Getting back to the original question. Could it be that the 3% difference separator on/off is due to account for the change in mass flow through the engine?
That reduced mass flow can be compensated for by just pushing the power lever slightly further forward - provided the engine is not already at one of its limits (torque, ITT or Ng). Most of the time, it will be possible to make such compensation.
Does an open separator create slightly more airframe drag? This could be the reason for the 3% factor.
GTang, thanks for that. What happens when you continue to advance the power lever beyond 96%? Does ITT, fuel flow and torque continue to rise, or do they also cease increasing?
If the answer to this is YES, then clearly you have an Ng instrument error.
If the answer is NO, and the torque is less than the limit (if the aircraft has a torque limiter fitted), then clearly the Ng governor has a fault.