I'm self-doubting whether to do a better 2x4" MEH or (using 16ohm 4FE35) 4x4" square MEH. With my "more coherent" 1st-order-notched T/MM I think I heard a bit of HF108 ketone polymer membrane sound signature, trading steely edged strings for hard plastickyness -- a little bit. My Minmeh 2nd-order-notched, flatter FR didn't do that.
Please consider giving "reflector point-source" topology a test. Just point a small car tweeter at midwoofer dustcap and move it around until their acoustic centers (one real, one reflection) became coincident and the sound holographic. Also LX which reduced the acoustic centers gap beween midtweeter and up-firing midwoofer.
Last edited:
Omnidirectional has it's use and users, but it can be hard to find a proper placement, taking up to much space (that could be used for a BIG horn 🙂. Depending on what ones intended use is, probably somewhere between critical listening and background noise, one should aim for a suitable system. I would prefer something that is true to the recording, so I would try to avoid early reflections (<10ms) to "hide" the room. My idea for a horn is to utilize the ability to (somewhat) control the beamwidht, and getting single-point source would be a big benefit. If in need of more reflections, one could toe the speakers in/out, but that would demand speakers suited for the listening room and/or vice versa. DIY can make this happen 🙂
Sure, depends on the room, placement, and seating. Many ways to reduce dispersion & reflections; not so many to get point-source (as you've said). LX cardioid-dipole can be close to walls. Point-source reflector size/shape is diy to taste (midwoofer face front or up). I'm doing MEH in a smallish room only 3.5x6m so directivity helps. Not really sure 8cm vs 12cm effective CtoC would make much difference far-field in a large space.
Last edited:
^ Agree. CtC-distances can be somewhat relaxed with correct filtering, details at Linkwitzlab. My listening room is somewhat like Yours, +low ceiling LWH 717x336x219cm, so for me a 64x46 degree horn would be nice, but it has to be BIG to get that pattern and have loading down to the Schroeder-frequency. One not-so-obvious possibility with MEH is the low group delay at low (bass) frequencies. Any abrupt changes in a horn will give reflections, reflections will give a ragged impedance (and freq) response, the same seems to apply for the mouth => a rounded mouth should be beneficial.
Sketch of predicted in-room reflexes. Dots are ego-centric listening-positions 😎
Sketch of predicted in-room reflexes. Dots are ego-centric listening-positions 😎
Updates on the throat adaptor experiments.
I made an R-OSSE waveguide in ATH and printed it in PLA using a 0.6 mm nozzle and 0.18 mm layer height. It resulted in a super smooth finish, the Bambulab A1 is such a good printer for the price. I was going to use a two-part epoxy to glue it together, but one part of the epoxy had hardened (probably too old), so I went with both superglue and contact glue. It’s probably not very strong, but it should work for this experiment.
The waveguide has a 1.4 mm entry with a 9.5 degree entry angle. I also designed an adaptor going from 1 inch to 1.4 inch with a 9.5 degree angle, it’s about 61 mm long. The throat adaptor with the mid entry point is 120 mm long, so the horn throat is 181 mm from the CD entry and 116 mm from the mid entry ports. It looks quite ridiculous, and I didn’t expect it to work very well. I used the mid rear chambers from the V1 MEH.
Here are the raw measurements:
A bit disappointing seeing the low extension of the mids but what can you expect with these small slits. However, the compression driver is performing well in this configuration, and it’s surprisingly flat. I did some rough EQ and brought the higher range of the mids down so I could cross them over at 150 Hz. Not ideal but it seems to work okayish. It would probably work better to use the mids between 600-1300hz, but my goal was to use them all the way down to 150hz.
Here is the response after EQ and crossover. There is a notch at around 310hz
Directivity response (normalized). The CD is doing great, but the mids are rougher.
For the next version, I’d like to skip the 1-to-1.4 inch converter and make a new MEH throat adapter that goes from 1 inch to 1.4 inch. The adapter's exit angle would match the throats entry of 9.5 degrees.
I’m also trying to figure out what changes would improve the mid tap ports. They definitely need to be bigger, and moving them closer to the horn throat should help to increase the low end.
The slit is currently 44 mm to 84 mm away from the compression driver. That corresponds to half a wavelength of 3897 Hz to 2041 Hz. This somewhat aligns with the notches in the response. Interestingly, the response around 3 kHz, right between those two notches looks good. Could I make the slit more gradual, shaped like an eye or a teardrop, to reduce harsh reflections? The edges of the slit aren’t rounded either, smoothing them out might improve the response.
If I move the mid ports further from the compression driver’s entry, I could likely shift the notch lower. Ideally, they should align with the crossover point. I assume they’d need to be half a wavelength from the CD entry to create a cancellation notch. So, for a crossover at 1300 Hz, the distance would need to be about 13.19 cm. But that would break Danley’s ¼-wavelength rule between the mid and the CD.
In the first throat adapter test, the CD’s response was my main concern and the mids looked pretty decent, considering they were open-back. Now the mids are worse. Is the distance from the horn throat the culprit? Given all this, I think placing the mid entry ports as close to the throat as possible is probably best. Maybe even on the horn walls? But then I’m back to square one. No… I’ll keep exploring this idea of putting the mid ports on the throat adapter.
I’ve been listening to it while writing this. The high frequencies are beautiful, smoother than on my previous MEHs, less metallic, which is great. The mids, though, are a bit muffled, not so great.
Anyway, I’m off on holiday now, so the exploration will continue in about two weeks.
I made an R-OSSE waveguide in ATH and printed it in PLA using a 0.6 mm nozzle and 0.18 mm layer height. It resulted in a super smooth finish, the Bambulab A1 is such a good printer for the price. I was going to use a two-part epoxy to glue it together, but one part of the epoxy had hardened (probably too old), so I went with both superglue and contact glue. It’s probably not very strong, but it should work for this experiment.
The waveguide has a 1.4 mm entry with a 9.5 degree entry angle. I also designed an adaptor going from 1 inch to 1.4 inch with a 9.5 degree angle, it’s about 61 mm long. The throat adaptor with the mid entry point is 120 mm long, so the horn throat is 181 mm from the CD entry and 116 mm from the mid entry ports. It looks quite ridiculous, and I didn’t expect it to work very well. I used the mid rear chambers from the V1 MEH.
Here are the raw measurements:
A bit disappointing seeing the low extension of the mids but what can you expect with these small slits. However, the compression driver is performing well in this configuration, and it’s surprisingly flat. I did some rough EQ and brought the higher range of the mids down so I could cross them over at 150 Hz. Not ideal but it seems to work okayish. It would probably work better to use the mids between 600-1300hz, but my goal was to use them all the way down to 150hz.
Here is the response after EQ and crossover. There is a notch at around 310hz
Directivity response (normalized). The CD is doing great, but the mids are rougher.
For the next version, I’d like to skip the 1-to-1.4 inch converter and make a new MEH throat adapter that goes from 1 inch to 1.4 inch. The adapter's exit angle would match the throats entry of 9.5 degrees.
I’m also trying to figure out what changes would improve the mid tap ports. They definitely need to be bigger, and moving them closer to the horn throat should help to increase the low end.
The slit is currently 44 mm to 84 mm away from the compression driver. That corresponds to half a wavelength of 3897 Hz to 2041 Hz. This somewhat aligns with the notches in the response. Interestingly, the response around 3 kHz, right between those two notches looks good. Could I make the slit more gradual, shaped like an eye or a teardrop, to reduce harsh reflections? The edges of the slit aren’t rounded either, smoothing them out might improve the response.
If I move the mid ports further from the compression driver’s entry, I could likely shift the notch lower. Ideally, they should align with the crossover point. I assume they’d need to be half a wavelength from the CD entry to create a cancellation notch. So, for a crossover at 1300 Hz, the distance would need to be about 13.19 cm. But that would break Danley’s ¼-wavelength rule between the mid and the CD.
In the first throat adapter test, the CD’s response was my main concern and the mids looked pretty decent, considering they were open-back. Now the mids are worse. Is the distance from the horn throat the culprit? Given all this, I think placing the mid entry ports as close to the throat as possible is probably best. Maybe even on the horn walls? But then I’m back to square one. No… I’ll keep exploring this idea of putting the mid ports on the throat adapter.
I’ve been listening to it while writing this. The high frequencies are beautiful, smoother than on my previous MEHs, less metallic, which is great. The mids, though, are a bit muffled, not so great.
Anyway, I’m off on holiday now, so the exploration will continue in about two weeks.
Nice to see You are testing out ideas 🙂 I think You are into something good with the CD adapter.
The rule of horn cross-section area(CSA=((c/f)^2)/4pi) where the ports enters the horn is challenged, maybe this can be part of the reason to the lacking low response of the mid's?
Can also be the volume of the mid's rear chamber? Distance between ports is definitely not to high.
Not to forget Danleys advice about flare rate wrt frequency, which I don't fully understand how to determine.
Please do test the idea of eye-shaped ports.
As I understand, the notch will be where the center of the mid ports are 1/4wL (90deg) from the reflective surface, which some says is the membrane of the CD, or the grill of the CD, or the horn apex... I think all of these give some reflections(frequency dependent) so the reflected wave will be somewhat smeared out in time and amplitude, which will have an effect on the acoustic lowpass rolloff steepness.
A narrow throat might give some interesting results for high SPL, maybe more distortion and intermodulation between CD and mids.
For aligning the pressure wave from the CD and the mid, I think the distance from the mid's voicecoil through the port and to a point on the horn axis can be made such that You might avoid using any time delay between CD and mid.
Enjoy Your holiday !
The rule of horn cross-section area(CSA=((c/f)^2)/4pi) where the ports enters the horn is challenged, maybe this can be part of the reason to the lacking low response of the mid's?
Can also be the volume of the mid's rear chamber? Distance between ports is definitely not to high.
Not to forget Danleys advice about flare rate wrt frequency, which I don't fully understand how to determine.
Please do test the idea of eye-shaped ports.
As I understand, the notch will be where the center of the mid ports are 1/4wL (90deg) from the reflective surface, which some says is the membrane of the CD, or the grill of the CD, or the horn apex... I think all of these give some reflections(frequency dependent) so the reflected wave will be somewhat smeared out in time and amplitude, which will have an effect on the acoustic lowpass rolloff steepness.
A narrow throat might give some interesting results for high SPL, maybe more distortion and intermodulation between CD and mids.
For aligning the pressure wave from the CD and the mid, I think the distance from the mid's voicecoil through the port and to a point on the horn axis can be made such that You might avoid using any time delay between CD and mid.
Enjoy Your holiday !
Deeply impressive printing and construction there, thank you for posting the pictures; it is inspiring to see!
FWIW bigger rear chambers can help the low end in (what are effectively) such bandpass arrangements. Larger (or at least less tiny) front chambers can help widen the pass band and increase efficiency (so cone fillers may not 'necessarily' be advantageous unless you're struggling to reach up to the compression driver's range).
But ultimately these little drivers are quite displacement-limited for bass duties, except at low SPLs. Their small Sd means the excursion needed shoots up at low frequencies; here is a 4FE35 with a 5L rear chamber and a tap size about 1/8th of Sd:
(EDIT: here the 4FE35 was simulated as a mid-range; it could likely be optimised better had it been intended as a 2-way)
So with EQ they would work at modest SPLs, perhaps mated to a subwoofer. But of course more or bigger drivers would offer more output; maybe four smaller or two 6" for my taste would seem a useful size. Though it is more tricky to reach up to a ~1khz crossover as they get bigger (one of the trade-offs with a 2-way MEH); the biggest I've seen achieved are 8" drivers in a 2-way - after that, a compression driver that can go lower seems necessary, except at very low spl. Meaty larger format compression drivers can be crossed lower to much bigger drivers, but can suffer at the top end; how much is somewhat related to the waveguide/horn type (for instance Geddes didn't like greater than 1" drivers on his oblate spheroid guides).
Often, moving the taps closer to the compression driver is about raising the top end of the cone drivers rather than the lower end. Due to the usual aim of keeping the taps within 1/4 wavelength of the compression driver's effective acoustic centre (which is generally somewhere inside the compression driver itself) and also within 1/4 wavelength of each other, although on narrower dispersion horns that 1/4 wavelength can be relaxed slightly. So the closer the taps are, the higher in frequency this ideal relationship will be maintained.
But secondly there is the null caused by reflections going down and bouncing back up the horn. This occurs when the total reflected distance (down and back) is half a wavelength, so in effect the one-way distance is again 1/4 wavelength. Though this time the distance is measured from where the reflection occurs, typically around the throat or the compression-driver screen in a unity horn though it varies - but might be quite different with your new throat profile. This null can be usefully placed to coincide with the crossover slope, or failing that to be pushed above (out of) the range at which the cone drivers are to operate. If your null is falling within the cone driver's intended range then yes, closer tap placement could certainly help.
Thirdly there is the expansion rate of the horn at the point where the taps enter. Arguably this is mostly about horn loading, which can also help the low end. Your throat adaptor seems to have the taps in a tubular non-expanding section of the throat? I haven't looked at the effects of that myself (and TBH am more into waveguides for their directivity, I don't know much about horn loading) so I don't really know what the effects might be, but feel it could be worth considering unless you've already done so.
FWIW bigger rear chambers can help the low end in (what are effectively) such bandpass arrangements. Larger (or at least less tiny) front chambers can help widen the pass band and increase efficiency (so cone fillers may not 'necessarily' be advantageous unless you're struggling to reach up to the compression driver's range).
But ultimately these little drivers are quite displacement-limited for bass duties, except at low SPLs. Their small Sd means the excursion needed shoots up at low frequencies; here is a 4FE35 with a 5L rear chamber and a tap size about 1/8th of Sd:
(EDIT: here the 4FE35 was simulated as a mid-range; it could likely be optimised better had it been intended as a 2-way)
So with EQ they would work at modest SPLs, perhaps mated to a subwoofer. But of course more or bigger drivers would offer more output; maybe four smaller or two 6" for my taste would seem a useful size. Though it is more tricky to reach up to a ~1khz crossover as they get bigger (one of the trade-offs with a 2-way MEH); the biggest I've seen achieved are 8" drivers in a 2-way - after that, a compression driver that can go lower seems necessary, except at very low spl. Meaty larger format compression drivers can be crossed lower to much bigger drivers, but can suffer at the top end; how much is somewhat related to the waveguide/horn type (for instance Geddes didn't like greater than 1" drivers on his oblate spheroid guides).
Often, moving the taps closer to the compression driver is about raising the top end of the cone drivers rather than the lower end. Due to the usual aim of keeping the taps within 1/4 wavelength of the compression driver's effective acoustic centre (which is generally somewhere inside the compression driver itself) and also within 1/4 wavelength of each other, although on narrower dispersion horns that 1/4 wavelength can be relaxed slightly. So the closer the taps are, the higher in frequency this ideal relationship will be maintained.
But secondly there is the null caused by reflections going down and bouncing back up the horn. This occurs when the total reflected distance (down and back) is half a wavelength, so in effect the one-way distance is again 1/4 wavelength. Though this time the distance is measured from where the reflection occurs, typically around the throat or the compression-driver screen in a unity horn though it varies - but might be quite different with your new throat profile. This null can be usefully placed to coincide with the crossover slope, or failing that to be pushed above (out of) the range at which the cone drivers are to operate. If your null is falling within the cone driver's intended range then yes, closer tap placement could certainly help.
Thirdly there is the expansion rate of the horn at the point where the taps enter. Arguably this is mostly about horn loading, which can also help the low end. Your throat adaptor seems to have the taps in a tubular non-expanding section of the throat? I haven't looked at the effects of that myself (and TBH am more into waveguides for their directivity, I don't know much about horn loading) so I don't really know what the effects might be, but feel it could be worth considering unless you've already done so.
Last edited:
I would like to hear that.
It's a very long throat.
I've gone away from that sort of thing to faster opening flare rates.
It's interesting.
It's a very long throat.
I've gone away from that sort of thing to faster opening flare rates.