteries are illustrated on the left, powering and/or being charged by the TPS65090. In the top left is an AC adapter. This is a DC supply at a voltage above that of the batteries. This supply charges the batteries and powers the system during charge. In this configuration, with the front-end PMU, the AC adapter also can be supplemented by the battery stack in a high-current load situation, such as a system start-up.

To the right of the batteries is a front-end PMU. To the right sits the system PMU, then the microprocessor. Neither need to tolerate more than five volts. At the bottom are the peripherals or modules. These can be powered through the frontend PMU’s load switches. Additionally, the front-end PMU contains a 10-bit, sixteen channel, analog-to-digital converter (ADC) and provides an I²C bus to allow for configuration and for the processor to read voltages and load currents from the ADC.

Click on image to enlarge.
Figure 1: Tablet design with multi-cell battery stack simplified block diagram.

The simplified functional block diagram pictured in Figure 2 illustrates these features, as well as the required external components.

Click on image to enlarge.
Figure 2: Functional block diagram of a frontend PMU
Two for the price of one
As tablet designs are often rooted from their predecessor, the smartphone, the primary design challenges are improving run-time and having a timely release to market. Shifting to a multi-cell battery stack and preserving the validated single-cell design is possible using a front-end PMU. A device like the TPS65090 allows for the upgrade to a multi-cell Li-Ion or Li-Polymer battery stack, with the front-end PMU protecting and managing the system while allowing the WLED driver to take advantage of the higher battery stack output voltage. This increases efficiency and run-time while simultaneously minimizing time-to-market by enabling the reuse of an existing power design.