If you'll pardon a little engineering jargon, here's something that might help.
Chemical engineers use two types of reactor models. All practical cases can be reduced to a combination of the two, and practical cases always fall somewhere in between. It is our job to get them to fall close to one of the two models, usually for control reasons, but production rate (thruput) often challenges that. (Ever try to out-run an explosion? Combustion control is exactly that. Yikes!)
---Batch model, also plug flow model (reactor is a long, small-diameter pipe)- - Put everything together, then wait for a certain amount of time (or until it comes out the other end of the pipe).
---Stirred-Tank model, where stuff is added and product is removed but the reactor volume is so well mixed that composition and temperature are uniform throughout.
Our case, smoking meat, lies in between. The air/smoke/humidity are fed, circulated, then removed but the meat stays in the smoking chamber. Think of a combination: a continuous, stirred tank reactor as far as the vapors go, a batch reactor as far as the meat is concerned. (Great fun, huh?
In a smoker, we'd like for the chamber to be well mixed (no hot spots, excess moisture, too much or not enough smoke, etc.). With the types of reactions and mass transfer that we're dealing with (relatively slow), composition and temperature uniformity can be approached by using an "adequate" circulation fan. We feed smoke and air mixture at a certain temperature, the chamber is reasonably well insulated, and if it's well mixed, the composition vented is about like that in the whole chamber.
But out here in the real world, it's never perfectly insulated, never well enough mixed... So you do what you can to lessen the problems. All sorts of tricks apply
- - adequate mixing (internal fan, etc.),
- - keeping the inlet smoke/air temperature relatively constant (separating the smoke pit/generator from the chamber is a good way to do this)
- - having a small flow in relation to the size of the chamber (think in terms of turnover time, or how many volumes of smoke/air are fed/vented per hour per unit of chamber volume).
That way, we approach the stirred tank model, where all the meat will see the same vapor temperature and composition for the longest time. At the other extreme, batch or plug flow model, vapor changes with time, which we don't want. Think temperature transients, hot spots, dead corners with no smoke and low temperature...
Sorry for the lecture. Believe it or not, that's what we use in the chemical industry. The thought process is simple, but not always obvious. (That way, we get paid more? "I wish." We get laid off a lot, too.)
To make a long story short: Ross sez (and I agree), "remote heat source and a forced air circulating system." The remote heat source helps avoid short-term transients, plus it gives you a shot at cooler smoke temperature. (You Ventura and Smoke Daddy guys have better control than an open fire, so you can close-couple to the smoke chamber and get away with it.) The forced-air recirculation system evens out temperature and concentration in the smoke chamber. I would add, an adequate (but not too big) vent (sometimes, leaks!), is necessary for control. This "wastes" heat, but gives you control.