Alright, you know how the saying goes - measure twice, cut once. Well, this is similar. With the UMIK connected to my noisy PC I couldn't make out the intermodulation distortion in the FFT, but with my laptop I can. They're still very low in level - the loudest tone is at -9db and another one at 6kHz at -10db, which should be well below the threshold of hearing, so I'm not entirely convinced that that's what I was hearing, but it might be more likely than hearing 30kHz at 80db. Note that to get the noise this low I put the mic as close to the driver as I could (less than 10cm from the voice coil). I measured the FR at that position and at my testing position 5m away from the driver and lowered the level accordingly (18db). If anything the tones are even lower in level than this, but the FR is close enough in the range of interest. The tones are about 80-85db lower in level than the 30kHz tone should be. I might add some -70dbfs white noise to try to mask the IMD and see if I can still hear something. Or I could try to cancel out each of the three tones. Here's the FFT: 1M Length and a couple averages, so a few minutes per each measurement.