Mic in the Box

a4eaudio

I have not seen this done much - I think @DaveFred tried it sometime in the last few years - but there is a method that theoretically can get you very accurate low-frequency measurements. It is the Mic in the Box (MIB) method and is discussed HERE in AudioXpress and Andrew Jones talked about it back at InDIYana a long time ago. [ @PWRRYD - I saw an old post on PETT where I think you tried it but did not get useful results.]

I think in most cases, our normal merging of nearfield and farfield get us "close enough", but sometimes there may be some uncertainty. I know people have tried ground-plane measurements which I have heard is a real pain. The MIB should be as accurate. Where I think this would be particularly useful is in cases with multiple woofers and/or multiple ports - where it may be hard to get a good measurement of one woofer/port without the other one "bleeding" into it. (Or the case of a front-firing port close to the woofer.) The other advantage is that you don't have to do calculations of driver diameter and port diameter, etc. Although some programs (VituixCAD's merger tool) does the math for you, D'Appolito points out that that combining of the woofer and port can be pretty sensitive to small errors in diameter estimates.

So, I bought a Dayton IMM-6S just to try this, I figured it would be easier to put through the port and dangle inside the speaker than a full-sized mic.

I decided to try it out this week because I was curious about a certain measurement. In the thread where Geoff Millar is designing with the RS180P, I simulated the in box in WinISD and with a tuning of 52Hz got a pretty flat response down to about 80Hz, with only a slight hump of 0.5dB centered around 100Hz. I tend to find the box models quite accurate. When building the box, I cut down the port a little at a time, measuring the impedance and looking at the actual zma files each time rather than guessing where the minimum impedance between the two peaks was. So I know the tuning is very close to 52.0Hz.

When merging my nearfield port, nearfield woofer and farfield woofer I got a gradual dip of about 1.0-1.5dB below 100Hz. This isn't a big deal, as it is not going to make much difference in the overall crossover design. But as a starting point I wanted to figure out what SPL level I wanted to target for full baffle step compensation. If I target the level at 200Hz, it is about 2dB higher than at 90Hz, yet my simulation had suggested I would have a bump right here.

So, I dangled my little IMM-6S mic into the speaker and took a measurement at -30dbFS and hit cancel pretty quickly as levels indicated clipping right away. I tried -40dbFS and encountered the same thing. I succeeded at either -50 or -55dbFS. I accidentally deleted the original measurement, but it basically looks like a mountain range at the tuning frequency. The image included below is actually the nearfield of the port, but it looked similar. In REW all I had to do was select "All SPL" (yellow circle) and click on the "Actions" button (red circle) to bring up the "Actions for All SPL" (green circle). Here I selected "Measurement Actions", also circled in green.

This brings up the "Measurements actions" menu (circled in pink). I selected my measurement that I had named "MIB" in the drop-down box that includes all of the measurements that I have opened in REW (green circle). Then clicked the "Mic in box correction" button (circled in yellow).

This created the new measurement and added a comment "Mic in box correction from MIB referenced to 50.0 Hz". Teal curve below.

Note, just like a nearfield woofer measurement, this is a 2pi measurement so I need to take into account diffraction/baffle step loss. I already have a diffraction model for the baffle saved in VituixCAD so just jumped into the merger tool and added effect of baffle diffraction. Magenta line below. (Arbitrarily set to be equal at 150Hz in REW.)

And finally, I compared this to my merged result in the second image above. Merged farfield and nearfield in red, MIB in magenta. In the D'Appolito article he takes the average of the two. I am not actually going to change anything, but I would expect the MIB is more accurate.

6thplanet

Good info! Nice write up / explanation.👍🏻