3D Printed Acoustic Lens for Sound Dispersion

[Phenoez8]

Hey guys, I am trying to reduce the high frequency beaming effect on a pair of speakers and I found this solution. I found this cool paper on a semi-sperecal shape acoustic lens.

Result from paper. I'm not seeing anything even close to this in my experiment.

I modeled my own version and printed it (installed on a 4 inch full range Dayton driver), but I didn't quite get the resulted the authors get. In my experiments the speaker performs the same or slightly worse with the lens on at 45 degrees off axis.

This is a plot of [OnAxisFR / 45degOffAxisFR]. My measurement shows that there are some resonances from the lens, and it doesn't really help with the beaming, since the response at 45 degrees rolls off at a similar rate compared to no lens.

I have attached the original paper with this post. What do you all think about this design? Maybe I am doing something wrong here. Maybe doing some FEA analysis would also help. This would be really cool if it works!

 

[Keith_W]

Rather than 1/12 smoothing, can you try ERB smoothing? Also overlay the on-axis response with the two 45deg measurements, all with ERB smoothing. It might be easier to see.

Really interesting paper BTW, thanks for sharing it.

 

[thulle]

Really interesting paper BTW, thanks for sharing it.

+1

The printing process for this lens takes about two weeks of time on an upgraded, open-source Lulzbot TAZ-3 printer.

Yikes

 

[alex-z]

It seems the experiment was done with a standard woofer targeting a 3000Hz maximum frequency. If you are using one of the Dayton drivers with a phase plug that will change the outcome. In fact I would expect the acoustic lens to perform worse above 3000Hz than simply having the bare phase plug.

 

[Phenoez8]

You need to look at the difference normalized to the on-axis SPL, the acoustic lens is naturally going to change the on-axis performance.

Polar maps are one of the best ways to visualize this. Take your 0, 15, 30, and 45 degree data then use VituixCAD to get the visualization.

Also, it seems the experiment was done with a standard woofer targeting a 3000Hz maximum frequency. If you are using one of the Dayton drivers with a phase plug that will change the outcome. In fact I would expect the acoustic lens to perform worse above 3000Hz than simply having the bare phase plug.

This driver does have a phase plug. That is a very good point. I will perhaps try it on a different driver. Thank you for your feedback!

 

[Hayabusa]

This lense tries to disperse in 3D ? Would a 2D version not be better? (only horizontal, this would also be very nice for line arrays)

 

[Hayabusa]

To see the effect on dispersion we also need a 0 degree measurement.

 

[Phenoez8]

To see the effect on dispersion we also need a 0 degree measurement.

I should have been more clear. What I provided is the difference between the 0 degree and the 45 degree measurement, with and without the lens. I would follow up with the separate measurements later.

 

[thewas]

Would be interesting to see some measurements from 0° to at least 90° in 15" increments.

Similar widening is often done with phase plugs which can be quite advanced like in this case https://us.kef.com/blogs/news/the-sweet-sound-of-kef-s-tangerine-waveguide

 

[Hayabusa]

What distance did you measure?

 

[Phenoez8]

What distance did you measure?

Just at 10cm from the center of the driver with a UMIK-1.

 

[Hayabusa]

Just at 10cm from the center of the driver with a UMIK-1.

You should measure more in the far field, minimum 1 meter I would suggest

 

[terryforsythe]

I have attached the original paper with this post. What do you all think about this design?

I don't think the author of the paper provided enough frequency response measurements to make a fully informed decision. Plus, the driver he used was not very good; the author should have started with a driver that has a flat frequency response. See Fig. 13. Nonethless, the lens accentuated the peak around 580Hz. It did bring down the peak at 3kHz, but also added some choppiness. Based on what data is there, it doesn't look like something I would use.

 

[Keith_W]

You should measure more in the far field, minimum 1 meter I would suggest

Agree, further measurement distance is needed. How far depends on how wide the baffle is, minimum 2x baffle distance. 10cm is too close to document any difference.

I am excited now, can't wait to see what he finds when he measures at a proper distance!

 

[Phenoez8]

Thank you all for the feedback. I have found a 6-inch paper cone midrange driver and I am in the process of making a larger lens to fit this and a baffle to do some more tests. I will keep you all updated.