Testing the ISO 18233 Approach to Transmission Loss-Experiment using REW

Matthew J Poes

Matthew J Poes

Senior Audioholic
my friend @shadyJ has been teasing that this has been my white whale. I have been trying to measure transmission loss using the old and the S/N ratio of this method was so poor that I couldn't play back tones loud enough (safely enough) to get an accurate measurement of TL. I had tried using the ISO 18233 method, using a Sine Sweep FFT measurement in each room, before, but got errors. I didn't (at the time) realize this was something others had done, so when it arose recently and I had a chance to talk this out with an expert and John Mulcahey, I figured out a method that did in fact work.

The gist of the approach is to use REW to capture a sine swept IR in each room and divide these to create the TL transfer function. You do this by measuring the IR of the source room (Theater) and receiver room (Room outside the theater). Ideally you would take many IR's around the wall at different heights and locations and average them, then deviide the averages. This is time consuming, so the measurement you will see is based on a single measurement on each side. I did test the average approach, but ran into some problems I didn't have time to trouble search.

Here is an example of the old method. The SPL levels are wrong, I calibrated the mic wrong, so add about 10dB's to know what they really were. Also keep in mind, the source (theater) room measurement had a mic about 25 feet from the speaker, so remember that when judging how loud all this is. It's hard to get most systems to play back loud enough and safely enough given the conditions.
TL Traditional.jpg
As you can see, the measurement with the RTA on the receiver side was mostly at the same level as the room's noise floor. Pretty much useless. That isn't a measurement of the walls TL.

Matt TL 2 Test.jpg
Now here is the Sine Sweep IR method, while the level is louder, the IR was the entire solution to the problem here. I even successfully took a measurement at 80dB's and still got relatively accurate results up to 3khz. This level was necessary to get accurate results above 3khz and to be certain I wasn't measuring noise.

STC Curve 64.PNG
Here is a calculation of STC from the above IR method, which comes in at around STC 64. The wall was built to a higher standard than that, but I believe this difference is everything from some measurement errors to normal and expected leaks in the construction that compromised the STC. I need to retry all of this by spatially averaging many IR's.

Now for those who want to try this themselves, I would just message me. It's a little complicated. Also, I'm not totally sure it can be done with a USB measurement mic, It's possible the noise floor is too high to get accurate results.

I posted this here also because I thought some folks like @gene with a more technical background might find it interesting. Doing measurements like this of this sophistication by more average joe's isn't common, so I always find it exciting when a simple solution is found that let's us do things that used to cost a fortune.

  • RBHsound.com
  • BlueJeansCable.com
  • SVS Sound Subwoofers
  • Experience the Martin Logan Montis