National Semiconductor Corporation announced the first in a new class of high-efficiency switched capacitor voltage regulators designed for dynamic step-up (boost) and step-down (buck) conversions. The LM3352 regulated 200mA DC-DC voltage converter offers portable system architects a high-efficiency alternative to LDO regulators, and enables users to take advantage of the full capacity of Lithium-ion batteries. The LM3352 is also a low-cost alternative to inductive switchers, minus the limitations of inductor height and manufacturing issues.

The new device achieves more than 80% average efficiency while producing a constant 3.3V output voltage over the complete discharge cycle of a Lithium-ion battery pack. The LM3352 produces up to 200mA of output current, the highest available in a regulated switched capacitor dc-dc converter and enough for the primary supply in cellular phones, hand-held organizers and digital cameras.

"The unique fractional gain switching architecture implemented in the LM3352 helps maintain high efficiency at high load currents through an entire Lithium-ion battery pack discharge cycle, even when the cutoff voltage is reduced to new lows," said Venkatesh Shan, National's power management marketing director. "With this design, OEMs can significantly increase battery life, while saving design cost and space," he added.

The LM3352 is the first switched capacitor to use an adaptive digital signal processing control system to optimize conversion efficiency and voltage regulation over a wide range of input voltages and load conditions. It is also the first switched capacitor architecture to use multiple fractional gain modes to provide a regulated output voltage between 1.8V and 4V from an input voltage between 2.5V and 5.5V. Output voltages of 2.5V, 3.0V and 3.3V are offered to match industry standard voltage requirements. Non-standard output options between 1.8V and 4.0V in 100mV increments are also available.

This is a technology story with a marketing whack. The architecture of the LM3352 is extremely clever with the adaptive, automatic, switching points optimized for the standard output voltages; this gives extremely high efficiency across the input voltage range for any of those outputs. Because it is a switched capacitor arrangement there is, of course, no bothersome inductor to deal with and the full range of a Lithium-ion battery's discharge cycle from 4.2 to 2.5 V can be used, whereas an LDO would have already shut off much earlier if your operating voltage is higher than 2.5 V. The three capacitors required can be 0.33 uF ceramics, with X7R dielectric recommended, and are obviously not a size problem. The capacitors do need to be physically close to the package because of the approximately 1-MHz switching frequency employed (the actual switching frequency varies with both the input voltage and the ambient temperature.

The fractional-gain technology used was introduced to us by National in the previously released LM3350 and LM3351. The adaptive-switching operation of this part is dependent on two feedback loops -- one derived from the output voltage, and the other from the input voltage -- which are effectively compared against an on-board reference to determine the switching control. The efficiency curves turn out to be jagged as each switch takes place across the varying input voltage and changes with the load. The absolute best average efficiencies seem to occur close to full load at 200 mA and close to a low load of 20 mA, a good range for a switch from transmit to receive, but all loads give results at approximately 90% at some points and as low as 70% at others.

It is difficult to judge what effect this technology is going to have on the market in the medium term (18 - 24 months) but long term I think it will be well-embraced by those parts of the market eager to get the most out of the battery cells; but there need to be some changes to the parts to secure the design wins that National obviously hopes for: the load current is a mite on the low side for the really high-volume applications even at 200 mA and, in practice, this is really a good 100 mA part with a "so-so" 200 mA performance (going, basically, in the hole at about 3.5 V on the input); I don't like the absolute maximum rating on the input pin of only 5.6 V; I read no mention of any reverse battery protection; the package is way too big with too many wasted pins (there are 5 ground pins plus 1 non-connect); the recommended output capacitor of 33 uF still gives a ripple voltage of 100 mV at 100 mA load; presumably that ripple has a fundamental of 1 MHz which is worrisome for use in RF-based products.