The low dropout regulator (LDO) is a wonderful component. It's simple to use, has just three terminals (dc in, dc out, ground); does only one thing; provides low-noise dc without associated EMI/RFI; requires no software; and if applied with reasonable caution, is well behaved. It's no wonder that the humble LDO is used like popcorn on circuit boards, to provide both board-level and localized power.
Since their initial development over 30 years ago, their design has been refined and enhanced, but with few fundamental changes. Dozens of vendors now supply them as commodities. You'd think that any new LDO would be a non-story. Sure, the specs and features might be a little better or more refined (efficiency, regulation, headroom, dropout voltage, and transient response, thermal and overload protection, to name a few) but otherwise, what's to innovate? But you would be very wrong.
In a recent conversation about the LT3080 LDO with Bob Dobkin, Vice President of Engineering and Chief Technology Officer at Linear Technology Corp., I got insight into how real innovation can occur even in a mature product class. Dobkin, who developed many of the original LDOs “back in the day”, looked at the LDO and understood its remaining shortcomings: inability to adjust the output down to 0 V rather than 1.2 V, needed for shutting down the load; non-surface-mount packages for higher-current LDOs; and the inability to combine multiple devices for greater current capability.
Could a new topology, using existing IC processes, provide relief? Dobkin came up with a design that overcomes these weaknesses. Independent LDOs can be paralleled using a nominal 10 milliohm “ballast” resistor between each, which is about a centimeter of PCB trace. As a result, the same LDO can be used for higher-current supplies, simplifying design and inventory; using multiple LDOs also spreads the thermal load across the board and reduces hot spots.
To further improve efficiency, the LDO can operate from two supplies, one at higher voltage for its internal control circuitry, and a lower-voltage one for the output-current pass transistor. Users can also place a resistor in series with the collector to move some of the dissipation out of the IC, further spreading the heat.burden.
The lesson of this improved design is that while everyone is looking for the next big thing, sometimes there's real benefit and often good money to be made by examining a well-established design, and seeing what's it's long-standing shortcomings may be. Even if users have become comfortable with these attributes, it's worth taking a new look, especially if the end product is radical internally while it looks conventional externally.