DIY Omni Dodecahedron Speaker

[Ze Frog]

Interesting project. I have 40 Peerless 2.5" driver's knocking about, was originally going to build a CBT array, then got sidetracked.

Something like this could be a good use for them, only God knows what else I can do with that many of the things.

 

[Naughtius]

Interesting project. I have 40 Peerless 2.5" driver's knocking about, was originally going to build a CBT array, then got sidetracked.

Something like this could be a good use for them, only God knows what else I can do with that many of the things.

Stereo pair!

2.5" would be on the larger side, driver-spacing-interference wise. The smaller you'd be able to build them the better. Mine ended up around 130mm / 5" in diameter, and start "breaking" up at around 4kHz. They could have been smaller if I didn't want the extra volume for the low end.

 

[kemmler3D]

Although I agree it probably won't sound how you want (although I am very curious) it's just plain cool. Glad to see you are doing it even though we tried to talk you out of it.

Given that your cancellation problems are mostly in the treble, and you're not using this nearfield at all, and since it's a "true" omni speaker, I wonder if you can't just correct the total response with FIR and hammer it flat, whatever it ends up looking like. Usually you have to consider directivity carefully when you're EQing a DSP speaker, but in this case you've brute-forced your way past traditional considerations of directivity.

 

[Naughtius]

Although I agree it probably won't sound how you want (although I am very curious) it's just plain cool. Glad to see you are doing it even though we tried to talk you out of it.

Given that your cancellation problems are mostly in the treble, and you're not using this nearfield at all, and since it's a "true" omni speaker, I wonder if you can't just correct the total response with FIR and hammer it flat, whatever it ends up looking like. Usually you have to consider directivity carefully when you're EQing a DSP speaker, but in this case you've brute-forced your way past traditional considerations of directivity.

Thanks! I appreciate that - as well as the mutual curiosity towards how stupid this idea will end up being.

I'm still worried about the comb filtering messing things up when moving my head around, but I feel EQing the general / average frequency response should indeed be breezy. No crossovers to worry about, the 'reverb tonality' will automatically match the main response, and quite some surface area to help boost the low end.

Also - I just got mail:

Really quite happy with the general material / surface finish. Tolerances have been a tiny bit too big, but that does mean everything fits.

Next up I'll have to belt sand the corners of the drivers a bit - 48 in total, and melt in the 48 brass inserts. Hope to have something that makes noise in a couple of days.

 

[Salt]

 

[kemmler3D]

I'm still worried about the comb filtering messing things up when moving my head around,

As long as your room isn't some reflective hell*, I think this will only be noticeable close to the speaker itself.

Those parts look pretty good - where'd you get them?

*In my old bedroom years ago, I had some extremely noticeable comb filtering / nulls when playing tones.

 

[Naughtius]

As long as your room isn't some reflective hell*, I think this will only be noticeable close to the speaker itself.

Those parts look pretty good - where'd you get them?

*In my old bedroom years ago, I had some extremely noticeable comb filtering / nulls when playing tones.

They're MJF prints from JLCPCB, $140 including shipping and taxes for the ball and 14 covers.

Wouldn't the problem arise far field as well? Looking at the far field simulation on the first page of this thread there are clear peaks and dips across the largest part of the frequency range above ~5kHz. I'd imagine all of those to be audible.

 

[kemmler3D]

Wouldn't the problem arise far field as well? Looking at the far field simulation on the first page of this thread there are clear peaks and dips across the largest part of the frequency range above ~5kHz. I'd imagine all of those to be audible.

Maybe? I think it depends on the direct / reflected ratio where you're listening. While you'll still have the FR peaks and dips, the shape of the dispersion will tend to smear out quite a bit once the sound has a chance to bounce a few times. That's why I think FIR will work especially well in this case.

But yeah if you stand right next to it and play 5khz tones, I'd expect to hear all those peaks and dips.

Also, to be fair, that kind of thing is much easier to hear with tones than music or voice.

 

[Salt]

When You're finished with construction and installation, take microphone and do some measurements by moving around (repeat mode) and correspond to Your subjectiv impression.
Any correction should be adjusted afterwards, if necessary.

 

[Ze Frog]

Stereo pair!

2.5" would be on the larger side, driver-spacing-interference wise. The smaller you'd be able to build them the better. Mine ended up around 130mm / 5" in diameter, and start "breaking" up at around 4kHz. They could have been smaller if I didn't want the extra volume for the low end.

Yeah, doesn't sound like a good idea then. Plus it's something a bit beyond my capability to be honest. Maybe I'll get to building the CBT array one day.

 

[Ze Frog]

They're MJF prints from JLCPCB, $140 including shipping and taxes for the ball and 14 covers.

Wouldn't the problem arise far field as well? Looking at the far field simulation on the first page of this thread there are clear peaks and dips across the largest part of the frequency range above ~5kHz. I'd imagine all of those to be audible.

Can the comb filtering be negated somewhat by having different drivers in different positions having a slightly different output? Like in a CBT array? After all, original CBT design was actually half of a sphere.

 

[Naughtius]

Maybe? I think it depends on the direct / reflected ratio where you're listening. While you'll still have the FR peaks and dips, the shape of the dispersion will tend to smear out quite a bit once the sound has a chance to bounce a few times. That's why I think FIR will work especially well in this case.

What frequency range are you thinking of in the context of these reflections?

Build pics!

 

[Killingbeans]

SLS printed nylon?

 

[Naughtius]

SLS printed nylon?

MJF, I mentioned a couple messages up.

I have sound! I've been playing / measuring / roughly tuning it for the past hour - and it doesn't sound like I expected it would.

• High end - When playing pink noise and spinning the speaker around I can subtly hear 'whooshwhooshwhoosh' as it rotates through the speaker's funky HF radiation pattern, but when playing music it's barely audible, except for when I really pay attention to cymbals and the like.
  • This is a lot less of a problem than I expected
• Low end - It's pretty bad. I imagined I could just tune down the mid range / high end until it sounded balanced, up to the xmax of 2.0mm, but these drivers really aren't built to cleanly make any significant excursion.
  • I also had to switch from a 30W amplifier to a 200W amplified in the process, as it was clipping at the slightest bit of bass.
• Reflections - Looking at impulse response measurements, the floor and ceiling are making a big mess of things. Any non-near field measurements also look chopped up by an axe. Sound wise, it notice it less than I thought I would. The first reflections come after 6ms (at the position I've been listening from) - Linkwitz's magic number - so I think they're delayed enough not to impact the direct sound too much here. It mainly just sounds 'spacey'.
• Listening - I've been comparing it to a Bose SoundLink Revolve+ II that I borrowed, similar size, and also somewhat omni, though not vertically.
  • Bass is disappointing - easily sounds distorted. I'd need good limiters to prevent this from showing up at louder listening levels using it for music.
  • Voice clarity is pretty damn good, a lot better than the Bose. Happy that this ended up nicely, podcasts and audiobooks being its main purpose.
  • The spaciness... it has something nice. A charming quality. I'm tempted to experiment with this more.

Graphs:

I'll have to build a dedicated DSP / amp combo for it still, clean up the cables on the inside so they won't rattle, tape the covers on, hang it, and then it's onto the next project.

 

[Ze Frog]

MJF, I mentioned a couple messages up.

I have sound! I've been playing / measuring / roughly tuning it for the past hour - and it doesn't sound like I expected it would.

• High end - When playing pink noise and spinning the speaker around I can subtly hear 'whooshwhooshwhoosh' as it rotates through the speaker's funky HF radiation pattern, but when playing music it's barely audible, except for when I really pay attention to cymbals and the like.
  • This is a lot less of a problem than I expected
• Low end - It's pretty bad. I imagined I could just tune down the mid range / high end until it sounded balanced, up to the xmax of 2.0mm, but these drivers really aren't built to cleanly make any significant excursion.
  • I also had to switch from a 30W amplifier to a 200W amplified in the process, as it was clipping at the slightest bit of bass.
• Reflections - Looking at impulse response measurements, the floor and ceiling are making a big mess of things. Any non-near field measurements also look chopped up by an axe. Sound wise, it notice it less than I thought I would. The first reflections come after 6ms (at the position I've been listening from) - Linkwitz's magic number - so I think they're delayed enough not to impact the direct sound too much here. It mainly just sounds 'spacey'.
• Listening - I've been comparing it to a Bose SoundLink Revolve+ II that I borrowed, similar size, and also somewhat omni, though not vertically.
  • Bass is disappointing - easily sounds distorted. I'd need good limiters to prevent this from showing up at louder listening levels using it for music.
  • Voice clarity is pretty damn good, a lot better than the Bose. Happy that this ended up nicely, podcasts and audiobooks being its main purpose.
  • The spaciness... it has something nice. A charming quality. I'm tempted to experiment with this more.

Graphs:
View attachment 356346View attachment 356348

View attachment 356349View attachment 356353

View attachment 356351


I'll have to build a dedicated DSP / amp combo for it still, clean up the cables on the inside so they won't rattle, tape the covers on, hang it, and then it's onto the next project.

Impressive build, certainly something different.

 

[Phenoez8]

Love this project. Keep us posted on updates!

 

[D!sco]

Nice job! If you were to wrap a foam sphere around the speakers, it would look like a real lifestyle product.
This might be the perfect opportunity for a pair of sealed 6-8" corner speakers. They would handle the midbass quite well as well as add some needed sub.

 

[Phenoez8]

I am also worried about the comb filtering problem from multiple drivers playing high frequencies together. I am working on an omnidirectional speaker project myself, with a small 2-1/2" woofer and two passive radiators at the top and some microspeaker tweeters at the base. Seeing your results gave me more confidence that this could sound acceptable.

 

[kemmler3D]

Really cool! The in-room response looks surprisingly decent, but it doesn't look like it's happy playing below 200hz which is fine for voice, not great for music. I guess it's fit for purpose, though?

I'm also impressed at how fast you got this project done. Usually these DIY threads drag on for months or years...

What frequency range are you thinking of in the context of these reflections?

In my head this was mids on up... based on your posted measurements I may have confused myself about how it would play out.

 

[Naughtius]

I am also worried about the comb filtering problem from multiple drivers playing high frequencies together. I am working on an omnidirectional speaker project myself, with a small 2-1/2" woofer and two passive radiators at the top and some microspeaker tweeters at the base. Seeing your results gave me more confidence that this could sound acceptable.

Took me a bit to find the tweeters, those are small! If you send me a STEP file I could run a quick simulation of what the radiation pattern would look like across the frequency range. Their baffles might also influence it a bit at the higher frequencies. Nice design with the passives in there.