ScottG,
I was not planning to go "dipoles" for the midbasses. I wanted to go closed box. Not too familiar with dipoles.
For about the last year, I have thought line arrays would give me the type of sound I am looking for. This project is a big one, and a little pricey. I would still do it if I would be happy with the outcome. Going with you design stated above, how good would these speakers be? The reasons I want line arrays are:
1. A very big sound. I listen to a lot of live rock recordings.
2. A speaker that can handle big dynamics with ease.
3. High effeciency, I want to use very high quality, low powered amps. Maybe a Pass Labs amp.
I guess I want the bigness of a professional system, with the accuracy of the best audiophile system.
You are right about the spectral decay on the Audax driver. I don't know that I have ever seen one that good. I need to double check to be sure they are using a similar scale on the graph to make equall comparisons to other drivers. A quick look at the parameters suggests that they would work well in a closed box.
I am going to read a few of the other references you mentioned.
Thanks.
I was not planning to go "dipoles" for the midbasses. I wanted to go closed box. Not too familiar with dipoles.
For about the last year, I have thought line arrays would give me the type of sound I am looking for. This project is a big one, and a little pricey. I would still do it if I would be happy with the outcome. Going with you design stated above, how good would these speakers be? The reasons I want line arrays are:
1. A very big sound. I listen to a lot of live rock recordings.
2. A speaker that can handle big dynamics with ease.
3. High effeciency, I want to use very high quality, low powered amps. Maybe a Pass Labs amp.
I guess I want the bigness of a professional system, with the accuracy of the best audiophile system.
You are right about the spectral decay on the Audax driver. I don't know that I have ever seen one that good. I need to double check to be sure they are using a similar scale on the graph to make equall comparisons to other drivers. A quick look at the parameters suggests that they would work well in a closed box.
I am going to read a few of the other references you mentioned.
Thanks.
I think you would get all 3.
1. A -V E R Y- big sound. (and unrealistically large image-wise if you listen to far away from them, which is a problem with all arrays).
2. It will handle high dynamics with ease.
3. Should be about 96 db eff.
The Audax's don't have to be in a dipole config., nor do you need a linesource dipole bass, its just that it would be that much better to go this route with little added expense and perhaps LESS difficulty with the "build" process (because your not makeing boxes - just baffles with rear "grills").
If it were me I'd start with the Neo8's (Non-PDR, 6 per side) and the Audax's (12 per side).
6*7.9 inches for the Neo 8 = 47.4 inch vertical length (leaves 7 inches for the super tweeter latter)
12*4.6 inches for the Audax = 55.62 inches.
Additionally the Neo8's are 4 ohms and the Audaxs are 8 ohms so that they can be wired to achieve a resistive "flat" impeadance of either 6 or 12 ohms (your choice).
Cost=
12*60 = $720 (ouch)
24*14 = $336 (excellent)
For this kind of money don't skimp on your high-pass caps - get the best there is (you'll only need 2 total):
http://www.e-speakers.com/products/mundorf-sup-so.html
as reviewed here:
http://home.zonnet.nl/geenius/Cap.html
Just make the front baffle (just about 10 inches in width with the midrange line next to the Neo8 line) for the box that you would normally make and see if you like the dipole sound from the midranges - note though that freq.s below 400 Hz will roll-off rapidly. If you like the midrange your getting with this then great - if not then try enclosing it with the box you had planned originally. Of course an enclose box will allow for, (offer extension to 80 Hz), a normal subwoofer, BUT a normal subwoofer will cost as much or more than the dipole line mentioned and won't sound nearly as good).
The dipole woofer line cost:
12*25 = 300 (dirt cheap)
Latter if you want you can add the super tweeter with caps and transformer attenuator.
Its certainly NOT a cheap speaker, but I'd think that the end result would commercially be more than 10 times the cost (perhaps 20-50 times).
Amp-wise.. a DIY gainclone would work very well and be very cheap to build.
For more money I'd look here (..a bargain like the Audax drivers):
http://www.transcendentsound.com/T16OTL.htm
HOWEVER if you wire the speaker for a higher impeadance (20 ohms) then you might be able to get away with:
http://www.transcendentsound.com/single ended.htm
They have an excellent pre-amp but I'd think you would be better off with this:
http://www.6moons.com/audioreviews/audiozone/amp2.html
(the pre-T1 made by fellow member Peter Daniels.. )
though you could make one yourself with parts from these people:
http://www.bentaudio.com/index2.html
1. A -V E R Y- big sound. (and unrealistically large image-wise if you listen to far away from them, which is a problem with all arrays).
2. It will handle high dynamics with ease.
3. Should be about 96 db eff.
The Audax's don't have to be in a dipole config., nor do you need a linesource dipole bass, its just that it would be that much better to go this route with little added expense and perhaps LESS difficulty with the "build" process (because your not makeing boxes - just baffles with rear "grills").
If it were me I'd start with the Neo8's (Non-PDR, 6 per side) and the Audax's (12 per side).
6*7.9 inches for the Neo 8 = 47.4 inch vertical length (leaves 7 inches for the super tweeter latter)
12*4.6 inches for the Audax = 55.62 inches.
Additionally the Neo8's are 4 ohms and the Audaxs are 8 ohms so that they can be wired to achieve a resistive "flat" impeadance of either 6 or 12 ohms (your choice).
Cost=
12*60 = $720 (ouch)
24*14 = $336 (excellent)
For this kind of money don't skimp on your high-pass caps - get the best there is (you'll only need 2 total):
http://www.e-speakers.com/products/mundorf-sup-so.html
as reviewed here:
http://home.zonnet.nl/geenius/Cap.html
Just make the front baffle (just about 10 inches in width with the midrange line next to the Neo8 line) for the box that you would normally make and see if you like the dipole sound from the midranges - note though that freq.s below 400 Hz will roll-off rapidly. If you like the midrange your getting with this then great - if not then try enclosing it with the box you had planned originally. Of course an enclose box will allow for, (offer extension to 80 Hz), a normal subwoofer, BUT a normal subwoofer will cost as much or more than the dipole line mentioned and won't sound nearly as good).
The dipole woofer line cost:
12*25 = 300 (dirt cheap)
Latter if you want you can add the super tweeter with caps and transformer attenuator.
Its certainly NOT a cheap speaker, but I'd think that the end result would commercially be more than 10 times the cost (perhaps 20-50 times).
Amp-wise.. a DIY gainclone would work very well and be very cheap to build.
For more money I'd look here (..a bargain like the Audax drivers):
http://www.transcendentsound.com/T16OTL.htm
HOWEVER if you wire the speaker for a higher impeadance (20 ohms) then you might be able to get away with:
http://www.transcendentsound.com/single ended.htm
They have an excellent pre-amp but I'd think you would be better off with this:
http://www.6moons.com/audioreviews/audiozone/amp2.html
(the pre-T1 made by fellow member Peter Daniels.. )
though you could make one yourself with parts from these people:
http://www.bentaudio.com/index2.html
That's a good idea, building the baffle first and doing listening tests. I have no experience with open boxes. Also, adding the supertweeter and 12 inch woofers later.
The cost is not bad, except for the Neo8s. But worth it I think. I strongly agree with using a very high quality cap in the crossover. I was thinking of the Theta cap, but the Mundorf Silver looks excellent. I may try it in a speaker I am using now. Still consider using an 3/way active crossover.
Going with your full example, what are the crossover frequencies and slopes? Are the 4 inch mids playing almost full range? My understanding is that the midwoofers should not play much above 2k~2.5k because of center to center distance. Maybe crossed over even lower to make good use of the Neo8s.
Also, why do you prefer the Neo8 over the PDR version? You said the off axis response of the regular unit is excellent. Is its dispertion comparable to good dome tweeters?
Testing the open box, you said bass will roll off quickly starting about 400hz. Would this be true for the finished product, or just this test?
I have the loudspeaker cookbook. I will look to see if there is a chaper on open boxes and dipoles. If they sound better, I'm interested. I just don't know much about them. I know LinkwitzLabs likes his Orion.
Have you built any speakers lately?
The cost is not bad, except for the Neo8s. But worth it I think. I strongly agree with using a very high quality cap in the crossover. I was thinking of the Theta cap, but the Mundorf Silver looks excellent. I may try it in a speaker I am using now. Still consider using an 3/way active crossover.
Going with your full example, what are the crossover frequencies and slopes? Are the 4 inch mids playing almost full range? My understanding is that the midwoofers should not play much above 2k~2.5k because of center to center distance. Maybe crossed over even lower to make good use of the Neo8s.
Also, why do you prefer the Neo8 over the PDR version? You said the off axis response of the regular unit is excellent. Is its dispertion comparable to good dome tweeters?
Testing the open box, you said bass will roll off quickly starting about 400hz. Would this be true for the finished product, or just this test?
I have the loudspeaker cookbook. I will look to see if there is a chaper on open boxes and dipoles. If they sound better, I'm interested. I just don't know much about them. I know LinkwitzLabs likes his Orion.
Have you built any speakers lately?
As to the exact crossover freq.s - don't know (or their exact summed slope), thats somthing you'll have to figure out - but if you've built a speaker b4 chances are you have a good idea. I use LoudspeakerLab (and a calibrated mic) for my measurements to determine crossover points. (I've given you my best guestimate already.)
The 4 inch midranges won't have an electrical crossover (or perhaps only a notch filter at those high freq.s on the csd plot), so essentially they will be run full-range. In reality though they will be sloping downward similar to the line array graph "link" I provided. The concern about 2.5 kHz is with regard to comb-filtering, but
1. it isn't aggresive at this freq. (only much higher in freq.)
2. you'll already be down about 6 db in level at this freq.
(i.e. it won't be a problem - and it might even provide a natural notch filter in the "hashy" region of the midrange's top-end)
The Neo8 white paper specifically mentions line arrays in its use. In addition those that have access to both and make line arrays use the non-PDR version. Finally, the underlying reason is that the increasing spl with freq. is greater than the PDR version - which works better with arrays because they DECREASE with freq.. Now the ON-AXIS response is to "peekey" at 14 kHz but thats OK because with the outside baffle edge approach you'll be listening at anywhere between 15-45 degrees off-axis. If you look at the response in the white paper you'll notice that its at about the right spl level. Its better than dome tweeters to this level, its signigicantly worse above this (but thats the reason for the super tweeter, its also why they made the PDR version).
-"Testing the open box, you said bass will roll off quickly starting about 400hz. Would this be true for the finished product, or just this test?"
true for the finished product as well (though its a guestimate) - i.e. for the midranges only. In otherwords the response from 400 to 1kHz will be the "flat" response handled by the midranges (with acoustic summation above and below these freq.s in conjunction with other drivers).
I've got the Vance Dickson book to, nothing of value there on arrays or dipoles. About the only good source on arrays if Jim Griffins paper. Dipoles (SL's site).
Lately I haven't, a couple years back I did a mock-up using cardboard for a PE array (the 69 cents ones) with a bunch of MCM film tweeters. I had a couple of problems though.
1. I didn't know about downward tilt that arrays produced which created the problem in #2
2. The film tweeters didn't go low enough for a viable 1st order crossover with the slope problem - in otherwords I would have to have made a very complex crossover to eq. both lines to have gotten it to work.
3. I use a psychoacoutic processor that did NOT work with the dipole configuration.
4. Because the line of tweeters was a point-source array, the line required a very close listening distance.
(lessons learned I guess, but with the exception of #3, I have tried to convey a solution to all of the above problems with the design mentioned, while providing MUCH better parts.)
The 4 inch midranges won't have an electrical crossover (or perhaps only a notch filter at those high freq.s on the csd plot), so essentially they will be run full-range. In reality though they will be sloping downward similar to the line array graph "link" I provided. The concern about 2.5 kHz is with regard to comb-filtering, but
1. it isn't aggresive at this freq. (only much higher in freq.)
2. you'll already be down about 6 db in level at this freq.
(i.e. it won't be a problem - and it might even provide a natural notch filter in the "hashy" region of the midrange's top-end)
The Neo8 white paper specifically mentions line arrays in its use. In addition those that have access to both and make line arrays use the non-PDR version. Finally, the underlying reason is that the increasing spl with freq. is greater than the PDR version - which works better with arrays because they DECREASE with freq.. Now the ON-AXIS response is to "peekey" at 14 kHz but thats OK because with the outside baffle edge approach you'll be listening at anywhere between 15-45 degrees off-axis. If you look at the response in the white paper you'll notice that its at about the right spl level. Its better than dome tweeters to this level, its signigicantly worse above this (but thats the reason for the super tweeter, its also why they made the PDR version).
-"Testing the open box, you said bass will roll off quickly starting about 400hz. Would this be true for the finished product, or just this test?"
true for the finished product as well (though its a guestimate) - i.e. for the midranges only. In otherwords the response from 400 to 1kHz will be the "flat" response handled by the midranges (with acoustic summation above and below these freq.s in conjunction with other drivers).
I've got the Vance Dickson book to, nothing of value there on arrays or dipoles. About the only good source on arrays if Jim Griffins paper. Dipoles (SL's site).
Lately I haven't, a couple years back I did a mock-up using cardboard for a PE array (the 69 cents ones) with a bunch of MCM film tweeters. I had a couple of problems though.
1. I didn't know about downward tilt that arrays produced which created the problem in #2
2. The film tweeters didn't go low enough for a viable 1st order crossover with the slope problem - in otherwords I would have to have made a very complex crossover to eq. both lines to have gotten it to work.
3. I use a psychoacoutic processor that did NOT work with the dipole configuration.
4. Because the line of tweeters was a point-source array, the line required a very close listening distance.
(lessons learned I guess, but with the exception of #3, I have tried to convey a solution to all of the above problems with the design mentioned, while providing MUCH better parts.)
Hi ScottG,
While researching line arrays I came upon the following statements you made and I would like to ask for more specific supporting data and references. These go back a number of posts on this thread.
I could not recall seeing any discussion of this on the Linkwitz site. So I took another brief look and still couldn't find the information you refer too. Could you please be more specific. I also believe that tomahack (referred to in your post) is running his NEO8's in dipole mode although I could be wrong here.
I couldn't find this information regarding 600Hz either. In addition, Linkwitz doesn't use "resistive cloth" behind his drivers and IIRC neither do any of the other dipole speaker builders on this forum. Do you know of any other supporting references for "resistive cloth"?
As to the idea of limiting rearward dipole radiation in the upper frequencies: I have owned full range dipole speakers by Magnepan, Infinity (QRS) and Dayton Wright and none of these limited the high frequency rear radiation.
Needless to say you have piqued my interest. May I say that your other comments and ideas have been very helpful and I particularly thank you for the tip on the Audax driver. Very nice.
Graeme
While researching line arrays I came upon the following statements you made and I would like to ask for more specific supporting data and references. These go back a number of posts on this thread.
ScottG said:
I could not recall seeing any discussion of this on the Linkwitz site. So I took another brief look and still couldn't find the information you refer too. Could you please be more specific. I also believe that tomahack (referred to in your post) is running his NEO8's in dipole mode although I could be wrong here.
ScottG said:
I couldn't find this information regarding 600Hz either. In addition, Linkwitz doesn't use "resistive cloth" behind his drivers and IIRC neither do any of the other dipole speaker builders on this forum. Do you know of any other supporting references for "resistive cloth"?
As to the idea of limiting rearward dipole radiation in the upper frequencies: I have owned full range dipole speakers by Magnepan, Infinity (QRS) and Dayton Wright and none of these limited the high frequency rear radiation.
Needless to say you have piqued my interest. May I say that your other comments and ideas have been very helpful and I particularly thank you for the tip on the Audax driver. Very nice.
Graeme
ahh, you are lucky I've been "sparring" with Thorsten lately, otherwise I never would have picked up this thread again..
hmm, now it seems I need to go plodding through SL's site.. urrrgg (your luck is my misfortune).
upper freq. dispersion..
This is rather implicit in his design, the tweeter he uses is a monopole. By effect a monopole tweeter (even one having a fairly broad dispersion) IS limited in its dispersion. The major reason SL uses the dipole method is to reduce off-axis freq. response - primarly below 300Hz or so. Now if you use a DIPOLE TWEETER then its rear antiphase output will "never" get to cancel-out the in-phase response - i.e. dispersion is enhanced with a dipole. Moreover the extended dispersion is out-of-phase creating a "Hall of Mirrors" effect as he describes it.
While I couldn't find anything stating exactly what I mentioned, here are a few things that were mentioned:
Hall of mirrors section D:
http://www.linkwitzlab.com/frontiers.htm#L
He does suggest that dipole opperation is ideal up to 5kHz BUT doesn't explain how he will achieve that giving the limitations of baffle width (and he doesn't achieve it with any of his designs)..
General design section 3:
http://www.linkwitzlab.com/design_of_loudspeakers.htm
Yes, Tomahack does indeed appear to be "running" them in full dipole "mode".
Resistive cloth thing..
note that the "resistive cloth" can and usually will do the same thing as SL suggests - (my description was more of an aggregate of dampening material and grill cloth.)
On this page:
http://www.linkwitzlab.com/m_panel.htm
Section J.
"Fill the cabinet cavity behind the driver lightly with polyester wadding to attenuate high frequency modes."
hmm, now it seems I need to go plodding through SL's site.. urrrgg (your luck is my misfortune).
upper freq. dispersion..
This is rather implicit in his design, the tweeter he uses is a monopole. By effect a monopole tweeter (even one having a fairly broad dispersion) IS limited in its dispersion. The major reason SL uses the dipole method is to reduce off-axis freq. response - primarly below 300Hz or so. Now if you use a DIPOLE TWEETER then its rear antiphase output will "never" get to cancel-out the in-phase response - i.e. dispersion is enhanced with a dipole. Moreover the extended dispersion is out-of-phase creating a "Hall of Mirrors" effect as he describes it.
While I couldn't find anything stating exactly what I mentioned, here are a few things that were mentioned:
Hall of mirrors section D:
http://www.linkwitzlab.com/frontiers.htm#L
He does suggest that dipole opperation is ideal up to 5kHz BUT doesn't explain how he will achieve that giving the limitations of baffle width (and he doesn't achieve it with any of his designs)..
General design section 3:
http://www.linkwitzlab.com/design_of_loudspeakers.htm
Yes, Tomahack does indeed appear to be "running" them in full dipole "mode".
Resistive cloth thing..
note that the "resistive cloth" can and usually will do the same thing as SL suggests - (my description was more of an aggregate of dampening material and grill cloth.)
On this page:
http://www.linkwitzlab.com/m_panel.htm
Section J.
"Fill the cabinet cavity behind the driver lightly with polyester wadding to attenuate high frequency modes."
Hi Scott,
Thank you for your reply.
You provided these references in a very assured manner which led me to believe that you could point them out quickly. I am, however, glad that you have reviewed the Linkwitz site and responded to my post.
My take on the "hall of mirrors" discussion is that it pertains more to room placement than speaker design.
Yes, I had previously found that quote. It's on the Phoenix design page. It is an amgibuous comment. "High frequency modes" could refer to room modes, which would support your statements. However, I took it to mean that the primary purpose of the light stuffing was to reduce cavity resonances in the baffle and perhaps mitigate diffraction effects.
In summary it sounds like neither of us was able to find any direct theoretical or experimental data on the Linkwitz site that supports the design principle of deliberately attenuating the rear propagation of high frequencies in OB speakers. I will admit that there are a number of ambiguous implications. The fact that Linkwitz does not use a dipole tweeter arrangement is insufficient to discredit the idea. Regarding Linkwitz not using a dipole tweeter arrangement: My guess - and I am guessing - is that high frequencies attenuate faster than mids and lows and by the time the reflected HF sound gets to the listener it is attenuated too far to make a valuable contribution to the "image". I find it difficult to believe that the onset of this effect - again I'm guessing (maybe theorizing is a more upscale term) that this would be as low as the xover point in Phoenix or Orion. The 5000Hz that you quote is way above the tweeter crossover point.
I find it interesting to note that the amount of effort and detail that Linkwitz puts into his designs seems to rise steadily as frequency drops.
I remain unconvinced that there is any proven design principle dictating that the HF rear sound propagation in a dipole should be deliberately attenuated. As I have stated before there is ample evidence that DIY and commericial builders have not done so. That being said I will give a couple of examples supporting your argument. First, the Magnepan tweeter: Its open on the front and the rear wave has to exit through slots in the chassis. Does this attenuate the back wave? I don't know. Second: The Infinity QRS and high-end IRS products had rear firing rows of EMIT tweeters connected out of phase with the front rows. However, there were only half as many on the back. Was this a deliberate attempt at attenuation or was Arnie Nudell saving a few pennies. I don't know.
Once again thank you for the response. I'm glad you spotted my post. This is an old thread and things seem to get to page 2 here distressingly quickly.
Cheers,
Graeme
Thank you for your reply.
ScottG said:
You provided these references in a very assured manner which led me to believe that you could point them out quickly. I am, however, glad that you have reviewed the Linkwitz site and responded to my post.
My take on the "hall of mirrors" discussion is that it pertains more to room placement than speaker design.
ScottG said:
Yes, I had previously found that quote. It's on the Phoenix design page. It is an amgibuous comment. "High frequency modes" could refer to room modes, which would support your statements. However, I took it to mean that the primary purpose of the light stuffing was to reduce cavity resonances in the baffle and perhaps mitigate diffraction effects.
In summary it sounds like neither of us was able to find any direct theoretical or experimental data on the Linkwitz site that supports the design principle of deliberately attenuating the rear propagation of high frequencies in OB speakers. I will admit that there are a number of ambiguous implications. The fact that Linkwitz does not use a dipole tweeter arrangement is insufficient to discredit the idea. Regarding Linkwitz not using a dipole tweeter arrangement: My guess - and I am guessing - is that high frequencies attenuate faster than mids and lows and by the time the reflected HF sound gets to the listener it is attenuated too far to make a valuable contribution to the "image". I find it difficult to believe that the onset of this effect - again I'm guessing (maybe theorizing is a more upscale term) that this would be as low as the xover point in Phoenix or Orion. The 5000Hz that you quote is way above the tweeter crossover point.
I find it interesting to note that the amount of effort and detail that Linkwitz puts into his designs seems to rise steadily as frequency drops.
I remain unconvinced that there is any proven design principle dictating that the HF rear sound propagation in a dipole should be deliberately attenuated. As I have stated before there is ample evidence that DIY and commericial builders have not done so. That being said I will give a couple of examples supporting your argument. First, the Magnepan tweeter: Its open on the front and the rear wave has to exit through slots in the chassis. Does this attenuate the back wave? I don't know. Second: The Infinity QRS and high-end IRS products had rear firing rows of EMIT tweeters connected out of phase with the front rows. However, there were only half as many on the back. Was this a deliberate attempt at attenuation or was Arnie Nudell saving a few pennies. I don't know.
Once again thank you for the response. I'm glad you spotted my post. This is an old thread and things seem to get to page 2 here distressingly quickly.
Cheers,
Graeme
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.