Algodoo

lol i think wood can't keep this forces in real life

LOL I think wood can support these forces in real life. Prove me wrong with some calculations. Most of the wood members are in direct tension or compression. At any rate, I couldn't find anything in the Terms Of Use agreement where everything made with Algodoo has to work in real life. Ideas that don't work in real life may lead to ideas that do.

I agree that wood cannot keep the forces, especially because arm is so light that it would be very thin when made with wood. But modern materials.. who knows? Carbon nanotubes have specific strength almost 100 times bigger than steel.

I'm not so impressed by the throw speed—you have an MR of 16k:1, of course it can throw really hard. What is pretty cool is that it steps down its throw speed almost linearly, with very little change in angle, down to a CW of only 500kg, which is a much more reasonable (though still extremely high) MR. It's still not as efficient as my FAKA, but that's because a FAKA is theoretically about the most efficient design known. The only more efficient design, in terms of MR, is a one-of-a-kind treb called Merlin, which I can't build in Algodoo because it uses curved load-bearing surfaces.

http://supertrebs.com/Merlin.html

BTW, feel free to take the basic FAKA design and see if you can tune it for high-MR throwing. Maybe add another section of arm, maybe just mess with the lengths... I bet you could get a supersonic throw with only a few-hundred–to–one MR.

Wow, i'll try to build Merlin - I think that joining many slihtly curved polys to get smooth enough curvature might do the trick.
MR is very high, but at 400 m/s projectile has kinetic energy of 160.000 * mass, that corresponds to 16k times heavier counterweight falling 1m. In my machine it falls about 5-6m so I have 15-20% efficiency. Supersonic speed with MR of less than few thousand will need much space for CW to fall down.

NanoTubes is too costly.
nuclear warheads more effective

This treb has about 7% efficiency, while the FAKA runs about 67% efficiency. This is based on the linear kinetic energy of the projectile divided by the potential gravitational energy of the weight (obtained from plots).

Actually, I misremembered. I have a Merlin with a one-piece track that doesn't mess up, it just kinda sucks for some reason. There's no real whip at the end and tons of energy left in the CW. Here:

http://www.algodoo.com/algobox/details.php?id=43520

The challenge, I think, is in the tuning—as the forces involved get larger the track may become an issue, but right now it's more that I'm horrible at tuning.
(You can make the CW a super-high-density box instead of the unwieldy baskets—I was working under a single-body weight restriction of 100kg.)

@s_noonan—never bothered to calculate the efficiency. Neat to know. Certainly the FAKA rattles around a lot, so it's nowhere near 100% (or even whatever the theoretical max is for the FAKA design). A perfectly tuned FAT, in particular, will have the counterweight come to a dead stop with no rebound when it lets go of the shot, while FAKAs are known for crashing around and derailing if not PERFECTLY tuned. Mine's not bad, but could be better.