My experience is that a large radius on the mouth fixes edge diffraction to a great extent.
For an example of a similar speaker with a passive crossover, have a look a this. Unfortunately the images have all disappeared. Look here (Dutch forum) for the images. I'm sure you'll manage to figure out what's what.
For an example of a similar speaker with a passive crossover, have a look a this. Unfortunately the images have all disappeared. Look here (Dutch forum) for the images. I'm sure you'll manage to figure out what's what.
@AllenB I have about 16cm worth of offset if I look at the impulse response only for about 200 degrees lag at 1200Hz, so I should increase asymmetry to 2 orders. With the woofer having some peaks above the crossover frequency it's looking more and more like a higher order filter.
@keyser: heel erg bedankt voor de link. Leuk projekt.
I am picking up on many alternate filter topologies to bend the tweeters to do ones will. Seems like 16-18 component filters aren't out of the ordinary once one moves up from Econowave territory.
That Dayton horn is not CD and it won't have the spherical segment wavefront that brings these kinds of artifacts to the fore. The upshot of the OS seems to be that the time domain is best behaved among the wide angle devices, especially if one adds foam.
Thing is, the baffle is already 16" wide for a 12" woofer and the waveguide makes use of most of that. There is 2mm all the way round to trim off excess still and already the terminus has just 5mm left without a cabinet roundover...
@keyser: heel erg bedankt voor de link. Leuk projekt.
I am picking up on many alternate filter topologies to bend the tweeters to do ones will. Seems like 16-18 component filters aren't out of the ordinary once one moves up from Econowave territory.
That Dayton horn is not CD and it won't have the spherical segment wavefront that brings these kinds of artifacts to the fore. The upshot of the OS seems to be that the time domain is best behaved among the wide angle devices, especially if one adds foam.
Thing is, the baffle is already 16" wide for a 12" woofer and the waveguide makes use of most of that. There is 2mm all the way round to trim off excess still and already the terminus has just 5mm left without a cabinet roundover...
I heard of people who like your new horn better for some reason, even if they measure bad. What do you think of it?
The idea is to listen to a system that has rising directivity and compare it to one with more constant directivity. It's something that has gripped me for some time and the only way to somewhat know is to listen how the characteristics compare. Perhaps it's unfair to pit a large well optimized 3-way horn at a 12"+ OSWG, but I just want to listen A/B how the upper mid and HF characteristics differ.
The dip on axis is not too worrying to me as it disappears by the time you're 10 degrees off axis and these are meant to be toed in heavily.
Here the Mid and High of the horn setup:
Please note, 2dB scale 2m in-room ungated. REW because Holm doesn't overlay as many measurements.
And smoothed to display the trends
Please note, 2dB scale 2m in-room ungated. REW because Holm doesn't overlay as many measurements.
And smoothed to display the trends
Ahh, I wasn't very clear at all. When I said variable radius I meant a radius that changes as you rotate about the mouth. For example you could have a 4"r horizontally which changes to a 2"r vertically.....so the radius itself is constant it just changes with angle.
Your radius is larger than it looks on my phone!
Each order would add 45 degrees where it cuts in, and 90 overall. Going lower in frequency makes the fixed distance less significant, but sees less of the natural woofer rolloff and more from the tweeter.
@Nate: I've had that thought before but for now I'm quite happy with my limited 3D CAD skills not to have to do a variable radius fillet. One could also emulate such a thing perhaps with patchy application of felt?
@AllenB: yeah, I think the 2 order asymmetry seems to work somewhat and I get some BSC thrown in the bag by choosing a low initial filter pole. The driver breakup and above stop-band rise also works against one to an extent as the acoustical response determines the actual phase and not just the filter. I should do some theoretical models and simulate up a time-delay filter. Phase tracking only seems viable over a narrow band and then there is also optimising the filter for power response and not just the primary listening angle.
In short, I'll have to construct a nice angular measurement setup and do the legwork. Might take a while but it's important to the end result so rather do it right.
@AllenB: yeah, I think the 2 order asymmetry seems to work somewhat and I get some BSC thrown in the bag by choosing a low initial filter pole. The driver breakup and above stop-band rise also works against one to an extent as the acoustical response determines the actual phase and not just the filter. I should do some theoretical models and simulate up a time-delay filter. Phase tracking only seems viable over a narrow band and then there is also optimising the filter for power response and not just the primary listening angle.
In short, I'll have to construct a nice angular measurement setup and do the legwork. Might take a while but it's important to the end result so rather do it right.
I've done up a quick and dirty crossover at a listening angle and it is sounding somewhat different compared to what I am used to. My wife who owns a better set of ears than I do noted on rock recordings that she expected classical voices to become too bright but when we did play she noted it sounded natural with some emphasis on the upper mid and low highs.
I've not gotten to do a full set of measurements yet with which to do a "proper" crossover. It will definitely have to include some RLC tuned filters to tame some effects of impedance spikes in the compression driver response and a bit of EQ around 4.5 kHz but to do so for a single axis only would be daft.
I've not gotten to do a full set of measurements yet with which to do a "proper" crossover. It will definitely have to include some RLC tuned filters to tame some effects of impedance spikes in the compression driver response and a bit of EQ around 4.5 kHz but to do so for a single axis only would be daft.
I listened for an hour yesterday and have the set running behind me this morning while I finish up some work. I'm typing this out while I wait for a cuppa tea. The CD HF takes some getting used to. More HF energy but not really obtrusive. Some EQ and a lot of listening is the only way to tell.
I'm not sure which I like more, but the horns do have a more direct sound and the same time are perhaps more laid back, likely due to the rising DI and directivity well into F_shroeder. Things will only become clearer over the course of listening to these and getting them to sound right, though even now there is nothing offensive other than a smidgen of upper mid low treble EQ.
I'm not sure which I like more, but the horns do have a more direct sound and the same time are perhaps more laid back, likely due to the rising DI and directivity well into F_shroeder. Things will only become clearer over the course of listening to these and getting them to sound right, though even now there is nothing offensive other than a smidgen of upper mid low treble EQ.
Seeing as this is what a crossover 'does', and seeing that you have your tweeter built and that the design is generally accepted as good in practice I would suggest that this process leads to you finding the appropriate angle on which to listen.
Peaks in the low kHz region can be touchy, but I'd further suggest trying a downward tilt in the constant directivity region if you're looking for a natural tone.
Update
It's amazing what a coat of paint does to a raw MDF speaker box.
I'm still getting the hang of the wide dispersion of these. There is a ton of HF energy and I'm rolling off the top end ever more (perhaps I should just drop some foam in there while I am at it)
photobucket link
It's amazing what a coat of paint does to a raw MDF speaker box.
I'm still getting the hang of the wide dispersion of these. There is a ton of HF energy and I'm rolling off the top end ever more (perhaps I should just drop some foam in there while I am at it)
photobucket link
Last edited:
Here the current crossover on these. Active makes things much simpler and there isn't the continuous thought of having forfeit 3/4 of the woofer input in some network. Ignore <100Hz due to reflections.
photobucket thumbnails seem not to be working, hmmm.
photobucket link - click to view full size
photobucket thumbnails seem not to be working, hmmm.
photobucket link - click to view full size
Last edited:
I've been listening to these for a while now and they're growing on me. I think a little bit more cabinet stuffing is called for, will have to identify where the modes sit and apply some targeted damping. My little 60Wpc Rotel multi channel amp certainly gets them going.
What's somewhat odd is that unlike with the horns I don't get an exact center image, more like an area from which the sound emanates. Tonally it is extremely neutral and sometimes I find that full baffle step compensation is too much, other times it's fine.
It's certainly been a fun build and the passive crossover has been a challenge. Just to take something with sensitvity in the high 90s down to the low 90s is a bit of a pity given that one idea is to run them off a 300B SET.
What's somewhat odd is that unlike with the horns I don't get an exact center image, more like an area from which the sound emanates. Tonally it is extremely neutral and sometimes I find that full baffle step compensation is too much, other times it's fine.
It's certainly been a fun build and the passive crossover has been a challenge. Just to take something with sensitvity in the high 90s down to the low 90s is a bit of a pity given that one idea is to run them off a 300B SET.
I found that comparing the SEOS waveguide to a good horn too. That's one reason why I prefer a horn over a OS waveguide To me the SEOS sounds like a dome tweeter where a good horn sounds more present and live. Funny though, I'm now using the JBL Cinema 2384 waveguide and love the sound. Very smooth and detailed yet precise imaging and lively presence like a good horn.
- Status
- Not open for further replies.