I'd love to know the answer to this question too. But I'm now thinking you've hit the nail on the head - an impact wrench works not by slowly applying torque as we would with a wrench. It works far more like beating on a wrench with a hammer. In the case of a frozen nut, we're hoping that the impact of the hammer on the wrench causes a very narrow in time torque spike that gets over the release torque of the frozen fastener, breaking it loose.
(Note to purists - it's possible that in talking torque, I'm not adequately describing the physics that are going on here. Instead of a torque peak phenomena, it may be more accurate to look at things in the energy domain - IE: it's not a torque spike, but an energy spike that overcomes the energy required to break the bond that's binding the nut. But, that's more esoteric that I think this discussion needs to be.)
The torque stick acts like a spring, winding up to limit the torque that's applied through it. Or, you could say that the torque stick drops that torque peak by spreading it out over time. But, the limiting effect absolutely is dependent on the impact applied by the impact wrench. If I take a small torque stick, and apply a monster impact to it, it won't apply the same amount of torque to the fastener if I used my el-cheapo Harbor Freight unit that barely puts out any impact.
ALL THAT SAID, I believe that you need to go somewhere else for your tires! Torque sticks and impact wrenches are not the proper equipment for the final torquing of tire lug nuts, for reasons you've noted - they can apply waaay too much torque! Always use a hand-applied torque wrench for final torque, never a torque-stick! I will attempt to find a cite for this.
However, something to note is that breakaway torque is not a super-accurate measure of how much torque was applied. It's about a 10% accurate measurement, at best! Your results of your breakaway torque clearly do show your lug nuts were overtightened. If the mechanic said he tightened to 100 FT-LBS and you measured anywhere from 90 to 110, I'd say that would be right.
(A side note - on my second job, I was given the responsibility of writing a company wide torque spec. I spent about a month researching torque, how accurate torque wrenches are, how accurate is torque as a measure of pre-load (it's not!), and various commercial and military specifications. Part of my research included a testing of break-away torque as a quality control method, as we had hardware fail NASA flight vibration testing and were using breakaway torque to try and understand if fabrication QA had failed. In our case, we couldn't demonstrate that breakaway torque was a good enough measurement to determine if or not fabrication QA had goofed.)
thanks much,
ben
