There is a new paper by Martin J King about OB-bass performance on his home page: http://www.quarter-wave.com/OBs/OB_Theory.html .
/Erling
/Erling
It is a nice article but I would take exception to a couple of points. First, the efficiency Martin calculates is the infinite baffle 2 Pi efficiency which doesn't reflect dipole behavior except at the dipole = monopole frequency. Changing the length of the H or U will shift that frequency up or down with result that the dipole efficiency will change as well as shown in this figure:
The figure assumes a driver with the T/S parameters for the Alpha 15A without any LP filtering.
The driver's 2Pi efficiency may not be a function of length (other than that due to changes in the air mass load) but the dipole efficiency increases by 6dB/octave as F increases, passing through the monopole = dipole frequency. Thus doubling the length shifts this frequency down an octave with result that the dipole system efficiency is increased by 6dB. As Martin correctly noted, increasing the length does shift the duct resonance downward limiting the useful range of the woofer system, however, if the desired x-o is around 125 Hz it is apparent that the correct x-o design would result in a woofer system requiring 6dB less input power to achieve the same SPL level, as well as requiring only 1/2 the driver excursion.
I would also add that damping a U is a critical part of the equation. Without the correct damping a U of a given length will function as an H dipole of the same length at low frequency and as the frequency rises the dipole nulls will “fill in”. However, Us with short length compared to the cross sectional dimension generally can not be satisfactorily damped. For this reason, damped Us should be built with the length approximately a minimum of 1.5 to 2 time the cross sectional dimension.
I would also add that the 15A is a rather low MMs driver for its size and its T/S parameters are more sensitive to air mass load than a driver like the Peeless SLS 15".
An externally hosted image should be here but it was not working when we last tested it.
The figure assumes a driver with the T/S parameters for the Alpha 15A without any LP filtering.
The driver's 2Pi efficiency may not be a function of length (other than that due to changes in the air mass load) but the dipole efficiency increases by 6dB/octave as F increases, passing through the monopole = dipole frequency. Thus doubling the length shifts this frequency down an octave with result that the dipole system efficiency is increased by 6dB. As Martin correctly noted, increasing the length does shift the duct resonance downward limiting the useful range of the woofer system, however, if the desired x-o is around 125 Hz it is apparent that the correct x-o design would result in a woofer system requiring 6dB less input power to achieve the same SPL level, as well as requiring only 1/2 the driver excursion.
I would also add that damping a U is a critical part of the equation. Without the correct damping a U of a given length will function as an H dipole of the same length at low frequency and as the frequency rises the dipole nulls will “fill in”. However, Us with short length compared to the cross sectional dimension generally can not be satisfactorily damped. For this reason, damped Us should be built with the length approximately a minimum of 1.5 to 2 time the cross sectional dimension.
I would also add that the 15A is a rather low MMs driver for its size and its T/S parameters are more sensitive to air mass load than a driver like the Peeless SLS 15".
Hi John,
I have to admit that your statement that I was calculating into 2 Pi threw me for a while. I took out the filter and my SPL was about 6 dB above your result. But then I realized that my model placed the H frame on the floor so I was taking advantage of the floor reflection. So I set the stand variable to 100 m ..... and got the same result. Opps, a mistake in the H frame worksheet. Fixed the mistake and now without the filter and with the stand height set to 100 m I am within a db or two of your result. I am not sure what your model includes so I consider this a consistent calculation. Running the other H frame yielded a similar comparison. I have uploaded a corrected H frame worksheet so you can try it and then see if we are in agreement. At some point we should compare what is included in the models which I hope will explain any other small differences.
I am not sure I agree completely with that paragraph. Making the H frame longer lowers the peak produced by the first standing wave but at the very low frequencies I still get about the same efficiency, within a db. I am thinking that this is primarily due to the build-up of the open end output due to the resonance. I do not get the 3 or 4 dB increase that your model shows, something is different in our calculation methods.
My goal was to have the U frame resonance above the crossover frequency so that damping was not required. I am not sure I see an advantage to using a longer U and then fighting to tame the fundamental standing wave by adding stuffing. Your description of how the U behaves, the dipole nulls filling in as frequency increases, matches what I have calculated.
Good comment, thanks. The 15A is a high efficiency and high Qts driver, I really like it for dipole applications. For a passive system, no EQ, I think that the 15A is an obvious choice that will work well in many different geometries.
Thanks again for the feedback,
john k... said:
I have to admit that your statement that I was calculating into 2 Pi threw me for a while. I took out the filter and my SPL was about 6 dB above your result. But then I realized that my model placed the H frame on the floor so I was taking advantage of the floor reflection. So I set the stand variable to 100 m ..... and got the same result. Opps, a mistake in the H frame worksheet. Fixed the mistake and now without the filter and with the stand height set to 100 m I am within a db or two of your result. I am not sure what your model includes so I consider this a consistent calculation. Running the other H frame yielded a similar comparison. I have uploaded a corrected H frame worksheet so you can try it and then see if we are in agreement. At some point we should compare what is included in the models which I hope will explain any other small differences.
I am not sure I agree completely with that paragraph. Making the H frame longer lowers the peak produced by the first standing wave but at the very low frequencies I still get about the same efficiency, within a db. I am thinking that this is primarily due to the build-up of the open end output due to the resonance. I do not get the 3 or 4 dB increase that your model shows, something is different in our calculation methods.
My goal was to have the U frame resonance above the crossover frequency so that damping was not required. I am not sure I see an advantage to using a longer U and then fighting to tame the fundamental standing wave by adding stuffing. Your description of how the U behaves, the dipole nulls filling in as frequency increases, matches what I have calculated.
Good comment, thanks. The 15A is a high efficiency and high Qts driver, I really like it for dipole applications. For a passive system, no EQ, I think that the 15A is an obvious choice that will work well in many different geometries.
Thanks again for the feedback,
Indeed I agree !
I challenged Brian Ding at Rythmik Audio for an experiment with his different Qts-value Servo Bass Units. In fact GR-Research
will have some units perfectly suited to the experiment. The Experiment would amount to see which unit in dipole configuration would draw the less servo feedback from the amplifier. That would also determine which unit would perform better without servo feedback. And in some way answer the Qts-question. You could also define other experiments. The link is here: http://www.audiocircle.com/circles/index.php?topic=47310.140
I suppose this is to ask too much from commercial enterprises. You have invested to much money into the ongoing production.
Some University could perhaps by units and conduct the research.
/Erling
I challenged Brian Ding at Rythmik Audio for an experiment with his different Qts-value Servo Bass Units. In fact GR-Research
will have some units perfectly suited to the experiment. The Experiment would amount to see which unit in dipole configuration would draw the less servo feedback from the amplifier. That would also determine which unit would perform better without servo feedback. And in some way answer the Qts-question. You could also define other experiments. The link is here: http://www.audiocircle.com/circles/index.php?topic=47310.140
I suppose this is to ask too much from commercial enterprises. You have invested to much money into the ongoing production.
Some University could perhaps by units and conduct the research.
/Erling
MJK said:
I'll take a look. But we won't beable to compare models directly. Back when we were colaborating on the firct U, H frame worksheet I was also encouraging the SoundEasy developer to incorporate H and U frames in SE. He did and that is the model I was using. I don't know the exact details other that it is based on Backman's model of acoustic resistance boxes.
Perhaps it is due to the simulated "listening distance"? I don't know how you change that. When I run your H worksheet I get the same SPL at the front terminus when I double the length. So if the length of the H doubles and the SPL at the terminus stays the same then when summed far from sources the dipole response should just be the point source dipole response roll off superimposed on the terminus SPL. The dipole transfer function will pass through the 0dB point at a frequency given as Fe = c/(6d) where d is the separation. Thus the dipole sensitivity at Fe will be equal to the the sensitivity of the the response at the terminus at Fe.
I agree that when I ran your H frame worksheet for L = 24 and L =12 I didn't see a change in dipole SPL. I was going to contact you about that. I don't understand why as I should increase.
Yep, if the U is undamped then at low frequency a 24" U should behave the same as a 24". But as the frequency rises the 24" U will have the 1/4 wave resonance peak at about 1/2 that of the H in addition to the lack of symmetry between front and rear SPL. The H retains the dipole response at all frequencies but the U will fill in to the sides due to the lack of front to rear symmetry. So if not damped, there is no advantage to using a U over an H if uniform polar response vs frequency is desired. But damping moves the low frequency response to a cardioid and from the front makes the on axis response of the U look like that on an H twice as long. Obviously it is very difficult to get the correct damping w/o measurements and even then perfect cardioid response at low frequency is impossible because the damping of the rear wave will also attenuate it some what even if the phase is perfect. But -15 dB is possible (as I have in the NaO II). Is we assume the phase is perfect for the -15 dB response eat the rear then this means the rear response is - 1.7 db compared to the front. When looking at the on axis response what this ultimately does is make the response a little less sensitive (by maybe a dB) and roll off a little slower with a plateau at low frequency. Hard to explain in words.
I agree, for no active eq the higher Q driver is better. Both the Peerless and the 15A have about the same mid band effeciency, but the low Q of the Peerless means a lot more eq is required around Fs.
I'll download the new worksheet and take a look. I'll try to look into this a little more with your work sheets to see what I find.
Martin, Perhaps the difference in SPL is due to consideration of the added air mass in the H or U frame altering the driver efficiency. Thus the increase in SPL expected form the greater source separation is countered by the lower driver efficiency as a result of the greater air mass load? When I look at a driver with greater moving mass like a 10" Peerless XLS is a U or H frame of smaller cross section (thus less volume) I do see the expected increase in SPL at low frequency. In this case the additional air mass is a much smaller factor.
Erling,
I think it is just two different approaches to try and get the same results. I am working (very slowly) on OB designs using low Qts drivers and EQ. I hope at some point to provide a document showing trade-offs and pro's and con's of each approach. I would not conclude at this point that one is superior to the other, I only recognize that there are different approaches.
I think it is just two different approaches to try and get the same results. I am working (very slowly) on OB designs using low Qts drivers and EQ. I hope at some point to provide a document showing trade-offs and pro's and con's of each approach. I would not conclude at this point that one is superior to the other, I only recognize that there are different approaches.
Martin,
I have interpreted John K, at least before his post here, to be in favour of low Qts drivers and EQ. I am only trying to inspire debate. But we certainly will await yours and others results. But I think that Brian Ding's statement that a higher Qts of 0.8 - 0.9 would result in better bass control with his Servo units is like telling the same story.
The resulting story will probably conclude that if available resoursces are sufficient there will be no hearable difference between the different bass approaches, only more or less difficult ways to achieve it.
/Erling
I have interpreted John K, at least before his post here, to be in favour of low Qts drivers and EQ. I am only trying to inspire debate. But we certainly will await yours and others results. But I think that Brian Ding's statement that a higher Qts of 0.8 - 0.9 would result in better bass control with his Servo units is like telling the same story.
The resulting story will probably conclude that if available resoursces are sufficient there will be no hearable difference between the different bass approaches, only more or less difficult ways to achieve it.
/Erling
skorpion said:
Really all I'm saying is that if you are committed to not applying active eq then you need to start with a suitable woofer which is one with a high Qts and appropriate Fs. The draw back of a passive system is that you will be stuck with a 3rd order roll off below the baffled Fs meaning greater gorup delay, if that matters to you.
What impact does this have w.r.t. your project 7?
MJK, I was just wondering whether your new results would have any impact on your project 7 build with the 2 alpha 15" per side and lowthers?
A secondary question is whether a combined approach (as you yourself used in your "brute force" design) might not lead to an optimal compromise.
This is not without self interest as I am trying to make 2 alphas per side an EX3 and an ft17h into a SO acceptable form with a relatively narrow baffle and effectively a 15cm frame along the sides and across the top using active EQ to boost bass.
Peter
MJK said:
MJK, I was just wondering whether your new results would have any impact on your project 7 build with the 2 alpha 15" per side and lowthers?
A secondary question is whether a combined approach (as you yourself used in your "brute force" design) might not lead to an optimal compromise.
This is not without self interest as I am trying to make 2 alphas per side an EX3 and an ft17h into a SO acceptable form with a relatively narrow baffle and effectively a 15cm frame along the sides and across the top using active EQ to boost bass.
Peter
Peter,
It definitely will impact my Project 7 design. I am planning on rebuilding the OB using a much smaller baffle and taking advantage of more of the gain in my crossover. I really need to down size to make things more mobile so I can move different speaker projects in and out, the big OBs cannot be moved by only myself. I have been contemplating using a modular OB contruction where I could use 1 or 2 Alpha's per side and different full range driver for the top. I am even thinking of building 4 H frames using the Alphas as the bass units in the system and individual smaller baffles for different full range drivers.
It definitely will impact my Project 7 design. I am planning on rebuilding the OB using a much smaller baffle and taking advantage of more of the gain in my crossover. I really need to down size to make things more mobile so I can move different speaker projects in and out, the big OBs cannot be moved by only myself. I have been contemplating using a modular OB contruction where I could use 1 or 2 Alpha's per side and different full range driver for the top. I am even thinking of building 4 H frames using the Alphas as the bass units in the system and individual smaller baffles for different full range drivers.
MJK said:
By "results" I imagine you mean FR, a snapshot in the photo album that seems to be the pervasive way of viewing things here. Take a look at group delay and better yet, distortion when comparing, then see if the results are the same. See how a driver with a more adavnced motor, cone and suspension stack up against a cheap weak motor speaker. You may be surprised. Or you may not care. Perhaps given your preference for fullrange drivers, distortions (amplitude, intermodulation, etc, etc) might not bother you as much. To each there own I suppose.
cheers,
AJ
Erling,
If the picture is what you are interested in studying, it should be easy to model and see what potential the geometry presents. First calculate an equivalent driver, wired in parallel or series, as outlined in my two drivers in one box document. Then using the equivalent driver model, use the H frame worksheet to model the actual physical H frame geometry. The results should be pretty accurate.
If the picture is what you are interested in studying, it should be easy to model and see what potential the geometry presents. First calculate an equivalent driver, wired in parallel or series, as outlined in my two drivers in one box document. Then using the equivalent driver model, use the H frame worksheet to model the actual physical H frame geometry. The results should be pretty accurate.
I read thru your new paper, thanks Martin.
Just a simple question: Why the 1 meter distance?
That's standard for box speakers but (in my experience) it seems to "trick" the mic into seeing the open baffle as bigger than it really is.
Moving the mic up close to the bass driver gives a very pretty curve that looks a lot like the published driver curve. Moving back to 2 or 3 meters does not, at all.
It seems to me that a mic to driver distance that is fairly close to, or smaller than, the dimension of the baffle front will give an FR plot with a lot more bass than the room actually hears. The mic picks up a lot more direct sound from the driver, as opposed to the wrap around from the back which cancels the lower frequencies at distance.
An issue - or not?
Just a simple question: Why the 1 meter distance?
That's standard for box speakers but (in my experience) it seems to "trick" the mic into seeing the open baffle as bigger than it really is.
Moving the mic up close to the bass driver gives a very pretty curve that looks a lot like the published driver curve. Moving back to 2 or 3 meters does not, at all.
It seems to me that a mic to driver distance that is fairly close to, or smaller than, the dimension of the baffle front will give an FR plot with a lot more bass than the room actually hears. The mic picks up a lot more direct sound from the driver, as opposed to the wrap around from the back which cancels the lower frequencies at distance.
An issue - or not?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.