Soongsc
Just so people know, I, for one, do not concur with your assesment in that thread. It is a guess without much supporting evidence. I think my reference presents a better case.
About the EM discussion - thanks, I completely concur that the issues are not new to microwave antenae. And the problems with dispersion - aka group delay - appear to be quite comparable.
Just so people know, I, for one, do not concur with your assesment in that thread. It is a guess without much supporting evidence. I think my reference presents a better case.
About the EM discussion - thanks, I completely concur that the issues are not new to microwave antenae. And the problems with dispersion - aka group delay - appear to be quite comparable.
gedlee said:
Hi,
HOM is an attention catching feature to advertise the OS waveguides as being problem solvers, high end, unique etc. The foam makes it hard to reproduce the concept elsewhere.
Unfortunately HOM exist. They can be looked at as deviation from pure axial wave propagation along the horn. But that is trivial. At last at the mouth all that theories crash. The fancy coordinate transformation that made Geddes solve the wave equation within the waveguide lacks validity outside of it. So we don' t know what happens when the wave leaves the horn.
But I can tell You with ease. The wave undergoes diffraction. That leads to interference with the axial part of the wave. Due to that it generates excessive group delay and ripples, also called HOM - as far as Geddes' competitors are addressed.
Foam doesn't help much, as the major part of HOM is generated at the outer circumference of the OS. And if there was some HOM inside the horn due to imperfections anywhere, it couldn't be attentuated effectively with that foam. The idea that HOM are waves that are bouncing through the waveguide like a ping pong ball is silly. And even if, the pathway would be about 150% longer. Given a dampening of 3dB to the axial wave the dampening to the "bouncing" waves would be something about 1dB more. Damp the major part by 3 dB to achieve an additonal smoothing of 1dB. Not that good of a deal, nasty.
You see, HOM is to some extent inevitable. The trick that people at ALTEC, EV, JBL have excersized for so long now is to smear that HOM thing all over the spectrum and directions below the threshold of perceptability. The goal is to end up with a horn/Waveguide whatever that is reasonably flat in all directions without compromizing efficiency to much. Recently the efficiency has been dealt with a bit caused by improved drivers. That let to devices as 18Sound XT1086 which is a tremendous competent waveguide. The Geddes bowls dont't come close to them in respect to smoothness on/off axis amplitde and group delay aka HOM-wise. Yeah, You got it right: the 18Sound XT1086 is less contaminated by HOM as the Geddes "OS". Same with PEAVEYs Quadric with foam at the walls, that I can't obtain unfortunately.
If Geddes was as competent on HOM as he ever claims to be, he could prove my assertion (if false, at will). Alas as far as I followed this and other discussion I don't expect to much. I'm afraid I have to face some ignorance once again.
Zilch is no horse - see the citation above. Geddes is not that titan that he would like to apear. Look close, he's just a man with one invention that he made.
so long
I think the issue is complicated and there are so many things mixed in the data. I don't think any single way of looking at data can tell very specifically . For example, in my own wave guides, there is less intersection between the SPL at various angles, but since the driver itself has a resonance that normally results in very nasty sound at certain times, I cannot say for sure if what I hear is the driver itself or HOMs until I solve the driver resonance at 20KHz. Of course, I am also starting too blow the impulse data up and see if I can find anything consistent that could be related directly with HOMs. Foam acts like a filter, so from the data in your ALMA presentation basically shows that. It is not conclusive indication of HOM either. But I also know that we all have to keep somewhat from revealing too much information. I am also looking at other parts of the impulse for trails. But only until various other design issues change will we find consistent trails which lead to HOMs.gedlee said:
I searched this thread and couldn't find any reference to cepstral analysis. Has anybody tried using this tool to evaluate HOM ?
If there is anybody not familiar with this technique, here is a link that makes easy reading.
http://www.libinst.com/cepst.htm
Peter
If there is anybody not familiar with this technique, here is a link that makes easy reading.
http://www.libinst.com/cepst.htm
Peter
I'm sure if you show some work with good explanations, the discussion would be very smooth.xpert said:
catapult said:
gedlee said:
catapult said:
gedlee said:
This is the type of crap that makes me take a set back from purchasing anything from you. I would like to dissociate the business transaction from the individual, and I tend that way every time I read a very informative post of yours (which happens often) - ironically. But to characterize a response to a challenge you issue as a "substantial modification" is disingenuous at best. It's a shallow attempt to discredit an FYI post, for what reason(s) allude me. Understandably, you're in a defensive mode, some people clearly have a personal issue with you. I'm sure you have the ability to differentiate. Take the time.
dfdg
I'm sorry, but I don't see what you see. To me the third paragraph above is a "substantial modification" of the first. The third will work (with some further slight "modifications"), the first won't - thats pretty "substantial". AND I acknowledged that the "modification" was quite insightful!! I think your are being very unfair here.
I'm sorry, but I don't see what you see. To me the third paragraph above is a "substantial modification" of the first. The third will work (with some further slight "modifications"), the first won't - thats pretty "substantial". AND I acknowledged that the "modification" was quite insightful!! I think your are being very unfair here.
I'm not sure what you are upset about. But when I look at the explanation on the Excel calculation, it does not seem to me it's an OS expansion profile. Even if it did fit the OS profile, CNC would be the only way that I know of to make it, there are lots of consideration about data point resolution and cutter offset issues to consider if you rely on Excel instead of a normal CAD program. Not something easily educated over the NET.dfdg said:
catapult said:
Catapult if you are going to call out Zilch then be fair and call out Geddes when he holds no punches with his blanket insults!!
gedlee said:
Geddes if you want people to keep on topic and debate why not stop posting what I think are extreme ego based remarks?
Do you really think you are better,smarter or remotely more succesful then everyone else?
Your research, experience, data and minor usage of subjective opinion speaks for itself. You have zero reason to insult us!
People are always going to disagree because that is the way the world learns and evolves.
And I want to continue to learn from this thread instead of it going into mud slinging mode.
Interesting POV
Any measurements that we can look at to compare the 1086 vs Geddes "Bowls"?
Also what is "PEAVEYs Quadric with foam at the walls"?
I think even if you could use it, it would take lots of data collection using different drivers on the same horn/waveguide to collect a common recognition pattern. It would be interesting. I might recommend that as a thesis for students to explore.PLB said:
Elliptical horns and vertical nulls
You know, it seems to me the ellipticals on the market now would be interesting to many of you. There are already products out there with 90°x60° patterns, or thereabouts. If you want this kind of device, seems to me that's the best compromise. It has lower vertical coverage than a 90° round WG, which reduces ceiling slap. I'm sure you can design a system that puts the vertical nulls out at the edge of the pattern, or close. Gives the best of both worlds.
I understand the competing priorities, and I think many of you do too. I choose to focus on the vertical nulls a little more than HOM, because I think the dirty pattern outside the nulls and the ceiling slap are problems worth focusing on. Some others, Geddes included, think the HOM are more offensive than anything else. I don't disagree that they are offensive, espcially since I never liked the sound of horns with diffraction slots and sharp edges on the mouth. But to tell the truth, I always saw those as discontinuities that caused impedance and response ripple, that being the main audible problem and the diffraction being an unwanted secondary issue. Still whichever it is, the diffraction isn't wanted. I've always dealt with that using flares that don't have sharp edges, and my favorites have radiused mouth terminations too.
The elliptical waveguide promises to be a good device that would appeal to both camps, those that care about the vertical pattern and those that don't so much. The elliptical waveguide can be made with a pattern that has enough asymmetry to reduce ceiling slap and a shape that makes putting the vertical nulls out near the edge of the pattern. The nulls might not be outside, might be inside a little, but near the edge. At least it's not cutting the pattern in half or two-thirds. Most of the pattern can be made clean, with a nice forward lobe. And the shape of the waveguide would not compromise the wave shape. It would not cause excessive internal reflections.
I think that is where the compromise is, you are looking for wavefront propogation to be uniform, without diffraction, and you are also looking for a pattern that stays within the nulls as much as possible. I think this is the shape that balances those two competing priorities the best, at least in the realm of the OS/PS waveguides. Seems like this would be an attractive solution for those that are contemplating round waveguides now. All the upsides, none of the downsides.
You know, it seems to me the ellipticals on the market now would be interesting to many of you. There are already products out there with 90°x60° patterns, or thereabouts. If you want this kind of device, seems to me that's the best compromise. It has lower vertical coverage than a 90° round WG, which reduces ceiling slap. I'm sure you can design a system that puts the vertical nulls out at the edge of the pattern, or close. Gives the best of both worlds.
I understand the competing priorities, and I think many of you do too. I choose to focus on the vertical nulls a little more than HOM, because I think the dirty pattern outside the nulls and the ceiling slap are problems worth focusing on. Some others, Geddes included, think the HOM are more offensive than anything else. I don't disagree that they are offensive, espcially since I never liked the sound of horns with diffraction slots and sharp edges on the mouth. But to tell the truth, I always saw those as discontinuities that caused impedance and response ripple, that being the main audible problem and the diffraction being an unwanted secondary issue. Still whichever it is, the diffraction isn't wanted. I've always dealt with that using flares that don't have sharp edges, and my favorites have radiused mouth terminations too.
The elliptical waveguide promises to be a good device that would appeal to both camps, those that care about the vertical pattern and those that don't so much. The elliptical waveguide can be made with a pattern that has enough asymmetry to reduce ceiling slap and a shape that makes putting the vertical nulls out near the edge of the pattern. The nulls might not be outside, might be inside a little, but near the edge. At least it's not cutting the pattern in half or two-thirds. Most of the pattern can be made clean, with a nice forward lobe. And the shape of the waveguide would not compromise the wave shape. It would not cause excessive internal reflections.
I think that is where the compromise is, you are looking for wavefront propogation to be uniform, without diffraction, and you are also looking for a pattern that stays within the nulls as much as possible. I think this is the shape that balances those two competing priorities the best, at least in the realm of the OS/PS waveguides. Seems like this would be an attractive solution for those that are contemplating round waveguides now. All the upsides, none of the downsides.
I grab this jpeg on the XT1086 from another thread.....
to me the XT1086 looks decent but its definitely not flat like the OS waveguide, is there where EQing comes into play?
The MCM/ring radiator tweeter measurements at the bottom are very interesting...jagged like the JBL but obviously no horn sound.
An externally hosted image should be here but it was not working when we last tested it.
to me the XT1086 looks decent but its definitely not flat like the OS waveguide, is there where EQing comes into play?
The MCM/ring radiator tweeter measurements at the bottom are very interesting...jagged like the JBL but obviously no horn sound.
Looks like the top set of curves has CD equalization and the next set doesn't. I'll bet if you provide the requisite 6dB/octave slope above mass rolloff to correct rthe driver's power response, the second set of curves look real nice. Except for that breakup at the top end, ouch! (But that's the driver, not the horn)
doug20 said:
Thanks,
The PEAVEY is their "Quadric Throat" model that measure quite well.
http://aa.peavey.com/downloads/pdf/qwp1.pdf
Nice reading. Nearly all in it is compromize and heuristics! Ripples smaller than ~ +/- 1dB, uniform directivity spatial and over frequency too. The periodicity of ripples is log not lin, so they won't occure due to internal reflection. It should all in all perform better than any upper most critical treshold for "HOM".
I have no curves to show for the XT1086. The box is taken apart for finishing, sorry. Ripples are below +/-1dB and are not periodic in any respect.
I liked the link that "PTB" gave:
http://www.libinst.com/cepst.htm
Nice reading too. From that "HOM" could be established - if seen as internal/mouth reflections - as a measurable variable. But what would it tell?
cheers
btw: is there any third party measurement of Geddes' waveguides?
EQing would be necessary unless the shape is not exactlly the OS profile to the lip.doug20 said:
The ring radiator curves look interesting, because Thiel speakers seem to use ring radiating for mid range. I would not expect the jaggedness of the curves to be offending. I suspect it's like mild case of comb filter effect due to the ring shape.
Wayne Parham said:
I agree that the top does look EQed, any non-EQed measurements of the OS waveguide anywhere?
I wish the XT1086 was measured with the BMS4552 instead or even the DE250 to get indentical comparisons. Maybe even EQed.
Re: Elliptical horns and vertical nulls
Which products are going to the least offensive diffraction though? That seems to be the biggest issue.
Wayne Parham said:
Which products are going to the least offensive diffraction though? That seems to be the biggest issue.
doug20 said:
Doug he's referring to eq'ing the low end gain caused by the horn loading. We just want to apply EQ (i.e. a cap and maybe shunt resistor) so the level from say 2-20khz is flat like it would be when you designed a speaker with it. We're not referring to EQ like a 31 band graphic to actually smooth out the response. That would make comparing different drivers or horns kind of pointless.
Any horn or waveguide that provides constant directivity (or even close) will need 6dB/octave augmention above the driver's mass rolloff point, usually around 3-4kHz for a 1" compression driver.
This is because the horn itself doesn't add any acoustic EQ. Some horns (exponential and tractrix) have collapsing directivity, so have a little acoustic EQ as a result. But CD horns do not.
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 an horn gains pattern control in one axis or both within the pattern. This is a form of collapsing directivity also.
Those sources of ripple are largely what this discussion is all about.
You can also see spikes at the top end from diaphragm breakup. Sometimes it is enough to boost the top octave enough that full 6dB/octave isn't required. In my experience, drivers with that behavior sound harsh, even if the compensation is dialed back enough to tone down the top end. But not everyone finds breakup in the top end to be a bad thing, so consider my comments in this regard to be a subjective opinion. Objectively, diaphragm breakup adds energy up high because of resonant ripples along the surface of the diaphragm.
However, the general trend is there in all CD horns. The sound within the pattern tracks the power response of the driver. It falls about 6dB/octave above mass rolloff, around 3-4kHz. After smoothing the ripples from diaphragm breakup (generally up high) or changing directivity as the horn gains control (usually down low or worse case, lower midband), the average curve should show a 6dB/octave downward slope above 4kHz on any constant directivity horn or waveguide.
If you see an on-axis response curve from a CD horn or waveguide that's flat, it has already had CD equalization applied. If it droops at 6dB/octave, no EQ has been applied. You can pretty much just add or subtract this first-order filter slope to any curve you see, mentally, if nothing else, to compare EQ'ed curves with non-ED'ed curves.
This is because the horn itself doesn't add any acoustic EQ. Some horns (exponential and tractrix) have collapsing directivity, so have a little acoustic EQ as a result. But CD horns do not.
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 an horn gains pattern control in one axis or both within the pattern. This is a form of collapsing directivity also.
Those sources of ripple are largely what this discussion is all about.
You can also see spikes at the top end from diaphragm breakup. Sometimes it is enough to boost the top octave enough that full 6dB/octave isn't required. In my experience, drivers with that behavior sound harsh, even if the compensation is dialed back enough to tone down the top end. But not everyone finds breakup in the top end to be a bad thing, so consider my comments in this regard to be a subjective opinion. Objectively, diaphragm breakup adds energy up high because of resonant ripples along the surface of the diaphragm.
However, the general trend is there in all CD horns. The sound within the pattern tracks the power response of the driver. It falls about 6dB/octave above mass rolloff, around 3-4kHz. After smoothing the ripples from diaphragm breakup (generally up high) or changing directivity as the horn gains control (usually down low or worse case, lower midband), the average curve should show a 6dB/octave downward slope above 4kHz on any constant directivity horn or waveguide.
If you see an on-axis response curve from a CD horn or waveguide that's flat, it has already had CD equalization applied. If it droops at 6dB/octave, no EQ has been applied. You can pretty much just add or subtract this first-order filter slope to any curve you see, mentally, if nothing else, to compare EQ'ed curves with non-ED'ed curves.
- Status
- Not open for further replies.