That is the "lead out". Earlier you seemed to already understand that the NG format encodes the lead out in the header rather than encoding it at the end where it logically fits.


Both regular and NG formats encode the wavelength of the carrier, rather than the frequency. So a normal duration can be computed by multiplying the duration number (the 15 or 16 or 40 etc.) by the wavelength and multiplying that by the unit time used to encode wavelengths (which is a different unit between old Prontos vs. NG, but the concept is the same).

But the NG encodes longer values differently. They don't depend on the wavelength. For some of them I think it multiplies the unit time by 256 and multiplies that by the number encoded after the B. That makes the relationship to the value in the Pronto Hex harder to see.

To further confuse things, I think the number of bits after the B varies and the multiplier (that 256) varies. You also should understand that those multiple bit fields need to be rereversed to form a number (the low bit fields come before the high bit fields).

In theory you could paste similar Pronto Hex in a few times varying just the size a one large gap in the middle of the signal and see how the XML changes. (In fact I did that). But that method runs into some PENG bugs and gives incorrect results. To really get the answer you would need to learn from the NG remote to some reliable IR learning system and see the actual gap lengths. I don't own any Pronto (NG or otherwise) so I never did that).