Alright. I'll be the first.
Roller dynos don't guess or estimate, they calculate torque and HP exactly based on the mass of the roller (not the car) and the energy necessary to accellerate the roller. Where the problem lies is translating that back to the engine. Just the same as the dynapack, they have to work in correction factors for STP. Then some bozo pulls a number out of his a$$ for gear losses, same as the dynapack.
The uncorrected numbers from the dynapack should not be significantly different from the uncorrected numbers from most any decent roller dyno with the exception of wheel slippage which is not a problem for the majority of cars.
I would expect the dynapack to be more repeatable, however, which is what you really need to compare before and after mods. But you didn't say that. We need to differenciate between accuracy and repeatability in these discussions. The Dynapack is no more accurate than the guesses that go into the correction factors, just like an inertial dyno.
|