That's almost 10 dB less roll-off than on your previous graph.
I still think that it might be the coupler itself (perhaps Luke is now using different microphone drivers in the couplers?), however to rule it out, it would be best to lower the number of averages (which minimises the chance to lose the seal during the measurement process) and to use sine sweeps instead of noise for the measurements.
I still think that it might be the coupler itself (perhaps Luke is now using different microphone drivers in the couplers?), however to rule it out, it would be best to lower the number of averages (which minimises the chance to lose the seal during the measurement process) and to use sine sweeps instead of noise for the measurements.