Sounds ambitious. A gram of Pu‑238 gives you about 0.5 W thermal, so to get 10 years of useful power you’d need a system that’s almost 100 % efficient, which is impossible with current thermocouple or thermophotovoltaic tech. The self‑healing and redundancy add mass and complexity that the tiny power budget can’t support. You could look at a small fission‑based reactor, but that’s a different design. If you’re set on RTG, you’ll need a few grams at least, and then focus on shielding and thermal management, not the “super‑compact” dream. Let's sketch the mass budget first and see where we hit the limits.

Okay, start with the basics: one gram of Pu‑238 gives about 0.5 W of thermal power. Even if you push the conversion efficiency to 20 % with a state‑of‑the‑art thermocouple, you’re looking at 0.1 W electrical. To run a rover at, say, 5 W average for 10 years you’d need roughly 500 grams of fuel, not a single gram. Mass of the generator housing, shielding, heat straps, and control electronics would push the total to maybe 3–5 kg per 5 W of output. So unless you can get a 50–60 % conversion efficiency and cut the mass of the shielding by a factor of ten, the “absurdly efficient zone” is a myth. Let’s draft a rough mass breakdown: Pu‑238 mass, converter block, thermal straps, shielding, electronics, redundancy modules. Then we can see where the big gaps are. And yeah, a little nuclear jazz might sound cool, but real physics doesn’t bend that easily.