I've come up with an idea even better than Particle Puff Propulsion. Now that I've thought of it, it seems so obvious:

Particle Puff Propulsion relied on doing something "clever" to the design of a nuclear bomb to make it concentrate kinetic energy into a small fast-moving puff. This meant doing something challenging in hundreds of thousands of devices.

Torusail Runway Pulse Propulsion, in contrast, takes the opposite approach--use normal nuclear bombs, and do something clever on the starship side of things to make use of the slow moving bomb products. This means doing something challenging in just one device.

Well, once I thought of how to design the "clever sail", it turned out to be pretty easy to do. One of the simplest magnetic geometries does it--a simple torus. Remarkably, a torus magnetic field deflects bomb products in a way that efficiently acheives any velocity, limited only by the speed of light.

http://www.geocities.com/mechdan/torusail/index.html

The theoretical efficiencies are close to the absolute ideal. The costs scale quadraticly with cruise velocity--proportional to the kinetic energy of the starship. In contrast, a rocket's costs scale exponentially with cruise velocity. If you want a rocket to go much faster than the exhaust velocity, you pay dearly for it. With torusail propulsion, you can go as fast as you want and you only pay the going conservation-of-energy rate for it.

What I love about torusail propulsion more than anything else is that the physics and technology involved are very simple so even I can understand and analyze it. The magnetic fields involved are relatively low strength and used in simple ways well within what's already been demonstrated in the lab. There aren't any supercompressed or superheated plasmas involved; no plasma instabilities to worry about. There aren't any novel nuclear bomb design issues to worry about--the fission-fusion-fission device is essentially off-the-shelf.

I think I've finally conceived of a real winner, here.