Engine Dyno, as used by most OEM -



hub dyno, as used by ASG, EV0R etc



Roller dyno, as used by MMS, Racetorque etc



An Engine dyno is the most repeatable, consistent, accurate dyno, but they are fairly expensive to operate and hence only used by car manufacturers / very high end workshops. An engine dyno reads directly from the back of the crank so is the most "accurate" for the "true" horsepower value. This is where subaru come up with a magical number (eg 195kw)

When you add to that the force required to spin the clutch, gearbox, tailshaft, 4 driveshafts, 4 hubs and brakes all attached to woobly stock rubber mounts on a hub dyno, you can see where some of the loss comes from. I've yet to see someone come up with an accurate, real world magical number on how much loss is involved, but there are certain rules of thumb (see kato and jdmstyles's posts) which will give you a rough idea. The values are measured at the wheel hub of the car, and thats why you get your lower figure (say 180kw)

The roller dyno is the least accurate of the 3 as it adds the rolling mass of the wheels. The benefit of a roller though is being able to get cars on much quicker (not needing to remove the wheels and align the hub packs). Some people think this is the most "accurate" final figure as its the power measured at the tyre treads and that's what the driver experiences in the real world, but its normally found via torque generated and then calculated back to give a "flywheel figure". Where a roller loses out is the 4 spinning wheels have a flywheel effect and can greatly smooth out tune inconstancies which would show up on a hub or engine dyno

questions?

__________________
MY93 Impreza GX 1.8L, 0hp, 0nm