Looks very cool! Remind me what printer you're using? I'm considering getting something with a large frame to be able to handle large objects in one (higher risk) shot.
I'm not familiar with Ratrig. Look like solid units if you know what you're doing!
Any idea whether budget units like the Elegoo Neptune 4 Max would be capable enough to print these?
Any idea whether budget units like the Elegoo Neptune 4 Max would be capable enough to print these?
Impressive work!
How do you like the sound quality compared to the larger JMOD MEH? Do you think the custom waveguide significantly improves sound quality compared to a standard MEH horn?
What type of filament are you using for the prints?
In your first design you positioned the mid-entry taps about ¼ wavelength from the CD exit. Does it work well?
I’ve been trying to figure out if the critical distance is between the ports and the CD exit or the ports and the CD diaphragm, as I've seen mentioned sometimes here. Based on your design, placing the ports ¼ wavelength from the CD exit seems the way to go.
I’m planning to build a compact 2-way MEH (+ subs) and am debating whether to go with a pyramid MEH horn or invest more time into learning ATH to design a custom waveguide.
Looking forward to seeing how this project evolves!
How do you like the sound quality compared to the larger JMOD MEH? Do you think the custom waveguide significantly improves sound quality compared to a standard MEH horn?
What type of filament are you using for the prints?
In your first design you positioned the mid-entry taps about ¼ wavelength from the CD exit. Does it work well?
I’ve been trying to figure out if the critical distance is between the ports and the CD exit or the ports and the CD diaphragm, as I've seen mentioned sometimes here. Based on your design, placing the ports ¼ wavelength from the CD exit seems the way to go.
I’m planning to build a compact 2-way MEH (+ subs) and am debating whether to go with a pyramid MEH horn or invest more time into learning ATH to design a custom waveguide.
Looking forward to seeing how this project evolves!
I imagine that elegoo would work quite well, yeah. The prints for this sort of thing really aren't too complex to do.
Thanks! I've A/B tested them in mono and while the new one does achieve the "pinpoint" effect that you get with this style of speaker quite well, the JMOD still has the edge, in my opinion. Better transient response and layering/depth to the sound. That being said, I do want to revisit A/B testing with the second revision. The first one was still using a cheap $60 Dayton audio KABD amp board with a pretty high noise floor and technically not enough watts, while the JMOD was hooked up to a Linea Research amp with FIR filtering, so not a fair comparison in that regard. I've got a Hypex board on the way that should level out the playing field a bit. Also I think the print rigidity on the first one wasn't great and was potentially causing unwanted vibrations/air leaks, so that should also improve.
Whether or not the waveguide profile is an improvement over a typical pyramid conical horn is still something I'm determining. We know that straight sectioned conical horns generally have polar responses that are pretty ragged. My hypothesis behind this whole design is to determine if a waveguide design that prioritizes smooth consistent polars will sound subjectively any better that the conical alternative. Keeping in mind that this approach loses out on the idealized LF loading you get with a conical horn.
In theory you could do a "best of both worlds" approach by designing an extended throat adapter in Ath that provides a better transition to the conical section (see below). I think something like this becomes possible when using coax compression drivers like the DCX464 since it eliminates the need for midrange drivers, freeing up considerable space around the throat. It's probably what I'll do for a rev 2.0 of the JMOD design if it goes as well as I think it will.
I'm using PLA+ for the prototypes since it prints easy, but will move to PETG (or PC if it cooperates) for the final version.
I follow the 1/4 wavelength from the CD exit rule. I know some people like to do it from the diaphragm, but I've had good results with my method that mostly agree with the simulations.
Prototype B is assembled. Did some preliminary testing/tuning to get a flat-ish FIR preset going on the hypex. Will take polars & distortion measurements on a later date.
You can see the on axis comparisons between prototype a/b below. The main differences being:
1. More bass extension as a result of a slightly larger enclosure & different port tuning
2. More leakage from the woofer taps >2khz as the acoustic lowpass effect is likely less effective than having the taps positioned at the edge of the cone (this was expected)
3. Slightly more sensitivity from the compression driver in the 2-5k range, not sure why, but it's small, so could just be measurement error.
There are some interesting high q dips at 3.3k, 4.1k, and 5k that may warrant some investigation, though I doubt they would be audible. They could be related to the way I measured (ground plane), so I'll try and get the speaker elevated and do a gated measurement to see if they're still there.
You can see the on axis comparisons between prototype a/b below. The main differences being:
1. More bass extension as a result of a slightly larger enclosure & different port tuning
2. More leakage from the woofer taps >2khz as the acoustic lowpass effect is likely less effective than having the taps positioned at the edge of the cone (this was expected)
3. Slightly more sensitivity from the compression driver in the 2-5k range, not sure why, but it's small, so could just be measurement error.
There are some interesting high q dips at 3.3k, 4.1k, and 5k that may warrant some investigation, though I doubt they would be audible. They could be related to the way I measured (ground plane), so I'll try and get the speaker elevated and do a gated measurement to see if they're still there.
Attachments
Polars/Spinorama and HD measurements of prototype B indicate a nice improvement from locating the taps on the knuckles. Notably there's much less pattern widening occurring at 1.5-4khz.
Interesting that H2 rises up at 8-10k. Maybe that could give a false impression of the DH450 having a higher extension than it really does?
Interesting that H2 rises up at 8-10k. Maybe that could give a false impression of the DH450 having a higher extension than it really does?
Further fine tuning and comparing the response I'm comparing my measurements of the DH450 on this waveguide versus the published B&C response (as measured on an ME45 horn) to weed out what shouldn't be there.
Mainly interested in investigating the nulls occurring around 3.3k, 4.1k, and 4.95k. These seem to appear in both the compression driver and the woofer's responses, and are not present in B&C's graph (on the right).
So my first thought is that this could be resultant of a throat reflection occurring due to diffraction at the edge of the tap. This would make sense if any potential diffraction sources lined up with the 1/4 wavelength of the nulls, but they don't appear to.
Upon further investigation however, it seems like this is far more likely the result of resonances from the waveguide material itself as evidenced by the long-tailed ridges in the CSD plot (thanks to @Nissep for setting me straight on this).
So the next step will be to measure the compression driver with the taps taped over. If the nulls still appear, it seems like the material resonating would be a cause. If they are eliminated, then it might be something else.
Mainly interested in investigating the nulls occurring around 3.3k, 4.1k, and 4.95k. These seem to appear in both the compression driver and the woofer's responses, and are not present in B&C's graph (on the right).
So my first thought is that this could be resultant of a throat reflection occurring due to diffraction at the edge of the tap. This would make sense if any potential diffraction sources lined up with the 1/4 wavelength of the nulls, but they don't appear to.
Upon further investigation however, it seems like this is far more likely the result of resonances from the waveguide material itself as evidenced by the long-tailed ridges in the CSD plot (thanks to @Nissep for setting me straight on this).
So the next step will be to measure the compression driver with the taps taped over. If the nulls still appear, it seems like the material resonating would be a cause. If they are eliminated, then it might be something else.
Hey, great work. Very impressive designs and overall concepts!
The effect of changing port location (and size?) is quiet remarkably. It is hard to see from the pictures, did you change the port size? The four ports seem bigger but are they double the size? What is the actual port size in comparison to Sd?
I could not find any distortion measurements of version A, did it change for the woofers with the different ports?
The effect of changing port location (and size?) is quiet remarkably. It is hard to see from the pictures, did you change the port size? The four ports seem bigger but are they double the size? What is the actual port size in comparison to Sd?
I could not find any distortion measurements of version A, did it change for the woofers with the different ports?
As far as the port area, the area of each port in revision A was 7sq cm. Since I halved the number of ports in revision B, the area doubled to 14sq cm per port. The Sd of each woofer is 132sq cm. Although the port size is a bit difficult to estimate since the geometry is weird, so take that with a grain of salt.
Okay that is very impressive! could be interesting to find out the reason for the resonance at 700 Hz Hz, It shows in the decay and only the in the sonogram.
Its on the todo list as a priority. I have a few ideas of what it might be, but I have more measurements to make before I can make a determination there. It didn't seem to be as prominent in the first prototype, so I'm curious if something shifted or changed in the driver itself. Either way i've got more DH450s on the way, so that should be an easy one to rule out.
Determined that the dips in the response at 1.8k, 3.3k, 4.1k, and 4.9k were due to diffraction from the ports. Back to redesigning them I suppose.
Unsmoothed, 14ms gate. Ignore absolute SPL as I separated the traces for readability.
Unsmoothed, 14ms gate. Ignore absolute SPL as I separated the traces for readability.
New revision with staggered ports of slightly varying distances from the throat. The shapes are a bit different too with the idea being that asymmetry will assist in "breaking up" the nulls that are happening.
Comparison below shows that the steepness of the nulls was reduced quite a bit.
For polars, the vertical axis now has a slight narrowing happening in the top octave. Horizontal is the same as the previous prototype.
(^^^ this one is labeled horizontal I know, but its really the vertical.)
I was able to get a bit flatter on-axis frequency response now that the cancellation notches are lessened. This spin was generated only with +-90 degree measurements, so need to do a full one once I get some time to take it.
Comparison below shows that the steepness of the nulls was reduced quite a bit.
For polars, the vertical axis now has a slight narrowing happening in the top octave. Horizontal is the same as the previous prototype.
(^^^ this one is labeled horizontal I know, but its really the vertical.)
I was able to get a bit flatter on-axis frequency response now that the cancellation notches are lessened. This spin was generated only with +-90 degree measurements, so need to do a full one once I get some time to take it.
Nearing completion of this project with a final configuration render below. I used STV's port optimization spreadsheet to settle in on the below port design. The first couple of cm will be machined plywood, with the remainder of the port being 3d printed.
A summary of updates since the last post:
Once I get the final box built and validated, I'm going to clean up the files and upload .STLs for people who might be interested in printing this. Otherwise, I'll have the waveguide assemblies available for purchase in the US since its not exactly an easy print.
I've got the measurements (so far) uploaded here. These were taken on the somewhat janky C prototype without the new port, so I'll be re-taking them with the final box when I have it built.
A summary of updates since the last post:
- The distortion spike that was occurring at 700Hz turned out to be debris on the compression driver diaphragm. Cleaning this off effectively eliminated that spike.
- I've been working with a local 3d printing shop to get a high strength resin version of the waveguide made. The below is the first-pass prototype of that, but I should be getting a new one with the revised port geometry any day now. This is using Anycubic's Rigid 100 resin. I've been able to thoroughly test it in a box at several live events over the last few months and its held up remarkably well.
Once I get the final box built and validated, I'm going to clean up the files and upload .STLs for people who might be interested in printing this. Otherwise, I'll have the waveguide assemblies available for purchase in the US since its not exactly an easy print.
I've got the measurements (so far) uploaded here. These were taken on the somewhat janky C prototype without the new port, so I'll be re-taking them with the final box when I have it built.
Right now I'm using FIR to flatten out the phase for the majority of the operating frequency range.
But you can see what the phase looks like for the woofers/CD without any processing below:
(ignore absolute SPL)