The
The CD has a deep notch above 1 khz, so I didn't even try with a lower XO. The next version, with a bigger front chamber will finish printing tonight, so hopefully I'll have time to do measurements tomorrow.
The CD has a deep notch above 1 khz, so I didn't even try with a lower XO. The next version, with a bigger front chamber will finish printing tonight, so hopefully I'll have time to do measurements tomorrow.
my previous post was pre-coffee, my prior experience makes me think of reflection null first thing, which somehow doesn't seem to be an issue here. I hope I'm fully awake now.
Let's think about the Helmholtz resonance for a moment.
The resonant freq is proportional to the sqrt of (Aneck/(Vcavity*Lneck)
We want the resonance to be below crossover because only the CD sees it. If it’s below the XO, it won’t affect the combined response. To the mid drivers, the cavity plus port creates an acoustic LPF. These two things work against each other because we want a short neck with a large Xsection area to pass the mid signal without exceeding 17m/s particle velocity.
What this suggests to me is an array of small holes instead of a single large hole. The small holes have a low resonant frequency because of their small area. The sum of the areas of the small hole should provide enough area to satisfy the mids.
You could do a quick and dirty experiment dividing your slits up into perhaps 4 segments with epoxy or hot glue. The relatively thin horn wall makes for a short Lneck which raises resonant frequency. You could probably make the wall thicker without affecting the mid response too badly.
One thing different about your design is the large cavity volume. Most of us minimized the cavity volume in an effort to get the mids to play higher. Raising the volume also lowers the Helmholtz resonance. I guess it’s a good thing so long as your mids play high enough. Given that the mids are excursion limited, you don’t even care if you need to boost the high end of their range to make the XO work, so long as you don’t exceed max power…
Let's think about the Helmholtz resonance for a moment.
The resonant freq is proportional to the sqrt of (Aneck/(Vcavity*Lneck)
We want the resonance to be below crossover because only the CD sees it. If it’s below the XO, it won’t affect the combined response. To the mid drivers, the cavity plus port creates an acoustic LPF. These two things work against each other because we want a short neck with a large Xsection area to pass the mid signal without exceeding 17m/s particle velocity.
What this suggests to me is an array of small holes instead of a single large hole. The small holes have a low resonant frequency because of their small area. The sum of the areas of the small hole should provide enough area to satisfy the mids.
You could do a quick and dirty experiment dividing your slits up into perhaps 4 segments with epoxy or hot glue. The relatively thin horn wall makes for a short Lneck which raises resonant frequency. You could probably make the wall thicker without affecting the mid response too badly.
One thing different about your design is the large cavity volume. Most of us minimized the cavity volume in an effort to get the mids to play higher. Raising the volume also lowers the Helmholtz resonance. I guess it’s a good thing so long as your mids play high enough. Given that the mids are excursion limited, you don’t even care if you need to boost the high end of their range to make the XO work, so long as you don’t exceed max power…
Yes, there are a lot of variables that have changed. The port area is smaller (5.77 cm² vs 8 cm²), the port is much closer to the CD, and it’s now an elongated slot instead of a circular hole. Also, the front chamber is quite a bit smaller, and the distance between the midwoofer and the port entry into the horn is shorter.
Thank you! I got a bit frustrated after the first two prototypes, I spent so much time on things like building the enclosure and painting the horn (which was printed with PETG CF, so it’s safer to paint it to avoid getting carbon fibers on your skin). After seeing @wchang’s quick test, I figured I could just focus on the 3D printed parts for now and make quick prototypes that I print overnight. That way, I can quickly test out ideas I haven’t been able to simulate properly and learn how different parameters affect the response.
My current fusion document is so messy so it's hard to present it in a understandable way, but here is the sketch of the horn profile from the middle to the throat. Most of the design is derived from this one sketch.
Here is the slot in the latest design:
That's an excellent idea! Thanks @nc535. I just did the measurements of the newest horn part, and the results were disappointing. The notch shifted slightly, but not enough to reach my target crossover around 900–1000 Hz. What I probably need is a longer port.
Here’s the response, comparing the new version to the previous one mentioned in post 99.
After reading your reply I designed this insert that I'll insert into the mid ports so I'm quickly able to compare the difference.
Measurements are ready:
These are the three versions of a similar horn part.
The first one (oldest, June 13) has a smaller front chamber and 5 mm port depth.
The second one (medium new, June 14) has a larger front chamber and the same 5 mm port depth.
The third one (newest, June 15) uses the same horn part as the second, but with added port adapters that convert the slot into an array of small circular ports, all 8.75 mm deep.
My conclusion is that I need more dramatic changes to bring the CD response down in frequency. The knee of the response is at roughly the same frequency in all measurements, what changes is the steepness of the falloff. I need to figure out how to bring the knee down in frequency. I also don't think I need to have this large front chamber, I'd prefer a mid response that is more flat up until the XO frequency, if possible. Back to Hornresp for more simulation.
These are the three versions of a similar horn part.
The first one (oldest, June 13) has a smaller front chamber and 5 mm port depth.
The second one (medium new, June 14) has a larger front chamber and the same 5 mm port depth.
The third one (newest, June 15) uses the same horn part as the second, but with added port adapters that convert the slot into an array of small circular ports, all 8.75 mm deep.
My conclusion is that I need more dramatic changes to bring the CD response down in frequency. The knee of the response is at roughly the same frequency in all measurements, what changes is the steepness of the falloff. I need to figure out how to bring the knee down in frequency. I also don't think I need to have this large front chamber, I'd prefer a mid response that is more flat up until the XO frequency, if possible. Back to Hornresp for more simulation.
I’ve done some more Hornresp simulations. The port length has a big impact on the notch, bringing it down as the port length increases. This version features a smaller front chamber but deeper ports, around 2.4 cm. I decided to keep the long slot like in the previous design, it was easier since I didn’t have to change the 3D model too much, and I think it will make for a fairer comparison with the previous versions to see what the increased port length does to the response. If the result isn’t as good as I hope, I’ll experiment with multiple smaller ports instead.
Here’s roughly what I’m hoping for in terms of response:
Here’s roughly what I’m hoping for in terms of response:
Success! Finally got the drivers to vibe together!
Here is the raw response from mids and CD:
Final response after XO:
Directivity:
Here is the raw response from mids and CD:
Final response after XO:
Directivity:
Nice work 🙂 That looks amazingly good!
Please do educate the rest of the bunch with a walk-through of Your theories and decisions, it will be of great value for further/upcoming designs 🙂
Please do educate the rest of the bunch with a walk-through of Your theories and decisions, it will be of great value for further/upcoming designs 🙂
Thank you!
It’s pretty close to my initial goal, but still not 100% perfect. If you look at the phase plot, it rises a bit around the crossover region. While the mids and CD now blend better, it’s still not as smooth as when I did the crossover for the throat adapter, where the path lengths were much shorter. My theory is that the distance between the mids and the CD is still a bit too long.
So I’ve made a new version with four ports that are more rounded and not as long as the previous slots. The idea is to shorten the path lengths from the mids to the CD enough to allow for a crossover around 1000 Hz. I was concerned these rounded ports might introduce more severe notches in the CD’s response, so I’ve staggered them and made each one a different shape, but their area remains similar (around 2 cm² per port, 8 cm² total).
In my previous version, the main change was increasing the port tube axial length (i.e., the distance from the compression chamber exit to the horn entry, not the slot length), which lowered the frequency of the CD’s notch. However, it was still slightly too high for a 1000 Hz crossover. In this new version the port is even deeper, around 3 cm. That said, making the ports too long starts to affect the mids’ SPL output, so I don’t want to take it much further. I also increased their size slightly (from 6.2 to 8 cm²) to help reduce the risk of port compression noise.
You can also see that I added rims to the port tubes inside the compression chamber. This allows me to extend the port length without having to move the mid drivers further away.
Here is the 3d model of the newest version:
It’s pretty close to my initial goal, but still not 100% perfect. If you look at the phase plot, it rises a bit around the crossover region. While the mids and CD now blend better, it’s still not as smooth as when I did the crossover for the throat adapter, where the path lengths were much shorter. My theory is that the distance between the mids and the CD is still a bit too long.
So I’ve made a new version with four ports that are more rounded and not as long as the previous slots. The idea is to shorten the path lengths from the mids to the CD enough to allow for a crossover around 1000 Hz. I was concerned these rounded ports might introduce more severe notches in the CD’s response, so I’ve staggered them and made each one a different shape, but their area remains similar (around 2 cm² per port, 8 cm² total).
In my previous version, the main change was increasing the port tube axial length (i.e., the distance from the compression chamber exit to the horn entry, not the slot length), which lowered the frequency of the CD’s notch. However, it was still slightly too high for a 1000 Hz crossover. In this new version the port is even deeper, around 3 cm. That said, making the ports too long starts to affect the mids’ SPL output, so I don’t want to take it much further. I also increased their size slightly (from 6.2 to 8 cm²) to help reduce the risk of port compression noise.
You can also see that I added rims to the port tubes inside the compression chamber. This allows me to extend the port length without having to move the mid drivers further away.
Here is the 3d model of the newest version:
I got the mounting hole distances wrong on this previous version, so I couldn’t mount the mids properly. With the help of some gaffer tape, I managed to take measurements, which turned out to be quite promising.
After that I made some updates to the 3D model. I reduced the size of the front chamber and removed the rims at the port entry. The result is a simpler and cleaner design overall.
Here are the renderings:
And here are the raw measurements:
And polars (normalized with 1/24 smoothing):
I’m so happy with the result. I can finally stop these mid port chamber experiments and focus on other things.
Next up is building a sub box which the horn will sit on. Then I can finally give this setup a proper listen.
After that I made some updates to the 3D model. I reduced the size of the front chamber and removed the rims at the port entry. The result is a simpler and cleaner design overall.
Here are the renderings:
And here are the raw measurements:
And polars (normalized with 1/24 smoothing):
I’m so happy with the result. I can finally stop these mid port chamber experiments and focus on other things.
Next up is building a sub box which the horn will sit on. Then I can finally give this setup a proper listen.