on paper this 15' driver looks good, no?
http://www.seleniumloudspeakers.com/site2004/catalogo/pdf/wooferpw3_15pw3_15pw3-slf_new.pdf
http://www.seleniumloudspeakers.com/site2004/catalogo/pdf/wooferpw3_15pw3_15pw3-slf_new.pdf
The "on paper" point is applicable to many drivers we've discussed. There are so many drivers out there and I guess we are just eliminating the ones that seem innappropriate at this point, and determining some that are worth considering.
I have a couple of Selenium drivers, and my experience is that they offer outstanding value for the money. Whether these are good enough for the highest end systems, I don't know. I have only tried the cheaper models also, some of which are quite good.
The drivers you mention do seem very good. Good response, good Qts, good price. ($60 ea,!) I especially like the way the response just drops off at the high end. I have some 1505 15" Seleniums (discontinued) and they do the same thing. Made getting the crossover right a LOT easier. In fact , I don't have a low pass filter on them! These you mention, being a cheaper model are lacking the weather resistant cones which is probably good, and have asmaller voice coil usually means higher frequency response.
They don't have much x-max which could be a big problem for the output Lynn wants,(A speaker I mentioned previously had a similar problem) but for our unmarried members, I can imagine a very impressive system with at least four (maybe eight!) of these on each channel.
Yeah, with 8 of them per channel you pretty much have the surface area of an electrostatic panel, with a LOT more excursion.
Whoooeee!!!
I have a couple of Selenium drivers, and my experience is that they offer outstanding value for the money. Whether these are good enough for the highest end systems, I don't know. I have only tried the cheaper models also, some of which are quite good.
The drivers you mention do seem very good. Good response, good Qts, good price. ($60 ea,!) I especially like the way the response just drops off at the high end. I have some 1505 15" Seleniums (discontinued) and they do the same thing. Made getting the crossover right a LOT easier. In fact , I don't have a low pass filter on them! These you mention, being a cheaper model are lacking the weather resistant cones which is probably good, and have asmaller voice coil usually means higher frequency response.
They don't have much x-max which could be a big problem for the output Lynn wants,(A speaker I mentioned previously had a similar problem) but for our unmarried members, I can imagine a very impressive system with at least four (maybe eight!) of these on each channel.
Yeah, with 8 of them per channel you pretty much have the surface area of an electrostatic panel, with a LOT more excursion.
Whoooeee!!!
gedlee said:
Have a safe, enjoyable, and interesting trip! Last time I was in Bangkok was 1991, I'm sure it's changed since - although it's plenty hot-n-humid this time of year. I wish your business venture all the best - I think all of us here on the forum are looking forward to the new AI product line. All the best, Earl!
It'll probably be a while until I take another air flight, and I guess I'll be flying Business Class from now on, to avoid complications from leg circulation and lengthy sit-times in a cramped seat. I'm 6 foot 2 inches, 240 lbs, so Economy seats and a long journey are never a happy combination for me.
The latest exchange has gotten me thinking about multipath distortion in loudspeakers. The departure from minimum-phase occurs when multiple minimum-phase paths meet at the detector - either a microphone or eardrum. An inverse DSP characteristic can correct the time distortion, but unfortunately only at one very small location in space.
In fact, any kind of multipath, from any source - internal cabinet reflections and standing waves, diffraction from cabinet or baffle edges, sound traversing multiple paths inside a horn, room reflections, all produce non-minimum-phase multipath distortion when summed at the detector.
But - it makes a difference when the delayed signals arrive. If they fall in a short interval - say, 0 to 2 mSec - they can interfere with localization and timbre. Earl has found level-dependent colorations arising from HOM's in horns - multipath in the very short 0.2 mSec range is apparently particularly destructive.
Longer-period multipath (in the 3 mSec to 20 mSec range) may not only be benign, but could provide a more spacious sound. This can be verified easily enough by moving the speakers outdoors and comparing them to the listening room. If you strongly prefer the outdoor sound, that puts you in the controlled/narrow directivity camp. If you find the outdoor sound dry, with headphone-like stereo, you're probably going to be in the wide-dispersion, dipole, or omnidirectional camp.
I don't think this is matter of moral character, religion, or legislation. I just think different people hear differently. For example, a friend of mine was part of a study by a Canadian telephone company that investigated the audibility of lossy digital compression vs the native language of the speaker. They found, to the surprise of the investigators, that native speakers of some European languages couldn't accept any lossy digital compression at all, without losing intelligibility (the no-go languages weren't English). I've never seen anything like this in the published literature, but considering the different character of audio culture in Japan, China, Germany, Italy, France, Britain, and the USA, I'm not too surprised.
I've also been thinking about the baffle material - it might be interesting to have a lossy, partially acoustically transparent baffle. Think of several layers of perforated metal, separated by layers of wool-fiber felt. If you wrap this into a cylinder, it's the same construction as a gun silencer, or the interior of a muffler in a car. The gun silencer has to deal with much louder transients than anything we deal with, and must have 40~50 dB attenuation, astonishing when you think about it.
The exterior appearance of the lossy baffle would be like the classic RCA Type 44 ribbon microphone, just on a much larger scale. (The Microphones site has sound samples of many different microphones, and they sound more different than you'd expect. This site is well worth a visit, and a listen!)
OK, I know, pretty low WAF, no matter what color the perforated metal mesh is - although I guess charcoal gray would be (slightly) more tolerable.
But - the point here is a baffle that doesn't have a sharp acoustic boundary. The sound gradually diffuses through it, some travelling along the surface, some getting absorbed by the layers of felt, and some making all the way through. This complex structure could be reserved for the outer edge of the baffle, with the area directly between the drivers made of the usual layers of plywood (perhaps separated by sand, as mentioned earlier).
I'm visualizing an inner D-shaped mesh, with an outer D-shaped mesh that surrounds it. One layer of felt slides into the innermost mesh, an outer layer of felt wraps like a blanket around the inner mesh, surrounded in turn by the outer mesh. This gives the lossy baffle a smoothly rounded edge, 4 layers of mesh, and 3 layers of felt (or foam, or whatever). The drivers can be mounted on an inner wood frame, with the edge of the speaker defined by the lossy baffle. With losses this large, the outer profile of the baffle becomes less important, and could just be a straight line from top to bottom.
Hi Lynn,
Might be possible to increase WAF by making the outer layer short pile carpet, or of furnishing cover material.
A thought came to mind today relating to phase plugs and the larger full range driver whilst I was maintaining the WAF of our kitchen.
Often peakiness prior to higher AF roll-off can be due to centre of cone/dust cap related problems. Many LS DIYers have been fitting phase plugs to correct and smooth the upper response.
Last year I found I prefered reproduction with a phase plug compared to a dust cap, but I also prefered the sligthly increased reduction of hf energy without needing to use additional crossover circuitry as was introduced by a roll of wool felt where the phase plug had been.
Today's idea is to cut and use a fluffy paint roller coming out from the magnet. They come in several diameters from 1" up, and I shall be trying this myself tomorrow.
Cheers ......... Graham.
Might be possible to increase WAF by making the outer layer short pile carpet, or of furnishing cover material.
A thought came to mind today relating to phase plugs and the larger full range driver whilst I was maintaining the WAF of our kitchen.
Often peakiness prior to higher AF roll-off can be due to centre of cone/dust cap related problems. Many LS DIYers have been fitting phase plugs to correct and smooth the upper response.
Last year I found I prefered reproduction with a phase plug compared to a dust cap, but I also prefered the sligthly increased reduction of hf energy without needing to use additional crossover circuitry as was introduced by a roll of wool felt where the phase plug had been.
Today's idea is to cut and use a fluffy paint roller coming out from the magnet. They come in several diameters from 1" up, and I shall be trying this myself tomorrow.
Cheers ......... Graham.
Hi,
I lost track on the speakers but don't remember RCF MB12G301 mentioned as WR
http://www.rcf.it/vediMacro.phtml/l...wFrequencyWoofer/Mid-BassMB12G301/product.htm
"http://www.rcf.it/vediMacro.phtml/lang/IT/IDMacro/2729/menuAttivo/2293/m1/1/m2/1/m3/1_1_5/LowFrequencyWoofer/Mid-BassMB12G301/product.htm"
http://www.rcf.it/download.phtml?id=1439
"http://www.rcf.it/download.phtml?id=1439"
maybe a little bit low Qts and Xmax but good FR up to at least 2kHz, good efficiency and low Mms
Greetings
Michael
I lost track on the speakers but don't remember RCF MB12G301 mentioned as WR
http://www.rcf.it/vediMacro.phtml/l...wFrequencyWoofer/Mid-BassMB12G301/product.htm
"http://www.rcf.it/vediMacro.phtml/lang/IT/IDMacro/2729/menuAttivo/2293/m1/1/m2/1/m3/1_1_5/LowFrequencyWoofer/Mid-BassMB12G301/product.htm"
http://www.rcf.it/download.phtml?id=1439
"http://www.rcf.it/download.phtml?id=1439"
maybe a little bit low Qts and Xmax but good FR up to at least 2kHz, good efficiency and low Mms
Greetings
Michael
Hi
From the RCF L15P530 I'd really would like to see directivity and CSD
http://www.rcf.it/vediMacro.phtml/l...owFrequencyWoofer/Mid-BassL15P530/product.htm
"http://www.rcf.it/vediMacro.phtml/lang/IT/IDMacro/1627/menuAttivo/2293/m1/1/m2/1/m3/1_1_2/LowFrequencyWoofer/Mid-BassL15P530/product.htm"
http://www.rcf.it/download.phtml?id=1435
"http://www.rcf.it/download.phtml?id=1435"
Greetings
Michael
From the RCF L15P530 I'd really would like to see directivity and CSD
http://www.rcf.it/vediMacro.phtml/l...owFrequencyWoofer/Mid-BassL15P530/product.htm
"http://www.rcf.it/vediMacro.phtml/lang/IT/IDMacro/1627/menuAttivo/2293/m1/1/m2/1/m3/1_1_2/LowFrequencyWoofer/Mid-BassL15P530/product.htm"
http://www.rcf.it/download.phtml?id=1435
"http://www.rcf.it/download.phtml?id=1435"
Greetings
Michael
Woofs
Russell ... The Selenium you looked at and most other PA speakers do not go very low, -3dB is around 40hz and -6dB at 30hz in OB. Are you implying to use these as the low acceptable frequencies for OB?
Personally I'm aiming for smaller (12") P.Audio UB speakers to come in around 80hz with a very deep sub woofer (-3dB @ 20hz) and use as needed. String Quartet, probably not! Stones, of course.
Note: All OB people do not just listen to Symphony music and Celine Dion. I still love my old rock and Disco and run my 22 year old son out of the house playing Disco Duck just to counter the crap he listens to.
Another problem with PA speakers and my Pro dealer/reconer (Lynn: Jamac) says it usually takes a lot of power to get them started as they are very stiff. Low listening levels could be difficult. They are not made for low levels. Have you tested this? I have not, still building.
Zene
Russell ... The Selenium you looked at and most other PA speakers do not go very low, -3dB is around 40hz and -6dB at 30hz in OB. Are you implying to use these as the low acceptable frequencies for OB?
Personally I'm aiming for smaller (12") P.Audio UB speakers to come in around 80hz with a very deep sub woofer (-3dB @ 20hz) and use as needed. String Quartet, probably not! Stones, of course.
Note: All OB people do not just listen to Symphony music and Celine Dion. I still love my old rock and Disco and run my 22 year old son out of the house playing Disco Duck just to counter the crap he listens to.
Another problem with PA speakers and my Pro dealer/reconer (Lynn: Jamac) says it usually takes a lot of power to get them started as they are very stiff. Low listening levels could be difficult. They are not made for low levels. Have you tested this? I have not, still building.
Zene
Russell Dawkins said:
I tested the 10" here: http://s139.photobucket.com/albums/q288/augerpro/Selenium 10PW3/
For $38 I paid it's a good enough driver. But the F3 and F5 (don't have higher order measurements) were kinda high for use in a "higher end" project. Also while the cone breakup doesn't appear too bad from teh SPL response the Z plot and CSD plot show it's really very uncontrolled. Anyway that's my .02
Phase plug
Hi
I also replaced the inverse dust cup of my Dynaudio 21W54 with sort of a phase plug as I had to fix a broken voice coil wire.
FR got changed significantly :
"http://members.aon.at/kinotechnik/diyaudio/dipol/21W54_phase_plug/frequency%20_comparison.gif"
Then I wanted to check distortion at the front and back - which is always good to know in OB's - I expected to get significant lower harmonics at the back due to the absence of coughing through the pole vent but I was completely wrong.
3rd harmonics were increased
"http://members.aon.at/kinotechnik/diyaudio/dipol/21W54_phase_plug/HD-3nd%20_comparison.gif"
and especially 2nd harmonics went "through the roof" from 500 Hz up peaking at above 1%.
Notice that there is strong filtering applied.for swept HD and SPL was around 90 dB at 1m.
"http://members.aon.at/kinotechnik/diyaudio/dipol/21W54_phase_plug/HD-2nd%20_comparison.gif"
Bottom line:
I will have to repeat with a driver before and after modification as soon as I have my setup reinstalled as the above measurements are from two different speakers.
Maybe there is also an effect from changing the geometry of the original very shallow membrane to a deeper cone ( HOM ? ). I never thought this could be an issue until Earl referred about that.
Greetings
Michael
Hi
I also replaced the inverse dust cup of my Dynaudio 21W54 with sort of a phase plug as I had to fix a broken voice coil wire.
FR got changed significantly :
An externally hosted image should be here but it was not working when we last tested it.
"http://members.aon.at/kinotechnik/diyaudio/dipol/21W54_phase_plug/frequency%20_comparison.gif"
Then I wanted to check distortion at the front and back - which is always good to know in OB's - I expected to get significant lower harmonics at the back due to the absence of coughing through the pole vent but I was completely wrong.
3rd harmonics were increased
An externally hosted image should be here but it was not working when we last tested it.
"http://members.aon.at/kinotechnik/diyaudio/dipol/21W54_phase_plug/HD-3nd%20_comparison.gif"
and especially 2nd harmonics went "through the roof" from 500 Hz up peaking at above 1%.
Notice that there is strong filtering applied.for swept HD and SPL was around 90 dB at 1m.
An externally hosted image should be here but it was not working when we last tested it.
"http://members.aon.at/kinotechnik/diyaudio/dipol/21W54_phase_plug/HD-2nd%20_comparison.gif"
Bottom line:
I will have to repeat with a driver before and after modification as soon as I have my setup reinstalled as the above measurements are from two different speakers.
Maybe there is also an effect from changing the geometry of the original very shallow membrane to a deeper cone ( HOM ? ). I never thought this could be an issue until Earl referred about that.
Greetings
Michael
Hi Michael, you bring up a very good point. All cone drivers are very shallow short horns to some degree. I suspect where they are different is what proportion of radiation is coming from the area close to the VC in the 500 Hz and higher region.
If the majority of radiation is coming from the near-VC region, then a phase plug will change everything due to the different loading seen by "ring radiator" that the diaphragm has become.
It's been considered good design practice (since the Thirties) to gradually decouple the outer parts of the diaphragm in the 500~1000 Hz region, leaving the inner portion to load into the short conical horn of the rest of the diaphragm. Straight-sided drivers uses radial rings stamped into the cone, and curvilear profiles have a modest degree of inherent decoupling.
If you think of the driver as a ring radiator loading into a short, shallow horn, putting a phase plug in the middle is going to change the "horn" profile quite substantially, changing impulse response, polar pattern, and distortion.
From the perspective of multipath, the slightly off-axis (say 5~10 degrees) impulse response should also change, since there may be shading effects with a phase plug that aren't there with the dust cap. If you have access to before-and-after impulse response, that might be most interesting.
I completely agree with Earl's suggestion to measure FR at 7.5 degree intervals - I'd take it further, and look at impulse response too. It's in the impulse response domain I'd expect to see trouble in moderately off-axis regions - reflections, cancellations, and so on. If the impulse response looks completely different in each 7.5 degree window, that's a really bad sign, indicating chaotic, non-coherent radiation. Similarly, if the impulse response looks almost identical, that's a very good sign, indicating a smooth, coherent wavefront. (Hey ribbons, I'm thinking of you.)
P.S. The suggestion to cover the metal mesh with grille cloth makes sense.
If the majority of radiation is coming from the near-VC region, then a phase plug will change everything due to the different loading seen by "ring radiator" that the diaphragm has become.
It's been considered good design practice (since the Thirties) to gradually decouple the outer parts of the diaphragm in the 500~1000 Hz region, leaving the inner portion to load into the short conical horn of the rest of the diaphragm. Straight-sided drivers uses radial rings stamped into the cone, and curvilear profiles have a modest degree of inherent decoupling.
If you think of the driver as a ring radiator loading into a short, shallow horn, putting a phase plug in the middle is going to change the "horn" profile quite substantially, changing impulse response, polar pattern, and distortion.
From the perspective of multipath, the slightly off-axis (say 5~10 degrees) impulse response should also change, since there may be shading effects with a phase plug that aren't there with the dust cap. If you have access to before-and-after impulse response, that might be most interesting.
I completely agree with Earl's suggestion to measure FR at 7.5 degree intervals - I'd take it further, and look at impulse response too. It's in the impulse response domain I'd expect to see trouble in moderately off-axis regions - reflections, cancellations, and so on. If the impulse response looks completely different in each 7.5 degree window, that's a really bad sign, indicating chaotic, non-coherent radiation. Similarly, if the impulse response looks almost identical, that's a very good sign, indicating a smooth, coherent wavefront. (Hey ribbons, I'm thinking of you.)
P.S. The suggestion to cover the metal mesh with grille cloth makes sense.
Lynn Olson said:
Hi Lynn,
I have been following this thread without comment from the start. However, you have alluded to this non-minimum phase behavior before and I would like to point out that the generalization is not correct. Whether or not the summation of multipath sources results in a minimum phase response or not is dependent on the relative strength of the sources. Baffle diffraction in particular does not generally result in a non minimum phase response. I would refer you to the article by Kates, Loudspeaker Cabinet Reflection Effects, JAES May 1979 for an analysis and discussion of the cabinet diffraction problem.
Room reflections may result in non minimum phase behavior, depending of the relative strength of the direct and reflected sources. Cepsrtal analysis can be applied to determine the behavior of the room response, if necessary.
Regards,
Helm
Lynn ... You might see if this will work for walls? System Q is your forte'
Excel sheet at bottom.
http://www.saecollege.de/reference_material/pages/Low Mid Frequencies.htm
Lynn ... You might see if this will work for walls? System Q is your forte'
Excel sheet at bottom.
http://www.saecollege.de/reference_material/pages/Low Mid Frequencies.htm
I understand that probably Lynn does not refer to "minimum phase" by it's technical definition. But I think his point is that there are many issues that effect the measured data. In another thread, I have posted some test results that changed the driver phase respnse significantly without changing the amplitude response that should normally be associated with the phase change. This seems to show that a speaker system may not be a minimum phase system as we mostly assume.john k... said:
From what I can gather from Cepstral analysis, it seems mostly applied to characteristic recognition but not system identification. I wonder if anyone has seen any report that can correctly conduct identification of two systems from a Cepstral analysis?
john k... said:
Yes, Feyz showed this with back on the MAD board, with the following analysis. Note that I also backed this up with empirical results, per the following. Lynn, take a direct on axis measure of any single driver in box speaker showing diffraction. Using your MLSSA, calculate the excess phase. Remove any linear vs freq excess phase, which indicates time of flight. You will see that there is no excess phase left. All the phase in the measure, once time of flight is removed, is represented by the Hilbert transform: i.e. a speaker in a box showing diffraction is still a minimum phase system. This only applies to single drivers at a time, not multiway systems. I alluded to this in PM a few days ago.
Lynn, BTW, that study of lossy codecs and nationality dependant outcomes was the study my audio team conducted back at BNR, and which became a point in one of you SoS articles. We eventually had to bring in slavic talkers, in order to tweak the bugs out of the codec, hiccups that english talkers couldn't uncover.
Anyway, here's Feyz:
Re: Perception of diffraction
Posted By: Feyz (85.96.154.238)
Date: 10/18 3:02p.m.
In Response To: Re: Perception of diffraction (Earl Geddes)
Assume the baffle is a circular one, and there is a point source in the middle of the baffle. The mic is on axis to the point source at some distance to the baffle. The sum at the mic is this:
H(s) - K * H(s) * e^(-s*Td)
where H(s) is the frequency response of the driver, the point source. The minus sign accounts for the phase reversal occurring at diffraction at the edge. K is a constant that accounts for the reduced amplitude of the delayed diffraction signal from the direct signal, because of the extra path the diffraction signal has to travel. Td is the time delay of the diffraction signal to reach the mic after the direct signal. Rearrange this sum:
H(s) * (1 - K * e^(-s*Td))
(1 - K * e^(-s*Td)) part is the baffle diffraction signature. This is what we are interested in. The test to see if this is minimum phase or not is to look at whether all its zeros and poles are on the left hand side of the s plane. And they are, as long as K is smaller than 1; and for a baffle diffraction case K is always smaller than 1. I had given the reason why K is always smaller than 1 in the past, but I think you can see it why. You have an AES paper on calculating baffle diffraction if I am not mistaken. And the situation is still minimum phase for baffles that have shapes other than circle, like rectangular etc, but I don't really feel like going over all that again.
A caveat, if the diffraction signal had seen something on the way that caused its frequency content (including phase) to change wrt to the direct signal, then the sum may or may not be minimum phase, depending on the transfer function that changed the diffraction signals contents.
soongsc said:
Cepstral analysis of a speaker measure will allow clear identification of the delayed reflections from room boundaries. Reflections are uniquely parsed out without the same overlap with the incident, as you would find in an impulse response. The difficult part is post editing the cepstrum, to remove the first reflection and thereby provide a quasi anechoic snapshot out to the second reflection. Once accomplished, conversion back to the time domain provides a longer impulse to window, with the usual benefits in resolution and lowest valid frequency.
The latest version of SE implements cepstral analysis and editing. This will slowly become an art in itself, as users, through continued experience, slowly learn the best ways to edit the cepstrum. I have some good ideas in this, but can't determine yet how to automate it.
DDF said:
The question is if one does not know the exact response of a one system, how can one reliably un-mix reflections from the low frequency system model from the reflections? At least one must be known, which I think should be the room.
As mentioned elsewhere, SE Cepstral implementation is a work in progress. Currently it's a good learning tool. I think our thoughts should not be bond by it's current status.
Hi Lynn,
It was only after I tried the absorbent felt centre plug and tried solid plugs again that I realised the a solid plug actually seemed to be introducing a slightly edgy and phasey central horn effect, (as you suggest at the centre cone ring around the plug) even though this was still preferable sounding compared to the original dustcap.
I also preferred reproduction with the pole piece / voice coil open, and that is what made me think about damping material in the little resonant chamber so formed, which (to me) then proved better.
There is much discussion about baffle edge and room effects, but no matter, it is the time aligned composite driver response that strikes our senses first, and we realise the others effects (phase modifying affectations which are measurable via steady sine analysis) arrive later and it is these which are difficult to counter. The porous 'D' baffle should sufficiently delay and damp baffle related re-radiation affectations.
I mocked up a carpet / mesh baffle last year and it allows the driver(s) to transduce much more correctly by reducing both edge effects and softening the increased (due to being enclosed) rear pressure differential which itself introduces effects upon the driver cone. The only thing is that a porous baffle is not easy to fabricate, so I can do no more that encourage you to please try it.
Hi Michael.
Nice plots.
Cheers ......... Graham.
It was only after I tried the absorbent felt centre plug and tried solid plugs again that I realised the a solid plug actually seemed to be introducing a slightly edgy and phasey central horn effect, (as you suggest at the centre cone ring around the plug) even though this was still preferable sounding compared to the original dustcap.
I also preferred reproduction with the pole piece / voice coil open, and that is what made me think about damping material in the little resonant chamber so formed, which (to me) then proved better.
There is much discussion about baffle edge and room effects, but no matter, it is the time aligned composite driver response that strikes our senses first, and we realise the others effects (phase modifying affectations which are measurable via steady sine analysis) arrive later and it is these which are difficult to counter. The porous 'D' baffle should sufficiently delay and damp baffle related re-radiation affectations.
I mocked up a carpet / mesh baffle last year and it allows the driver(s) to transduce much more correctly by reducing both edge effects and softening the increased (due to being enclosed) rear pressure differential which itself introduces effects upon the driver cone. The only thing is that a porous baffle is not easy to fabricate, so I can do no more that encourage you to please try it.
Hi Michael.
Nice plots.
Cheers ......... Graham.