Don`t forget to give melamine foam a try. Its easy to cut to anything from thin slices to plugs with a hole punch. And its more repeatable than making a ball from polyfill.
And since I know you are from Sweden, https://www.clasohlson.com/se/Fläcksuddare-4-pack/p/34-6395?gad_source=1&gclid=CjwKCAiAibeuBhAAEiwAiXBoJLq18P2ZBKym8VG-FFcdEZDFuq47fofmhEBtXaVaXx99ZSro7oxZ0BoCXMQQAvD_BwE
And the real deal can be bought thru amazon or:
https://www.skumacoustics.com/gb/3-acoustic-foam
And the info
https://plastics-rubber.basf.com/global/en/performance_polymers/products/basotect
Haha yeah, I'm aware that all my "discoveries" have been done before. For me this is mostly a fun exploration. I really enjoy experimenting and learning the different aspects of the design. And hopefully one day I’ll have a MEH I’m truly happy with.
That’s an excellent tip! I have some of those at home.
I’ve been working on a new version where I place the mid ports further out in the horn. I experimented with changing different parameters in Hornresp, figuring out a design that would affect the CD’s response as little as possible. The mids’ front chamber is much larger and wider than in previous designs. I decided to go with a circular port directly in front of the mid woofer, as I believe that would give the best low response.
Here is the Hornresp sim:
Good idea 🙂 Seems like You are making use of a reflection inside the front chamber, is this to have a more steep cut-off or to kill a resonance?
Exactly! I'm hoping the front chamber cancels out most of the output above the crossover frequency of 1000hz. It'll be interesting to see if the measurements align with the simulation.
I don't think you can get away with having the mid ports so far down the horn. If within the 1/4 wave spacing, the 3 sources - CD+mids combine into a single source. Outside that distance, you will need to model/calculate the response as separate sources. Doing that, you will see the directivity departing from that given by the horn shape. At 90 degrees off axis, for example, you should see a null at the frequency where the horizontal distance between midports is 180 degrees. Short of 90 degrees, you will see lesser cancellation....but still enough to be of concern.
It occurs to me that if you rotate the horn 90 degrees so the mids are on top and bottom instead of the sides, the effect is much easier to live with because you never listen very far off the vertical axis.
It occurs to me that if you rotate the horn 90 degrees so the mids are on top and bottom instead of the sides, the effect is much easier to live with because you never listen very far off the vertical axis.
I think only measurements can give an answer to how strong the (notch) reflection from the throat/CD is on a particular horn profile-CD combination. Some claims to have no reflections
others use EQ on an unwanted weak reflection. Maybe looking at the horn's acoustic impedance vs freq can give a clue?
I think it was mark100 who presented the phase wheel (post #76) when discussing port distance on a MEH, and Danley mentions 1/4-1/3 wL. Again, I think only measurements can give the correct answer. Driver placement should depend on what drivers are used(cone sag).
others use EQ on an unwanted weak reflection. Maybe looking at the horn's acoustic impedance vs freq can give a clue?
I think it was mark100 who presented the phase wheel (post #76) when discussing port distance on a MEH, and Danley mentions 1/4-1/3 wL. Again, I think only measurements can give the correct answer. Driver placement should depend on what drivers are used(cone sag).
Exactly! My theory is that if you bring the resonant frequencies below the frequency range of the compression driver, the mid ports/chambers will have minimal interruptions on the CD’s response.
That’s what I’m trying to find out with this newest version of the horn. Danley says that the mids and CD must be within 1/4 wavelength of each other, so you might be right. But I have also heard that the phantom center of the CD can be further out in the horn, not inside the driver itself. So if that’s true, then this might work. And if it doesn’t work, I’ll print another horn part with a shorter throat where the mid ports are exactly 1/4 wavelength from the CD’s entry and see if that gives a better result.
I don't know how much of this is applicable for a MEH, but from the figure on page 6 it looks like the distance between mid ports can be anything below 1/3wL
https://www.linkwitzlab.com/Horbach-Keele Presentation Part 2 V4.pdf
https://www.linkwitzlab.com/Horbach-Keele Presentation Part 2 V4.pdf
In couple of topics is discussed that mid notch is on higher freq than 1/4 wl calculation gave. Most plausible explanation is that horn aperture become to small in relation to wl, so sound waves bounce back earlier, making the distance apparently shorter.
Conical horns (like Danleys) have very rapid changing of flare rate (fast opening), so this effect is not so much pronounced. With ATH types horns, especially with throat extensions, the effect probably will be much stronger.
Before doing the measurements, I wrote down a few points I wanted to explore with this version:
Compare the CD’s response with the response of a horn without any mid ports. How much do the mid ports affect the response, and does it match the Hornresp simulation? Does it support my theory that if the resonant frequencies are brought below the compression driver’s range, the mid ports and chambers will have minimal impact on the CD’s response?
Will the mids get more low output now that the ports are larger and placed further out in the horn (compared to the MEH throat adapter mentioned earlier in this thread)? How does it compare to the Hornresp simulation?
Finally, how much impact do the mid ports have on directivity?
And the results? Both good and not so good.
The compression driver:
The graph shows the CD’s response on the current horn and on the same horn without any MEH ports. This definitely exceeded my expectations, much closer than I had hoped. A few tiny notches, but nothing major.
It works very well in this setup because the mid ports are relatively small, 8 cm² in total where they enter the horn. If the ports were larger, as needed in MEH designs that go lower and play louder, additional resonances would be introduced. So it’s a balancing act between keeping the ports small enough to get a clean CD response and large enough to play low and loud. For home hifi, this is perfect.
The mid output is also looking great, but my hornresp sim is off and the actual response is lower in frequency compared to the horn resp simulation, which means that the front chamber is probably slightly too big. The notch in the cd’s response is also lower. This means that I’ll have to go with a lower crossover than intended, I’ll try with a crossover around 700hz.
And now the bad news… I can’t seem to get a smooth crossover like I’ve usually been able to. I can’t get both the frequency response and the phase response flat. Either I get the frequency response flat, but then there’s a delay between the mid and CD. If I try to make the phase response flat, a notch appears at the crossover frequency.
My theory is that the CD and mids are too far apart, so when they play the same frequency at the same time, there’s some cancellation around the crossover region.
So you were right @nc535. I’ll have to get the mids closer to the CD.
Here is the response after the crossover.
And finally the directivity. Interestingly there is nothing major going on around the crossover region, but the high frequencies are much more messy than I hoped for. The peaks line up with the peaks in the mid response.
Compare the CD’s response with the response of a horn without any mid ports. How much do the mid ports affect the response, and does it match the Hornresp simulation? Does it support my theory that if the resonant frequencies are brought below the compression driver’s range, the mid ports and chambers will have minimal impact on the CD’s response?
Will the mids get more low output now that the ports are larger and placed further out in the horn (compared to the MEH throat adapter mentioned earlier in this thread)? How does it compare to the Hornresp simulation?
Finally, how much impact do the mid ports have on directivity?
And the results? Both good and not so good.
The compression driver:
The graph shows the CD’s response on the current horn and on the same horn without any MEH ports. This definitely exceeded my expectations, much closer than I had hoped. A few tiny notches, but nothing major.
It works very well in this setup because the mid ports are relatively small, 8 cm² in total where they enter the horn. If the ports were larger, as needed in MEH designs that go lower and play louder, additional resonances would be introduced. So it’s a balancing act between keeping the ports small enough to get a clean CD response and large enough to play low and loud. For home hifi, this is perfect.
The mid output is also looking great, but my hornresp sim is off and the actual response is lower in frequency compared to the horn resp simulation, which means that the front chamber is probably slightly too big. The notch in the cd’s response is also lower. This means that I’ll have to go with a lower crossover than intended, I’ll try with a crossover around 700hz.
And now the bad news… I can’t seem to get a smooth crossover like I’ve usually been able to. I can’t get both the frequency response and the phase response flat. Either I get the frequency response flat, but then there’s a delay between the mid and CD. If I try to make the phase response flat, a notch appears at the crossover frequency.
My theory is that the CD and mids are too far apart, so when they play the same frequency at the same time, there’s some cancellation around the crossover region.
So you were right @nc535. I’ll have to get the mids closer to the CD.
Here is the response after the crossover.
And finally the directivity. Interestingly there is nothing major going on around the crossover region, but the high frequencies are much more messy than I hoped for. The peaks line up with the peaks in the mid response.
Also have this on the mid driver of my CCDIY (diy coaxial compression driver). Measured the vollume in cad and modelled in Hornresp. Expected 10 dB drop at 4k but in reality it is at about 3k.
Maybe those peaks are vissible on the phase plots of the mid and HF. Could be phase shift issues from reflections and diffraction related to mid port length and sound reflecting back to the HF diafragm from the mid exit.
You can see the peaks of the polar also being at the same frequency as the small nulls in the frequency response of the HF.
Very interesting results 🙂
Does the physical dimensions correlate with Your measurements (port lenght, diametre, distance from CD, others)?
Are You doing further testing with reduced front chamber volume?
I've read (somewhere...) that having the peak of the mid's bandpass at the same freq as the notch would make life easier wrt xo and time delays, You're pretty close.
Phase seems close to 90deg for a large span. Did You try Chris A's procedure for alignment? First flatten FR, then adjust time delay, last adjust/trim phase at xo with filters or lots of PEQ over/under xo? Phase is just freq dependent time, so maybe this will work?
Does the physical dimensions correlate with Your measurements (port lenght, diametre, distance from CD, others)?
Are You doing further testing with reduced front chamber volume?
I've read (somewhere...) that having the peak of the mid's bandpass at the same freq as the notch would make life easier wrt xo and time delays, You're pretty close.
Phase seems close to 90deg for a large span. Did You try Chris A's procedure for alignment? First flatten FR, then adjust time delay, last adjust/trim phase at xo with filters or lots of PEQ over/under xo? Phase is just freq dependent time, so maybe this will work?
Yes, the values in the Hornresp simulation are as close to the real model as possible.
I think Hornresp calculates the front chamber as a perfect cylinder and the sound source as a perfectly flat vibrating surface. In reality the geometry is much more complex, the woofer membrane is not flat and the volume is not a perfect cylinder. So I think these complex geometries make it harder to calculate the actual response and some of the reflection paths are longer in reality than in the simulation.
Yes the plan is to a new design where the ports are closer to the CD, the mid chamber is a little smaller and the mid entry ports are long slots instead of circles.
The method I’m currently using is as follows:
- First, measure the drivers using this method: https://www.audiosciencereview.com/...ents-spinoramas-with-rew-and-vituixcad.21860/
- EQ them flat in REW and export the EQ settings for RePhase.
- In REW, within the EQ settings, click “Generate measurement from predicted.”
- Click on the predicted measurement, go to “All SPL,” click “Actions,” then “Measurement Actions,” and select “Excess Phase Version” within the Measurement Actions.
- Export the Excess Phase version as a text file.
- Open RePhase.
- Click “Import measurement” and import the Excess Phase measurement file.
- Under “Paragraphic Gain EQ,” click “Tools,” then “Open REW filter settings,” and select the EQ settings from REW.
- Then choose the crossovers under “Linear-Phase Filters.”
- Export as a .dbl file using these settings:
and import into CamillaDSP and apply to the correct driver.
Do this to both drivers and you should have a flat frequency response and flat phase.
Thanks for Your comprehensive answer. I realize You use FIR, Chris A(Cask05) used IIR and/or PEQ.
The distance from ports to CD is the probable cause for the persisting 90deg phase, and removing the 90deg with a delay should not change the FR.
Maybe measuring with the mic on axis at the mid ports would give a different FIR?
The distance from ports to CD is the probable cause for the persisting 90deg phase, and removing the 90deg with a delay should not change the FR.
Maybe measuring with the mic on axis at the mid ports would give a different FIR?
Very nice thread!
One tip; you can optimyse the midport reflections by placing damping in the ports or by placing a membrane on the ports on the hornwall with a resistive material i.e. glassfiber cloth
One tip; you can optimyse the midport reflections by placing damping in the ports or by placing a membrane on the ports on the hornwall with a resistive material i.e. glassfiber cloth
Interesting. Have You done measurement with and without the cloth? I would expect that this cloth would alter the acoustic bandpass filter and give compression at higher SPL. Any links to the used cloth?
I made a new version, but it’s not perfect. The front chamber ended up too small, which pushed the notch higher in frequency than I had hoped. Because of that, I couldn’t get a good crossover at the notch (1350 Hz). I did a polar measurement, and they look surprisingly good, the peaks and dips aren’t as intense as in the previous version.
I'm printing another version now with a bigger front chamber:
I'm printing another version now with a bigger front chamber: