It can only richen so much however. I remember reading somewhere (that I can't find now) that small air leaks are somewhat masked by AT to the point where the user may not know there's a problem. Major leaks probably will kill the saw if the user isn't paying attention.
I may be wrong but I thought it's been said that all AT tests for is RPM, adding or removing fuel to maintain it at WOT. I guess technically that is the same as checking the mixture, but that's not what it's looking at.
The primary feedback system is to look at rpm, but it is not attempting to control the saw to a specific rpm or anything like that. What it does is to momentarily lean out the mixture every so often, and then it looks at the rpm to see what happens. Maybe a bit oversimplified, but if the saw speeds up then it was too rich, if it slows down it was too lean, etc. In reality I'm sure they're probably looking for a set small increase in rpm, but you get the idea. The reaction of the rpm to the "lean out test" tells the system where the mixture is. No input sensors are required, by looking at rpm under WOT conditions it is essentially looking mostly at the mixture.
That certainly makes sense to this feeble old brain.
