NASA looks to be developing larger (12-40KW) radioisotope thermoelectric generators for the moon's surface. Hopefully NASA finishes this work as they have a history of starting big RTG projects and not completing them. Also, it probably would be better to use the Hyperion uranium hydride reactor instead, which has been covered extensively on this site and should also be ready in 2013 and would provide 1000 times more power
The primary components of fission surface power systems are a heat source, power conversion, heat rejection and power conditioning and distribution. Glenn recently contracted for the design and analysis of two different types of advanced power conversion units as an early step in the development of a full system-level technology demonstration. These power conversion units are necessary to process the heat produced by the nuclear reactor and efficiently convert it to electrical power.
The first design concept by Sunpower Inc., of Athens, Ohio, uses two opposed piston engines coupled to alternators that produce 6 kilowatts each, or a total of 12 kilowatts of power. The second contract with Barber Nichols Inc. of Arvada, Colo., is for development of a closed Brayton cycle engine that uses a high speed turbine and compressor coupled to a rotary alternator that also generates 12 kilowatts of power. Testing of the non-nuclear system is expected to take place at Glenn in 2012 or 2013.
NASA new effort is probably going for 35-45% efficiency in power conversion using Brayton cycle or piston engines. This would mean 500kg of nuclear fuel to get 100KWe or 200kg for 40KWe.
NASA worked on larger 100KWe-1000KWe nuclear RTGs in 1978-1985 (SP-100) and again in the 90s and again around 2005 (Prometheus). Each time they started and did not complete the work. They did not complete them. Russian RTGs (Topaz-1 and Topaz-2) reached 10KWe of power.