http://www.thunderbolts.info/wp/forum/phpB ... 42#p116242
...Shawyer also says the two end plates are spherical sections (concentric, from what I can see) a microwave resonant distance apart, and it's only the interaction between the end plates which generates the thrust, at least ideally. I still say the fact that group velocity decreases with aperture size represents a momentum transfer, but if you're going to consider just the end plates, then there's still the following to consider.
When you strike a convex surface, the impact drives the convex surface together, reinforcing and hardening the surface against the impact, and distributing the impact over the convex surface more effectively than in less convex surfaces. When you strike a concave surface, just the opposite happens. The impact pushes the impact site away from the rest of the concave surface, weakening and softening the surface against the impact, and localizing the impact to the impact site more effectively than in less concave surfaces. So again, the smaller, convex end spreads less easily and receives more momentum, and the larger, concave end spreads more easily and receives more heat.
Here's Shawyer's explanation video:
https://www.youtube.com/watch?v=wBtk6xWDrwY
His explanation of how the large end gets hit harder than the small end but the chamber moves in the direction of the small end thoroughly does not satisfy me.
Here's a link containing a diagram of the heat distribution:
http://www.nextbigfuture.com/2015/02/mo ... ation.html
Also, it occurs to me that if you throw a racquet ball directly at a wall in a ship so that the racquet ball bounces directly back at you, then the racquet ball, by virtue of it still being in motion relative to the ship, did not dump all the momentum of the throw into the ship when it bounced off the wall, and did instead provide its own internal angular deflectors by virtue of its own bounciness/stretchiness, and will not have dumped all its momentum into the ship until it stops moving relative to the ship (and even then, its bounciness on impacts will have laterally dissipated some of the momentum given to it by the throw).
So, in order to provide an example of my explanation which hopefully makes the allocation of deflectiveness and momentum clearer, consider having your back up against the inside front wall of a spaceship in zero-g, placing a bowling ball under your feet with you directly between the ball and the wall behind you, and extending your whole body as hard as you can directly away from the wall to push the bowling ball directly towards the back wall of the ship, giving the ship a good amount of forward momentum, and the ball a good amount of rearward momentum. The bowling ball travels straight for a while down a tube, then comes to a gentle curve in the tube, gentle enough to change the direction of the bowling ball 90 degrees, heading towards the side wall of the ship, while reducing the momentum of the bowling ball by only half, and converting some of that loss of momentum into thrust into the ship’s rear wall, thus reducing but not eliminating the ship’s forward momentum. The bowling ball is then heading directly towards the side wall at about half its former speed, and it passes through a gently springing gate which takes a little more momentum out of the ball to get out of the ball’s way and then shut behind the ball again. Then the bowling ball hits a big spring on the side wall, and bounces gently back towards the gate, which has a big spring on the side the ball is heading towards, and the gate does not swing in the direction the ball is now heading. The ball then dissipates its remaining momentum harmlessly in successive lateral bounces off the gate and side wall. When the ball comes to rest, an operator picks it up and throws it to you at a speed which exactly cancels out the forward momentum the ship gained. Oh no, is the process ruined? Nope, you’ve got a bowling ball coming at you, and it is easy to see that, the more exactly you catch the ball in the reverse process of how it was thrown to you, then the more exactly the forward momentum will be restored when you catch the ball. You are then free to repeat the process, this time with the ship moving forward faster relative to the last time you did it. Of course, as before, in order that your ship does move straight forward in this example, you have another team doing the same thing at the same time in a bilaterally symmetrical chamber next to yours.
And again, as usual, if I am missing something in one or more of my examples or elsewhere, please let me know.