Sphere shaped speakers with CHN110 drivers

[ppataki]

I got inspired by an earlier thread here and decided to build my own sphere shaped speakers too
The idea is the same, I will be using the largest Ikea Blanda Matt bamboo salad bowls (having an outer diameter of 28cm)

The drivers will be Markaudio CHN110 full range models (I have used these beauties in a couple of earlier projects, they are one of the best full range drivers in this size)

Here is approximately how the cabinet will look like:

The xmax limited max input power will be approx. 150W at 80Hz and that will yield 106.4dB per speaker at 1m
(they will be crossed with the sub at 80Hz)

I will post more pictures as we proceed with the build with my neighbor

 

[norman bates]

Baffle step ?

Anything knocking down the internal standing wave in a sphere ?

 

[ppataki]

Baffle step ?

Do you mean baffle step compensation?
There will be heavy DSP optimization in the main listening position so I guess that should mitigate that

Anything knocking down the internal standing wave in a sphere ?

On the inside there will be one layer of 4mm thick vibrodamping sheet then 10mm thick felt and then melamine foam filling

 

[Tangband]

I got inspired by an earlier thread here and decided to build my own sphere shaped speakers too
The idea is the same, I will be using the largest Ikea Blanda Matt bamboo salad bowls (having an outer diameter of 28cm)

View attachment 348305


View attachment 348306

The drivers will be Markaudio CHN110 full range models (I have used these beauties in a couple of earlier projects, they are one of the best full range drivers in this size)

View attachment 348307

Here is approximately how the cabinet will look like:

View attachment 348310

The xmax limited max input power will be approx. 150W at 80Hz and that will yield 106.4dB per speaker at 1m
(they will be crossed with the sub at 80Hz)

View attachment 348313


I will post more pictures as we proceed with the build with my neighbor

Very interesting !

 

[ppataki]

Started treating the back-facing hemispheres

4mm thick vibrodamping sheet:

Then 10mm thick felt:

Then finally some melamine foam:

Next step is to make the cutout on the front-facing hemispheres then treat those as well then finally put together the two

Hopefully next week I will be able to finalize both speakers

 

[ppataki]

Just for fun: if you drop a melamine foam it falls kind of like an umbrella
It really seems to 'slow down the air'
I guess this shows that it would be an excellent sound absorption material

New video by Peter Pataki

photos.app.goo.gl

 

[ernestcarl]

When you’re done try measuring the on- and off-axis curves, say, 10-15 degree increments up to at least 60 degrees. It would be interesting to see the effect of equalization applied to all curves simultaneously as well.

 

[ppataki]

I have never done that before, but I will give it a try this time
Moreover, it would be great to create something like this for these speakers and for my current 15" Fane speakers too:

One moment, please...

(to compare them)

I would expect the smaller CHN110s to have a better polar plot vs the big Fanes but we shall see

Any watchouts/suggestions for doing these measurements/settings in REW?

 

[norman bates]

Very nice

 

[ernestcarl]

Any watchouts/suggestions for doing these measurements/settings in REW?

If you're using a USB microphone nearfield, better perform short (e.g. 128k-256k) repeated measurements. A timing reference shouldn't be absolutely necessary. Discard "bad looking" IRs, 'cross correlate align', then vector average and/or RMS average -- the vector avg may be useful if you want to also examine the time domain info.

In room, position the speaker well away from all boundaries as much as possible as they will influence the measurements. You could set the desired angle increments around the the speaker using a tape and marker beforehand or use a protractor ruler (use an extension guide for longer distances).

Compromised in-room:


More ideal is measuring in a BIG room and with a high stand:

Same speaker measured in-room without any windowing/gating ~30 cm distance in middle of the room:

Gating (5.5 ms) applied with/out manually created and semi-automatically generated inverse filters:


In-room moving microphone measurements (MMM) with/out semi-automated inverse "speaker" filter applied:


I prefer gentler "manually created" filters (say, the 2 or 7 PEQ examples above) for this 2-way coax speaker which has already been DSP equalized. You also probably should not completely boost some of the much deeper dips (to avoid increasing distortion at higher SPL levels) -- note that you can't really "correct" for diffraction errors.

 

[ppataki]

If you're using a USB microphone nearfield, better perform short (e.g. 128k-256k) repeated measurements. A timing reference shouldn't be absolutely necessary. Discard "bad looking" IRs, 'cross correlate align', then vector average and/or RMS average -- the vector avg may be useful if you want to also examine the time domain info.

In room, position the speaker well away from all boundaries as much as possible as they will influence the measurements. You could set the desired angle increments around the the speaker using a tape and marker beforehand or use a compass ruler (use an extension guide for longer distances).

Compromised in-room:
View attachment 348924 View attachment 348925 View attachment 348926

More ideal is measuring in a BIG room and with a high stand:
View attachment 348927

Same speaker measured in-room without any windowing/gating ~30 cm distance in middle of the room:
View attachment 348940

Gating (5.5 ms) applied with/out manually created and semi-automatically generated inverse filters:
View attachment 348941 View attachment 348942 View attachment 348943 View attachment 348944 View attachment 348945 View attachment 348946

In-room moving microphone measurements (MMM) with/out semi-automated inverse "speaker" filter applied:
View attachment 348947 View attachment 348948

I prefer gentler "manually created" filters (say, the 2 or 7 PEQ examples above) for this 2-way coax speaker which has already been DSP equalized. You also probably should not completely boost some of the much deeper dips (to avoid increasing distortion at higher SPL levels) -- note that you can't really "correct" for diffraction errors.

Thanks @ernestcarl for the details, appreciated
I was wondering: since I always optimize for the MLP (and I don't intend to change that), shouldn't I perform these 0-15-30-45-60 degree measurements at the MLP?
First with zero DSP and then with full DSP?

 

[ernestcarl]

Thanks @ernestcarl for the details, appreciated
I was wondering: since I always optimize for the MLP (and I don't intend to change that), shouldn't I perform these 0-15-30-45-60 degree measurements at the MLP?
First with zero DSP and then with full DSP?

It’s a sphere so you could even measure just one side. The only major difference might be the vertical where it meets the stand/ground. The angle increments could be anything you want, but you probably only need a small window that does not veer too far off the central position.

The DSP is the the result from a prediction so you can just convolve your EQ filter with the original uncorrected curves using the A x B arithmetic function. Almost no reason to re-measure.

*The first set were done nearfield, the second in-room at the MLP could also work if using MMM or single-point but you will get a mix of the room —more so with a single point measurement even if windowed/gated.

 

[MAB]

I have never done that before, but I will give it a try this time
Moreover, it would be great to create something like this for these speakers and for my current 15" Fane speakers too:

One moment, please...

(to compare them)

I would expect the smaller CHN110s to have a better polar plot vs the big Fanes but we shall see

Any watchouts/suggestions for doing these measurements/settings in REW?

Napilopez has a turtorial, you can actually make a full spinorama if you want!

How to make quasi-anechoic speaker measurements/spinoramas with REW and VituixCAD

Quasi-anechoic measurements are a way to capture the 'real' frequency response of a speaker without the influence of a room. They are called 'quasi-anechoic' because they have the potential to get you really close to the results one might achieve with a non-reflective environment like an...

www.audiosciencereview.com

I use a camera tripod with a sturdy rotating head for small speakers:

And a stool that rotates for larger speakers:

Agree with ernestcarl, would be good to see off-axis measurements on these!

Looks like a fun project!
Agree with your comments on melamine foam, I have wondered about using it for enclosure damping.

 

[ppataki]

Some progress in the meantime:

 

[Tangband]

Some progress in the meantime:

View attachment 349259

View attachment 349260

View attachment 349261

Very nice ! Did you secure the two bowls with wood glue and was is easy ( no woodwork needed )?

 

[DanielT]

Good job ppataki. Damn good looking globe speakers I must say AND anyone can go to the nearest IKEA and buy a couple of these wooden bowls and do the same thing. Inspiring.

AND another interesting thing. These ball speakers should spread the sound in a different way than a speaker with a flat baffle, especially compared to a large wide flat baffle, so it will be fun to test different sound structures in your listening room. My suspicion, but now I'm just guessing, that in a very sparsely furnished room with few furniture and few carpets, ball speakers don't fit. There will be too many reflections, too smeared sound. Okay, it's a matter of taste so the only thing you can do is test and see what you like.

Here, a Swedish speaker designer, Olle Mirsch, and his last (I think) project, the Rondo speakers. Stylish, futuristic I must say.

FR is measured in that graph. I translate what the text under the graph :

"In simulated free-field measurement straight from the front exhibits Rondo a very straight and gently falling frequency response.The treble drop is a deliberate choice to compensate the treble rise that Rondo due to his construction provides in the listening room. 60° from the side, the frequency curve looks almost identical!"

https://www.google.com/url?sa=t&source=web&rct=j&opi=89978449&url=https://www.hifigoteborg.se/pdf/Mirsch_Rondo_test.pdf%3Ffbclid%3DIwAR08zOVvsuLMjI-xi1R35-2omvpu9zmhEJWRHPhUXLVXY3JD8PO5ABOWows&ved=2ahUKEwi2paGZ7qeEAxWQFBAIHQCRC_UQFnoECBUQAQ&usg=AOvVaw0qX_ARroNxo_2m1nkPzVSN

 

[ppataki]

Very nice ! Did you secure the two bowls with wood glue and was is easy ( no woodwork needed )?

No, that is yet to come
I am planning to use wood glue - fingers crossed

Now I will be finishing the insides then do the paint-job and then will try to put the two sides together
I really love the color of the bowls but unfortunately they will not match any of our rooms in the house so I am planning to use a special paint (have no idea how it is called in English) that will keep the texture of the bamboo and will only change the color (to white)

 

[ppataki]

My suspicion, but now I'm just guessing, that in a very sparsely furnished room with few furniture and few carpets, ball speakers don't fit. There will be too many reflections, too smeared sound.

FWIW, my experience in the last 4 years with full range drivers is that since they beam like hell they tend to have near-zero reflections
You can very clearly see this in the IR graph if you measure a full range driver (you can also check out my other projects' measurements)
In the same room/same position they have much less reflections compared to a multiway speaker
The bigger the full range driver is the more it beams the less reflections it will create

 

[DanielT]

FWIW, my experience in the last 4 years with full range drivers is that since they beam like hell they tend to have near-zero reflections
You can very clearly see this in the IR graph if you measure a full range driver (you can also check out my other projects' measurements)
In the same room/same position they have much less reflections compared to a multiway speaker
The bigger the full range driver is the more it beams the less reflections it will create

Sorry, I was thinking about those Rondo speakers. They are two-way speakers.

What you say sounds reasonable given the drivers you are using.

How do you think they sound?

 

[ppataki]

How do you think they sound?

I had used the CHN110 drivers in two previous projects (both are published in the DIY section)
They are one of the very few full range drivers that sounds pretty good even with zero DSP (you can listen to the recordings I have included in those other projects)
When you correct them they will sound unbelievably great - ask @Tangband about the comparison to some Genelecs