[Cuki79]

You are looking in linear scale. The tail looks insignificant but it is not. Also you can clearly see a pseudo period corresponding to the main low frequency resonance. (About 5ms which is 200Hz)

You don’t hear the difference because:
long IR only matters for the low end. If you don’t focus on the low end you might not notice. Remember long IR enables to resolve long periods which correspond to short frequency. For example 100 Hz has a period of 0.01=10ms. With a 40ms IR you basically allow 100Hz to ring 4 times before vanishing…

Most people are deaf at 100Hz.

The more you blend the closer to the pickup. My goal is to alter the most the pickup without actually increase too much the actual impulse response of the system.

When you use FIR filter, you basically trig the IR. In order to have a resonance at 100 Hz, you need 40ms Tail with those 4 periods. It means that every time you hit the note, these 4 periods will be added

I think it does not work like that in reality. In reality, my guess is that the top and the air inside the box store mechanical energy. Like the swing on the play ground. As they are excited they accumulate more energy that enable some frequencies to ring longer. That « memory » of the energy stored can not be modelled by a FIR filter without adding unrealistic tail. That is why Fishman, from day one, used IIR filters to model the low end.

Note that however we join somehow at some point. The more you let IIR do the job (EQ) the flatter the Short IR frequency response will be.

PS: I am not advising against « blending »: to this day it is the best way to use our IR in a Live context. I am just trying another path.
PS2: Jon when one blends interferences might occur between the IR convoluted and original signals. Check the spectrum for 10 blend settings. Peaks or notches can appear in your spectrum when the two contribution are balanced in amplitude and opposed in phase.