That's an very good point about the radiators—you need some overcapacity with a fast-decaying radioisotope, so you will need more heat dissipation.

It's still not a close call, as far as I can tell (I'm not a domain expert). If you plan for even 20 years operating life, then your initial heat dissipation (for Pu-238) would be 1.17 of the nominal power at t+20 years, due it its radioactive decay. The largest RTG, GPHS-RTG, has 13.0 kg of "housing and fins" [0]; if you scale that linearly by 1.17 that would add +2.2 kg.

On the flip side, the isotope mass that emits 4,500 Watts of heat goes from 8.3 kg of Pu-238, to 39.5 kg of Am-241—that's +31.2 kg.

This is relative to a total mass (GPHS-RTG) of 56.0 kg.

(To preempt anyone complaining that this are small quibbles: the total dry mass of New Horzions [1], which contained one of these RTG's, is 401 kg).

[0, pdf] https://ntrs.nasa.gov/api/citations/20080003866/downloads/20... (table 1 on page 3, "State-of-the-art RPS. Comparing existing GPHS-RTG with near-term MMRTG and SRG110")

[1] https://en.wikipedia.org/wiki/New_Horizons

Strontium-90 has advantages. At density 2.6 vs 12, you can afford to loft more of it to compensate for its 28y half-life, and still come out ahead provided you can dispose of the extra heat at first. Few missions need to run decades.

Plutonium-238 decays to uranium-234: https://www.planetary.org/space-images/decay-of-pu-238-to-u-...

U-234 has a 246,000 year half life and can be safely ignored when calculating the remaining heat output.