Blitz, do some modeling with GGDT to see why steam would be a good idea.
I also think a ball-valve would be shorting yourself. If you're gonna work on a steam cannon, might as well make some sort of high-temperature piston valve, (investigate into some high-temp gasketing materials for the seal), and get the most "bang" for your buck. From what I hear from Navigator7, steam is pretty mean even at relatively low pressures. Also, the pressure is easier to acquire than compressed air. No moving parts like a bike pump to break, and no expensive air compressors. Everybody and their brother can buy a bag of charcoal, and has some sort of foot pump to provide an air blast to bring the heat up when they feel like it, (although not too much).
EDIT: goshdammit, I didn't see there were three pages. BLB already said why steam is better than air on page two, for anybody interested. Also, I was talking to "Blitz" on page 1 in the beginning of my post.
EDIT2: Here's a theoretical layout:
<img src="
http://www.markfh11q.net/images/steamcannon.png">
"A" is a pressure gauge. "B" is a temperature gauge. The piston and sealing face are silicone. Charcoal is dumped into the hole, and a shop-vac exhaust is fed into the hole. The temp. shouldn't be allowed to exceed around 350 F. That will keep everything within ratings, hopefully. Everything is iron, and the bushing at the front is a "double-tapped bushing" from McMaster-Carr. The back is a regular bushing.