Re: Re: Elliptical horns and vertical nulls
According to who? (Rhetorical question)
Let me ask you this. Lets say you have a device that produces no diffracton whatsoever, but response is jagged for watever reason, interference nulls, whatever.
Is this a useful device? (Another rhetorical question)
I'm not throwing stones or saying one approach is terrible and the other great. I am saying if there are competing priorities, you have to balance your goals, at least in my opinion.
doug20 said:
According to who? (Rhetorical question)
Let me ask you this. Lets say you have a device that produces no diffracton whatsoever, but response is jagged for watever reason, interference nulls, whatever.
Is this a useful device? (Another rhetorical question)
I'm not throwing stones or saying one approach is terrible and the other great. I am saying if there are competing priorities, you have to balance your goals, at least in my opinion.
Correction to post above, should have read:
"You can sometimes see some ripple within this general trend on horns that have diffraction slots. This is because the diffraction causes a discontinuity that shows up in the impedance and ultimately in the amplitude response. You can also see ripples in some horns that have other causes of changing directivity, like when a horn gains pattern control in one axis or both within the passband. This is a form of collapsing directivity also."
Sorry for the confusion. Just noticed that and couldn't edit it.
"You can sometimes see some ripple within this general trend on horns that have diffraction slots. This is because the diffraction causes a discontinuity that shows up in the impedance and ultimately in the amplitude response. You can also see ripples in some horns that have other causes of changing directivity, like when a horn gains pattern control in one axis or both within the passband. This is a form of collapsing directivity also."
Sorry for the confusion. Just noticed that and couldn't edit it.
augerpro said:
We just had a thesis posited here that if the ripples are an artifact of diffraction, which implies HOMs, and we EQ them out, the diffraction is still there, and thus, so are the HOMs.
I'm suggesting that this may be a rational means to isolate the elusive little buggers and assess their impact, and, potentially, their source, as well.
The "Here's a measurable difference with the foam in place versus without it" approach is not getting me much of substance, as conclusions rely on an unproven hypothesis as to what's actually occurring with(in) the foam. I'm still scratching my head regarding the apparent difference in response measurements shown in Earl's ALMA presentation.
xpert said:
Explain this, please, the log vs. lin part, particularly, as I'm seeing ripples everywhere, and very much desire to distinguish among them. Are my ripples bad ones, or merely innocuous ones...? 😉
Re: Re: Re: Elliptical horns and vertical nulls
lol, you are asking the wrong guy....Im the guy with the questions and no answers. You have to have the answers for me 😉
Maybe I missed something but I thought the biggest drawbacks of several waveguides is the diffraction in the throat or on the face, this is what is causing the HOM?
There are things we need to compromise, no problems there with me. Im just trying to get the facts and re-confirm the facts so I understand what is being posted.
Wayne Parham said:
lol, you are asking the wrong guy....Im the guy with the questions and no answers. You have to have the answers for me 😉
Maybe I missed something but I thought the biggest drawbacks of several waveguides is the diffraction in the throat or on the face, this is what is causing the HOM?
There are things we need to compromise, no problems there with me. Im just trying to get the facts and re-confirm the facts so I understand what is being posted.
It is an OS profile everywhere around the ellipse. I checked it at several angles to make sure it worked. I used Excel because it's easy but it was just to test the concept and draw some lines, not something to make a mold from.soongsc said:
Actually it would be quite easy to ghetto a horn that came pretty dang close to the OS shape with styrofoam and a hot wire. As the pic shows, the OS is pretty close to a straight cone, a hole drilled through the center and a gentle curve to feather the two together. So clamp a block of styrofoam between two patterns with the hot wire running through the center. The rear pattern is just a hole a little bigger than the wire. The front pattern is the shape of the mouth -- round, elliptical, whatever. Run the wire around the mouth pattern to cut out your 'cone.' Then drill a hole the size of the throat from the back side. Next, sand a smooth transition in the throat area. Finally round over the mouth.
Attachments
ZilchLab said:
I wasn't speaking to that question. Just clarifying the use of the term EQ as it relates to a CD device in the above conversation.
"Explain this, please, the log vs. lin part, particularly, as I'm seeing ripples everywhere, and very much desire to distinguish among them. Are my ripples bad ones, or merely innocuous ones...?"
You always do even when it's not a horn. Seems to me the time domain is where you would see something. CSD seems to make sense but it's value seems to be limited. You would need a very low noise enviorment to take measurements in so you could gate longer and keep resolution up. Looking at horn CSD's I have wondered many times in HOM's are hidden in them with the noise.
That's my line of thinking anyway.
Rob 🙂
You always do even when it's not a horn. Seems to me the time domain is where you would see something. CSD seems to make sense but it's value seems to be limited. You would need a very low noise enviorment to take measurements in so you could gate longer and keep resolution up. Looking at horn CSD's I have wondered many times in HOM's are hidden in them with the noise.
That's my line of thinking anyway.
Rob 🙂
augerpro said:
Yes, that's high-frequency compensation, and the contour is a conjugate of the un-EQ'd power response.
ZilchLab said:
Hi,
The ripples with PEAVEYs horn are somewhat shallow, You see some waviness, deep deep blue sky, You get tired, feel warm sun on Your face, even more tired .. You will send me all Your money and purchase that damned thing when awake .. please, we love You .. snip.
O/k, what is it all about that is the ripples perhaps. The link to cepstral analysis was a kind contribution directing attention to periodic ripples as being annoying. From that the ripples should be equally spaced (linear spacing not logarithmic) to bother the listener. Periodicity is needed to link to some speculative "Darwinistic" functions of hearing. Very spectacular speculations indeed.
The PEAVEY QT doesn't show linear spaced ripples. It ain't diffractive either. It must sound good. It is elliptical directivity wise. See that the foam is applied in total contradiction to Geddes'? At the rim only and not within the cavity! And it is heavy foam. On the other hand, that the soundwave was standing perpendicular on the walls ("... look on the graph!") anywhere is a silly notion in the paper given, quite amusing ...
I'm sorry I cant't show You the XT1086. Additionally I googled but found only PA applications that experienced severe reflections from ragged front baffles. With a HD1050 driver featuring a patented phase plug on a typical hifi-baffle it does much better. I've some minor experience with bigger to huge EV and JBL horns I own. The most annoying with them is that I burn out my hearing quite often. It's simple volume, not the hithero unknown. So said my doctor. He is able to measure the fatigue of the hearing cells, fantastique! Of course, hearing with a tired ear is annoying, HOM or not HOM.
It is time to get the Geddes waveguides from their pedestal and measure against designs that are born out of practical iterations rather than idealistic assumptions. Albeit being insulted and/or ignored viciously several times it is not meant as a personal revenge here. When I first read from OS it was really enlightning. many engineers felt the same. They engaged that concept in their work and here we are.
so long
Wayne Parham said:
Edit: oops, I had the wrong BMS driver -- it's the 4540 not the 4552. They must have used a screw-on to bolt-on adapter or something.
I've never heard anyone call the BMS 4540 harsh. Quite the opposite. Of course, I can't hear up to 20K anymore. 😉 There's clearly a resonance up there but I think it's a fundamentally different thing with their flexible annular diaphragm (ring radiator) than it would be if it were a metal diaphragm breaking up. A metal breakup would show a big spike in the 3rd harmonic at 6-7K and the BMS doesn't really show any sign of that.
An externally hosted image should be here but it was not working when we last tested it.
The sharp edges used for pattern modifiers in some CD horns are a problem, agreed. They cause discontinuities that show up as impedance ripples and ultimately ripples in response. They cause astigmatism, a different apparent source location off-axis in the vertical than in the horizontal. And they also cause diffraction, another consequence of the discontinuity. Same stuff seen in a different way, sort of like mass and weight.
What's worse, HOM or ripples in amplitude response? I don't know but I know you can see the impedance and amplitude ripples easily. Might not matter which is worse, sort of a chicken and egg argument.
Then there's the matter of interference with other sound sources. This is a summing problem, and it is (almost) unrelated. The biggest thing here is the vertical spacing, distance to the sound sources and crossover phase. You can't get two round devices as close together as you can get oval or rectangular devices of the same area, so that's a disadvantage. It squishes the forward lobe. The other thing, arguably worse, is the large vertical pattern increases ceiling reflections.
Then there are other issues too. The axisymmetrical horn loses pattern control at the same frequency in both horizontal and vertical planes. So whatever the passband is, you can be certain that pattern control is maintained in the vertical as long as it is in the horizontal. That means mouth edge diffraction can be kept largely out of band.
By going to a non-symmetrical mouth, the shorter dimension (vertical, in the cases we're discussing here) will lose pattern control first. If you crossover above that point, fine, but that decreases the null angle. And since you want the horizontal pattern to be equal to the pattern of the next adjacent driver, the crossover frequency is pretty much set within maybe an octave if using a direct radiating midwoofer and a horn/waveguide tweeter.
If you crossover below (but still above the point where pattern control is lost in the horizontal), then directivity shifts in the vertical within the passband. The mouth acts as a diffraction slot. All of this makes ripples, stronger if the transition is abrupt and wekaer if the transition is more gradual. You can curve the vertical surface to help, possibly round the edge too. This loosely describes a radial horn, CD in the horizontal and collapsing gently in the vertical.
There are a lot of factors to consider. What you do in one area generally impacts performance in another. To me, the goal is to balance these competing priorities. I want to reduce diffraction, but not at the cost of all other things. I do not want tall vertical coverage and nulls that dirty two-tirds of the pattern, for example. Nor do I want vertical control at the expense of horizontal or by creating excessive diffraction. It's a matter of balance to me. In my opinion, the ellipticals and some radials make the best compromises between smoothness of response, uniformity of horizontal beamwidth, a nice clean forward lobe with widely-spaced vertical nulls and low diffraction.
catapult said:
You could be right. Might not even be the DUT. I've seen measurement artifacts up around 20kHz, possibly something digital, possibly microphone. The digital ones I've seen were always the result of the signal used, always real-time stuff (noise, chirp, MLS, etc). So that kind of artifact goes away with a swept/stepped sine. Microphones sometimes get weird up high too.
Those might be what we were seeing on your charts, I don't know. I don't have experience with that driver. I'm familiar with Eminence, JBL and B&C drivers, where I've done most of my work.
My comments about diaphragm breakup were sort of generic, apply to anything with breakup. There's a lot of them out there. Most of the cheaper drivers have pretty strong breakup up high, and some people like the sparkle and zing that gives.
I know the early Eminence drivers had no top end and sounded dull. Later drivers had some breakup, and they actually sounded better. But more expensive drivers with diaphragms that are either damped better or made with beryllium that shifts breakup out-of-band sound much smoother to my ears.
doug20 said:
I think that people are taking the "horse to water" comment the wrong way. This is a standard phrase meaning, in this case, I can post the link to the data asked for but I can't make anyone read it. That data has been there for years, so when people say "there is no data", I was simple pointing out that there is. I did not mean an insult to Zilch by this comment as is being assumed.
The comment to Xpert was intentional, for obvoious reasons, and I won't appologize for it. His comments are ridiculous and his rudeness is intollerable so I won't discuss his errors with him.
As to HOM, ALL the data that there is can be found on my web site. If you don't accept it that's fine (not everyone accepts evolution either) but there is nothing else. Hence when I quote my position it is based on this data. If you hold another position, like HOM don't matter, then I would simple ask; "Where is your data?"
Wayne, I edited my post when I realized I had the wrong driver. (Either that or the graph is labeled wrong -- still looks like a 4552 to me.)Wayne Parham said:
I agree about most high-Q breakups. You can see that 3rd harmonic really spike at 1/3 the resonance frequency. For some reason the BMS drivers don't show that.
doug20 said:
Since directivity is the goal, I am *personally* looking for two things:
#1 - response at 45 degrees that has the same shape as the response at 0 degrees.
#2 - Response at 45 degrees that is a couple of DB lower than the response at 0 degrees.
That second one is arguable. The reason that I prefer it is that it allows you to sit off axis from one speaker, and still get a solid image. (IE, if you're closer to one speaker than the other, you still have an image.)
Based on that criteria, the two best plots in this group are the the vertical response of the XT1086, and the horizontal plot of the ring radiator on the cheap MCM waveguide.
It's LOLs that the XT1086 is better vertically than horizontally, but I'm not surprised... That stupid diffraction slot in the throat only exists in the horizontal plane. In the vertical plane, the walls of the waveguide are flat.
Patrick Bateman said:
Hi John
I'd be a bit more concerned with point #2 as I believe it to be fairly important to have a good fall off. The actual number is not as critical as the fact that it is there, but I'd say that it does have to be AT LEAST 6 dB down at 45 degrees. Otherwise you can't avoid a near sidewall reflection without etreme toe-in and then the whole concept begins to fail.
Patrick Bateman said:
Me too.
I am under the assumption that most people interested in this thread would have that in their list of most important features. I think most people interested in this topic find uniform power response and smooth polars to be very important.
The upper set of curves is my measurement of a BMS 4538 on an XT1086 (surface mounted on an 11" baffle):
The curves probably go out to 60 degrees, and I was eyeballing the angles and rotating the speaker by hand so the angle increments aren't accurate. Also, the ripples are likely to be in my measurement setup, or diffraction off the narrow baffle.
An externally hosted image should be here but it was not working when we last tested it.
The curves probably go out to 60 degrees, and I was eyeballing the angles and rotating the speaker by hand so the angle increments aren't accurate. Also, the ripples are likely to be in my measurement setup, or diffraction off the narrow baffle.
Re: Elliptical horns and vertical nulls
While the response curves of the XT1086 are a bit iffy in the horizontal plane, the vertical curves are just gorgeous. It seems to confirm that the use of an ellipitical waveguide gives the best of both worlds.
I personally don't believe the on-axis dip of the round waveguides is the "end of the world." (I don't listen on axis.)
But it does appear to be absent in the XT1086, likely due to it's shape.
So you have a couple of choices:
1 - XT1086, with curves that are CD in one axis, but not the other.
2 - Round waveguides that are CD in both directions, but have a dip.
Someone really needs to file down the neck of an XT1086 and measure it.
I'm trying to decide if it's worth $88 to find out if that will create the "ultimate waveguide", I have an XT1086 sitting here...
Wayne Parham said:
While the response curves of the XT1086 are a bit iffy in the horizontal plane, the vertical curves are just gorgeous. It seems to confirm that the use of an ellipitical waveguide gives the best of both worlds.
I personally don't believe the on-axis dip of the round waveguides is the "end of the world." (I don't listen on axis.)
But it does appear to be absent in the XT1086, likely due to it's shape.
So you have a couple of choices:
1 - XT1086, with curves that are CD in one axis, but not the other.
2 - Round waveguides that are CD in both directions, but have a dip.
Someone really needs to file down the neck of an XT1086 and measure it.
I'm trying to decide if it's worth $88 to find out if that will create the "ultimate waveguide", I have an XT1086 sitting here...
Re: Re: Elliptical horns and vertical nulls
John
I agree that the dip is not a seriuos issue. It looks worse than it ends up being in practice. If one switches the measurements to the listening axis then there is no dip. There is also no vertical lobes. Thats what I intend to do.
But an elliptical mouth will make the dip benign. It actually doesn't go away, it just broadens and becomes very shallow with eccentricity. It case you didn't see it, I posted a picture of a CNC'd elliptical mold awaiting completion (somewhere, don't rememebr). No promises on timing, but I am going to look at it.
Patrick Bateman said:
John
I agree that the dip is not a seriuos issue. It looks worse than it ends up being in practice. If one switches the measurements to the listening axis then there is no dip. There is also no vertical lobes. Thats what I intend to do.
But an elliptical mouth will make the dip benign. It actually doesn't go away, it just broadens and becomes very shallow with eccentricity. It case you didn't see it, I posted a picture of a CNC'd elliptical mold awaiting completion (somewhere, don't rememebr). No promises on timing, but I am going to look at it.
- Status
- Not open for further replies.