All three mics tested in these two reviews show 2nd harmonic that falls with SPL down to 100 dB where the noise specified at about 31 dB takes over. This makes perfect sense qualitatively, even if I'm still struggling to understand it quantitatively. How can it match the noise spec almost perfectly when the analyzer will only allow a tiny slice of the spectrum at 2 kHz +/- a few Hz?
However, the 3rd harmonic goes through a minimum of about -90 dB at 120 dB and then goes up again as SPL falls further, rising about 10 dB for every 10 dB drop in SPL. This does not look noise-related, as the rise begins at much higher SPL than for the 2nd harmonic.
https://www.isemcon.com/datasheets/iSEMcon EMX-7150.pdf
https://www.isemcon.com/datasheets/iSEMcon EMM-13D082.pdf
What is going on here? What nonlinearity mechanism is there (other than noise) that actually increases as the signal goes down? Mechanical hysteresis in the membrane? Sinister storage effects in the electret? Or is there a more mundane explanation?
However, the 3rd harmonic goes through a minimum of about -90 dB at 120 dB and then goes up again as SPL falls further, rising about 10 dB for every 10 dB drop in SPL. This does not look noise-related, as the rise begins at much higher SPL than for the 2nd harmonic.
https://www.isemcon.com/datasheets/iSEMcon EMX-7150.pdf
https://www.isemcon.com/datasheets/iSEMcon EMM-13D082.pdf
What is going on here? What nonlinearity mechanism is there (other than noise) that actually increases as the signal goes down? Mechanical hysteresis in the membrane? Sinister storage effects in the electret? Or is there a more mundane explanation?
Attachments
Last edited:
This is from the EMX-7150 data sheet. Note that this is just THD, so no information about HD2 and 3.
Seems to approach 0.1% = -60 dB THD at 110 dB but shows no tendency to rise towards smaller SPL and is actually a tad worse than the -65 dB THD given in the previous graph. Strangely, the B&K reference mic shows a shallow minimum from 120 to 130 dB with a small rise towards lower SPL, ending up at the same -60 dB at 110 dB.
Seems to approach 0.1% = -60 dB THD at 110 dB but shows no tendency to rise towards smaller SPL and is actually a tad worse than the -65 dB THD given in the previous graph. Strangely, the B&K reference mic shows a shallow minimum from 120 to 130 dB with a small rise towards lower SPL, ending up at the same -60 dB at 110 dB.
Attachments
Last edited:
Effective noise level may be slightly different for H2 and H3 in the measurement setup used for whatever reason, but I have little doubt that the initial decrease is entirely down to things being noise-limited. If you were to plot the absolute SPL of distortion products instead, things would start out flat until they begin to emerge from the noise floor.
The mic seems to exhibit textbook small signal behavior (described by IP2/IP3) until it begins to diverge around 146-147 dB SPL. Nominal max SPL spec would be 136 dB (1%) / 130 dB (0.5% THD). The relatively high level of H2 suggests that the electronics are the limiting factor.
The mic seems to exhibit textbook small signal behavior (described by IP2/IP3) until it begins to diverge around 146-147 dB SPL. Nominal max SPL spec would be 136 dB (1%) / 130 dB (0.5% THD). The relatively high level of H2 suggests that the electronics are the limiting factor.
I struggle to understand how noise at 3 kHz can be a whopping 20 dB different than at 2 kHz (assuming that a pretty narrow filter is used around 2 resp. 3 kHz to determine the harmonics). Maybe something strange going on in the analyzer?
The two measurements are in good agreement at 140 and 145 dB, wherease at 150 dB, ProductionPartner magazine reports -25 dB = 5.5% and continuing slope out to 160 dB wheras the data sheet just goes off the chart at 146 dB. Did they get improved electronics?
BTW, what does IP2/IP3 stand for?
Similar thing going on with the review of the M50:
Fresh From the Bench: Earthworks M50 Omnidirectional Measurement Microphone | audioXpress
Appears to saturate at 0.17% = - 55 dB at 100 dB, with a hint of a further rise towards lower SPL. Again, the graph reads THD rather than THD+N, so noise should be largely excluded. Noise is 22 dB(A) which should be about 30 dB which seem to be too little to explain a 45 dB gap at 100 dB even if the analyzer was wide open. I don't think the distortion of the source was subtracted in this case.
The two measurements are in good agreement at 140 and 145 dB, wherease at 150 dB, ProductionPartner magazine reports -25 dB = 5.5% and continuing slope out to 160 dB wheras the data sheet just goes off the chart at 146 dB. Did they get improved electronics?
BTW, what does IP2/IP3 stand for?
Similar thing going on with the review of the M50:
Fresh From the Bench: Earthworks M50 Omnidirectional Measurement Microphone | audioXpress
Appears to saturate at 0.17% = - 55 dB at 100 dB, with a hint of a further rise towards lower SPL. Again, the graph reads THD rather than THD+N, so noise should be largely excluded. Noise is 22 dB(A) which should be about 30 dB which seem to be too little to explain a 45 dB gap at 100 dB even if the analyzer was wide open. I don't think the distortion of the source was subtracted in this case.
Third-order intercept point - Wikipedia
These graphs look like amplitude chirp tests at a guess. You need a long long chirp for both good power resolution and good frequency resolution - in other words the bandwidths for the harmonic detection will be wide to allow the chirp to progress across the range of powers quickly.
What transducer has low enough distortion to show the Mike distortion? I would want a lot more on those measurements to accept them. It's a difficult problem to check for acoustic distortion and even air nonlinearity will limit you.
Perhaps the distortion from the DUT is cancelling the analyzer residual, so the reading reaches a minimum at some magic level where the residual and DUT distortion are equal but of opposite polarity?
Regarding my own measurements with small capsules and MEMS-microphone the increase of the noise in the low frequency region is caused by the lower sensitvity in this region. The increase in the higher freqs is the known issue of these delta-sigmas used inside them.