Danny is just saying that the sampling window should only include the actual test signal as it arrives at the mic--not noise before or reflections after. So setting the window at 2.7 ms will get things going just before the signal arrives, +plus 4 ms for the signal itself=6.7 ms.
Gotcha. I have no idea if you can do that with REW, but thanks for the explanation.
Rick, correct me if I am wrong, but the subwoofer is reaching down that low due to room gain, no? At a nearfield measurement position he won't see flat depth like that, but rather a slow roll-off starting much higher. This would be more like an anechoic measurement. Is that correct?
Room gain is often quoted but in reality often doesn't add that much bass extension. Many things are involved - size /shape of the room, speaker placement, openings to the room, wall / floor leakage,etc.
What happens is that as the software loses resolution it overstates db level. For instance, if the curve says 90db @40hz and the data below that isn't good, then everything below 40hz will tend to tend to stay around the same 90db level except for frequencies where there's a significant room mode (peak or dip).
The nearfield curve is like the anechoic measurements, actually in some ways better, because it takes a very large anechoic chamber to be accurate to 20hz and below. A groundplane measurement outdoors is also a great way to measure as long as it's in a quiet area far from other boundaries (such as a parking lot). Either way, the idea is to remove the room from the equation and make the signal time as long as possible for an accurate measurement.
