A couple of weeks ago I post the measurements of the Monoprice THX-365IW. For that, I built a leaky back wall in an attempt to approximate the actual application. Alas, that provided such messy measurements that I thought I start over using a new methodology/system. Klippel NFS has a module for measuring drivers in a baffle. It is a combination of a special baffle and different measurements system where only the surface of the baffle is scanned. I asked for an evaluation license for the module which normally costs more than 4,000 Euros. The baffle is even more expensive so I decided to build my own. First attempts were total failures as the module assumes you have their fixture and its special alignment points. I made a second try and almost succeeded:
I say almost as the software performs an alignment check and complained that accuracy was not as good as it should be. It was right: it pushed the microphone into the baffle at some point.
Still, the results are much cleaner and I thought I share this with you and make a collective decision if we need to keep going this way.
As a way of background, Klippel NFS scanner makes measurements in a semi-hemisphere. This way, it doesn't matter what radiates in the back. To the extent any of that reflects from the walls, NFS computationally eliminates it. Furthermore, since edge diffraction occurs outside of the scanned area, it is able to eliminate that as well. Fort his to work though it requires a lot of distance between the "driver" and the sides of the baffle and hence the wide width you see. The width is also dictated by the height of the in-wall speaker which is naturally much taller than what Klippel had in mind with a single round driver.
Measuring and cutting the baffle was one thing. Getting it to mount and stay stable on a 5 inch stand was another! In addition to this, you have to be mindful of any resonances in the baffle being captured and reported as part of the speaker response. You can't see the back side of my baffle but I managed to deal with these effectively as well.
Monoprice THX-365IW Measurements
Before I show you the spin graph, let's view the radiation pattern at 1 kHz so you get oriented correctly on how we are measuring:
This would be looking into the left side of the baffle. All energy is naturally directed forward.
Here is our spin graph now:
Remarkably clean, isn't it? We just have one resonance around 3.8 kHz or so but otherwise, speaker is doing well as is the measurement scheme.
I should note that this is a much higher resolution measurement than normal by accident. I was playing around and forgot to change the chirp signal to what I normally use. So if you are processing the exported data, be mindful of this.
Back to our spin, we see perfect directivity because there is no back radiation and what is facing forward is quite broad. You can see it better in beam width:
And horizontal directivity:
Note that we are now correctly bound by ±90 degrees due to measurement scheme and how the speaker is used.
Early window response now puts the "rear wall" response right on top of the rest:
Predicted in-room response is very nice as well:
Of course the math for this is for normal speakers in rooms. The angles may need to be adjusted. I welcome our mathematicians suggesting what should be changed.
Back to directivity, here is our vertical:
Grill or no Grill; That is the Question
For these measurements, I decided to not use the grill as to get the core measurements right and not worry about its effects. Once done though, I went back and made a couple of in-room measurements with and without the metal grill:
As expected, the effect is small. What I don't know however is whether it will impact NFS more in the way it causes diffractions.
Conclusions
The new measurement system gives us a much more close to real life results. The cost is an afternoon to build a precision baffle for every in-wall speaker to measure and thousands of dollars in software licensing. I am game pursuing this unless someone sees any serious issues here which I don't.
------------
As always, questions, comments, recommendations, etc. are welcome.
Appreciate any donations using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/
I say almost as the software performs an alignment check and complained that accuracy was not as good as it should be. It was right: it pushed the microphone into the baffle at some point.
Still, the results are much cleaner and I thought I share this with you and make a collective decision if we need to keep going this way.As a way of background, Klippel NFS scanner makes measurements in a semi-hemisphere. This way, it doesn't matter what radiates in the back. To the extent any of that reflects from the walls, NFS computationally eliminates it. Furthermore, since edge diffraction occurs outside of the scanned area, it is able to eliminate that as well. Fort his to work though it requires a lot of distance between the "driver" and the sides of the baffle and hence the wide width you see. The width is also dictated by the height of the in-wall speaker which is naturally much taller than what Klippel had in mind with a single round driver.
Measuring and cutting the baffle was one thing. Getting it to mount and stay stable on a 5 inch stand was another! In addition to this, you have to be mindful of any resonances in the baffle being captured and reported as part of the speaker response. You can't see the back side of my baffle but I managed to deal with these effectively as well.
Monoprice THX-365IW Measurements
Before I show you the spin graph, let's view the radiation pattern at 1 kHz so you get oriented correctly on how we are measuring:
This would be looking into the left side of the baffle. All energy is naturally directed forward.
Here is our spin graph now:
Remarkably clean, isn't it? We just have one resonance around 3.8 kHz or so but otherwise, speaker is doing well as is the measurement scheme.
I should note that this is a much higher resolution measurement than normal by accident. I was playing around and forgot to change the chirp signal to what I normally use. So if you are processing the exported data, be mindful of this.
Back to our spin, we see perfect directivity because there is no back radiation and what is facing forward is quite broad. You can see it better in beam width:
And horizontal directivity:
Note that we are now correctly bound by ±90 degrees due to measurement scheme and how the speaker is used.
Early window response now puts the "rear wall" response right on top of the rest:
Predicted in-room response is very nice as well:
Of course the math for this is for normal speakers in rooms. The angles may need to be adjusted. I welcome our mathematicians suggesting what should be changed.
Back to directivity, here is our vertical:
Grill or no Grill; That is the Question
For these measurements, I decided to not use the grill as to get the core measurements right and not worry about its effects. Once done though, I went back and made a couple of in-room measurements with and without the metal grill:
As expected, the effect is small. What I don't know however is whether it will impact NFS more in the way it causes diffractions.
Conclusions
The new measurement system gives us a much more close to real life results. The cost is an afternoon to build a precision baffle for every in-wall speaker to measure and thousands of dollars in software licensing. I am game pursuing this unless someone sees any serious issues here which I don't.
------------
As always, questions, comments, recommendations, etc. are welcome.
Appreciate any donations using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/
