Re-mic’ing the NS-10s for the third time
Four weeks of A/B/X with three small-diaphragm condensers, two ribbons, and a $42 measurement mic. The measurement mic won.
I have re-mic’ed the NS-10s twice before and both times I gave up around the eight-hour mark and put the measurement mic back where it was. This time I made it four weeks. The result is the same. The result is also more interesting than I expected.
The setup: a pair of NS-10s on Auralex MoPADs, 1.2m apart, 0.9m from the listening position, on-axis at 1.4m. The room is treated front-and-back but the ceiling is bare. The system is calibrated to 79dB SPL pink at the listening position. None of this is unusual.
What I was actually trying to do
I have a recurring problem: a mix that translates to headphones, the car, and the kitchen Sonos, but is consistently 1–2dB hot in the 2–4kHz band when I open it the next morning. I assumed the room. I have spent eight years assuming the room.
This time I assumed the measurement.
The class-1 measurement mic I use is a Beyerdynamic MM–1 (the cheap one, not the new MM-1 II). Its on-axis response is flat to within ±1dB from 20Hz–8kHz. Above 8kHz, it rolls off about 2dB by 12kHz and another 3dB by 16kHz. I knew this. I had ignored it because the room’s problems are all below 400Hz.1
The A/B/X protocol
Five microphones, in randomised order, blind. Each mic placed at the listening position pointing at the left speaker only. 30-second sine sweeps, 20Hz–20kHz, log. Averaged across 8 sweeps per mic. Captured at 96kHz/24-bit through a Lynx Aurora(n).
- Beyerdynamic MM–1 — the current reference.
- Earthworks M30 — borrowed for two weeks.
- iSEMcon EMX–7150 — the $42 measurement mic.
- Royer R–121 — ribbon, included as a control.
- AKG C414 (cardioid) — included as a control.
The two condensers and the ribbon were not seriously expected to be useful for measurement — they were included so I could see what “wrong” looks like in this protocol.
The result
The iSEMcon EMX-7150 had the flattest response from 2kHz to 18kHz of any mic in the room. Including the Earthworks. The MM–1’s 2–4kHz region was 1.4dB hotter than the iSEMcon’s, with a 0.6dB ripple I had been reading as the room.2
In other words: I have been correcting a measurement error for eight years. The room was telling the truth. The mic was not.
What I changed
I re-ran Sonarworks SoundID with the iSEMcon, fed the new correction curve into the JBL 7 Series, and re-balanced the analog trim. The translation problem went away in 90 minutes. I did not believe it. I let four mixes sit a week before I wrote this up.
The takeaway is not “cheap mic is better.” The takeaway is that I had been using a mic outside the band it was specified for, and trusting the spec sheet. The $42 mic was honest about its limits and happened to be flatter inside them.