[MJK]

I am curious if this was simply just a model or if you had built and measured one.

John,

My results are from my MathCad models. I have modeled dipole designs and correlated the results using measurements taken after building the systems. Others have done the same using my worksheets. I consider the calculated results to be validated based on these correlations.

I think we have both layed out good arguements with respect to the prefered driver properties for an OB system. I have read your detailed response and understand your position. But in the end, I still believe that the less expensive Eminence Alpha 15A and a couple of the Goldwood drivers will work better in an OB. At this point I do not see me changing direction with respect to driver selection in my next OB design.

Martin

[John_E_Janowitz]

However there is one major point I wish to pick up on - efficiency/sensitivity.
Yes parallel AE drivers will perform well on OB with the more sensitive widerange drivers, but they will need greater,and thus separate amplification of say 250W, whereas 2x Alpha will be fine with just a single 50W amplifier driving everything.

You are correct that they will need more power, but don't necessarily need a separate amp.  If you had a single 250W amp for a channel and are running a pair of woofers in parallel for 4ohm and an 8ohm mid/tweeter you could be just fine.  When doing a passive xover it's ok to have slightly more efficiency in the mid/tweeter as you can attenuate those.  You can't attenuate the woofer though if it ends up being more efficient.  Many we have worked with are using multiple amp channels and something like the DCX2496 for processing.  This gives you a lot more flexibility.  Another nice thing is that in the active system you can passively flatten the impedance curve first.  Then work with your xover and EQ.  You can end up with a nearly resistive load to each channel of amplification. 

I agree with Martin that the low Fs is wasted - and a similar AE driver would actually perform better on an OB with Fs a little higher, say with a lighter cone.  If LF power is available I would suggest choosing a driver with an Fs circa 25Hz;  if not 30 to 40Hz is a good start.
I also agree with Martin that once you get to 15" you can get satisfactory 'in-room' LF listening levels without needing more than say 4mm of X.max, especially with paralleled drivers.  Of course we need to be realistic about desired as opposed to satisfactory listening levels.

Again, the IB15's weren't originally intended for open baffle use.  They are intended for infinite baffle subwoofer installations.  Using 4-8 drivers in a manifold with 150W per driver will give you extreme levels down to the mid teens with no need for EQ to boos the low end.   It just happens that with the Q, power handling, and displacement capability at a very reasonable cost, many have picked them up for this application.  Being able to take them up to as much as 500hz cleanly has some benefits as well. 

In reality though this is why we made the OB15.  It has 80grams less mass then the IB15, about 4dB more efficiency, and a stiffer spider to put the Q at the proper level.  This was more designed to meet the need you're mentioning. 

While 4mm of excursion may meet the requirements for some, for others it will not come close.  This again is part of the DIY process, determining what your needs are.  If you want concert level output and ability to reproduce low bass guitar notes, you need much more.  Again the other thing is to look at what happens to the driver parameters at that 4mm Xmax.  If the Bl and compliance were linear there would be a large difference from what happens in the real world.  Models suggest that as you increase power, the output just increases.  In reality your response curve changes greatly and the low end output begins to compress first.  If you want to listen at a loud average listening level, you need more excursion available for the peaks and dynamic content.  Headroom is always a good thing.  Again, how much you need is up to you.

Further to this discussion, I would suggest that the AE driver sound will not integrate as well as an Alpha with a widerange on an OB, and that using 2 in parallel will not improve this aspect.

I'm curious as to the basis of this.  The Dipole15, IB15, OB15, etc all have a much higher available bandwidth than the Alpha15A.  They have significantly lower inductance and distortion so they can cover a wider band to begin with.  I'm not sure how lower distortion, higher bandwidth, more linear inductance, etc would possibly cause any issues with integration vs the Alpha15.

John

[jonasz]

I'm curious to see this explanation also as I can't see ANY andvantage to the Alpha-15A over the IB15... 

[Graham Maynard]

Explanation comes via doing and listening, because SPL plots tell you zilch about acoustic phase and relative group delays.

[John_E_Janowitz]

My results are from my MathCad models. I have modeled dipole designs and correlated the results using measurements taken after building the systems. Others have done the same using my worksheets. I consider the calculated results to be validated based on these correlations.
Martin

Hi Martin,

I'm going to be putting together a full dipole system soon.  I'll be using the pair of Dipole15's for the midbass section, our new 6.5" midrange driver, and possibly a B&G Neo3 for the tweeter section with the back cup removed.  I'm planning a flat front baffle with wings going back on each side.   A U baffle from above.  If looking at the pair from the center of the room, the outer vertical edges of the baffle will have a 3" radius and the inner will have a 6" radius.  Total depth will extend to about 15" on the bass section.  Is this something you can accurately model?  If so, can you tell me at what distance I would need to place the mic to get measurements to correlate with your models?

John

[AK]

I agree with Martin's choise of woofers. System is designed for music listening in regular home environment.
All 15in 20mm xmax woofers are way overkill for home stereo, unless you have really huge living space.
Only reason I would get hi end 15in woofer is it's attractive looks. If speaker is used in main living room, it must look good.

[John_E_Janowitz]

I agree with Martin's choise of woofers. System is designed for music listening in regular home environment.
All 15in 20mm xmax woofers are way overkill for home stereo, unless you have really huge living space.
Only reason I would get hi end 15in woofer is it's attractive looks. If speaker is used in main living room, it must look good.

Well, again i think this is preference and really depends on what you're going for.  If you follow Lynn Olson's Beyond the Aerial thread, he's considering 2-3 Dipole15 per channel currently.  http://www.diyaudio.com/forums/showthread.php?postid=1619154#post1619154

What kind of driver you need is clearly defined by the clean output levels you want to achieve.  This formula back calculated from a Linkwitz formula tells how much displacement you need for a given SPL at a given frequency:

Max SPL (at freq f) = 48.4 + 20 log (Vd x f^2), where Vd is the Sd x peak-peak Xmax in liters.

Plugging in 30hz and saying you want 110dB SPL you get 1.3358L of displacement required.  Now say the baffle you are choosing makes your response 10dB down at 30hz.  You now need to calculate for 120dB levels.  This requires 4.22L of displacement to reach your goals on the given baffle.  Each Dipole15 at 12mm Xmax displaces 2.06L peak to peak, so a pair on this baffle would roughly reach this goal.  This keep in mind is 110dB at 1m.  If you want 110dB at 2m, 4m, etc you are subtracting 6dB per doubling of distance and upping the displacement requirements.

Again the key too is that you can reach these levels while BL and compliance remain generally constant.  Reaching 12mm the Dipole15 retains between 90-95% of the rest BL value meaning the parameters are close to the same at 12mm as they are at rest.  If you have a driver that varies greatly in parameters as it gets to high excursion, things quickly get out of whack.   

John

[BrassEar]

using multiple amp channels and something like the DCX2496 for processing.  Another nice thing is that in the active system you can passively flatten the impedance curve first.  Then work with your xover and EQ.  You can end up with a nearly resistive load to each channel of amplification. 

Interesting. How does one passively flatten the impedance curve first?

TIA

[John_E_Janowitz]

Interesting. How does one passively flatten the impedance curve first?

There are two things to flatten, depending on the driver.  The one is the impedance peak at resonance.  This would require an LCR filter to flatten out.  You'd then implement a zobel network to compensate for the inductive rise going up higher in frequency. 



The RC at left is the zobel and the CLR at right is the notch filter. 

Here's a good page explaining the zobel calculations:
http://www.trueaudio.com/st_zobel.htm

Here is some info on the notch filter.  http://www.carstereo.com/help/Articles.cfm?id=23
John

[MJK]

I'm going to be putting together a full dipole system soon.  I'll be using the pair of Dipole15's for the midbass section, our new 6.5" midrange driver, and possibly a B&G Neo3 for the tweeter section with the back cup removed.  I'm planning a flat front baffle with wings going back on each side.   A U baffle from above.  If looking at the pair from the center of the room, the outer vertical edges of the baffle will have a 3" radius and the inner will have a 6" radius.  Total depth will extend to about 15" on the bass section.  Is this something you can accurately model?  If so, can you tell me at what distance I would need to place the mic to get measurements to correlate with your models?

John,

I am not sure I completely understand your geometry, the different radius values for the vertical edges has me confused. When I model an OB that has side wings extending back on each side I try to bracket the problem. At one extreme I model only the front baffle which is the smallest that the OB could be and then I model the front and the sides as one larger flat baffle which is the biggest the OB could be, the answer is probably between these two cases. With my personal copies of the OB worksheets I can model a general OB shape with as many sides as required to represent the design. The angular location and distance to the microphone is entered by the user, it is not at a fixed position.

Martin

[MJK]

What kind of driver you need is clearly defined by the clean output levels you want to achieve.  This formula back calculated from a Linkwitz formula tells how much displacement you need for a given SPL at a given frequency:

Max SPL (at freq f) = 48.4 + 20 log (Vd x f^2), where Vd is the Sd x peak-peak Xmax in liters.

Plugging in 30hz and saying you want 110dB SPL you get 1.3358L of displacement required.  Now say the baffle you are choosing makes your response 10dB down at 30hz.  You now need to calculate for 120dB levels.  This requires 4.22L of displacement to reach your goals on the given baffle.  Each Dipole15 at 12mm Xmax displaces 2.06L peak to peak, so a pair on this baffle would roughly reach this goal.  This keep in mind is 110dB at 1m.  If you want 110dB at 2m, 4m, etc you are subtracting 6dB per doubling of distance and upping the displacement requirements.

John,

I think we have had a similar discussion before. I am not doubting your formula or calculations but we should all recognize that the results are correct if a pure 30 Hz sine wave is being reproduced. For a music application where the bass is composed of many different low frequencies I am not convinced that the summed displacement for a given volume level is as high as you are reporting. I have played my Lowther OB system at very high volume levels and the Eminence Alpha 15A woofers barely move even for bass heavy material. A system playing music that hits 110 dB levels is probably not all bass frequencies. I still believe that these types of calculations are extremely conservative and that the maximum deflections caused by most music playing at these volume levels will be much lower for a bass driver(s) crossed over to a mid range driver at about 200 Hz.

Martin

[John_E_Janowitz]

I think we have had a similar discussion before. I am not doubting your formula or calculations but we should all recognize that the results are correct if a pure 30 Hz sine wave is being reproduced. For a music application where the bass is composed of many different low frequencies I am not convinced that the summed displacement for a given volume level is as high as you are reporting. I have played my Lowther OB system at very high volume levels and the Eminence Alpha 15A woofers barely move even for bass heavy material. A system playing music that hits 110 dB levels is probably not all bass frequencies. I still believe that these types of calculations are extremely conservative and that the maximum deflections caused by most music playing at these volume levels will be much lower for a bass driver(s) crossed over to a mid range driver at about 200 Hz.

Yes, we have had the discussion and I still respectfully disagree.  I would point out again the difference between cumulative SPL and SPL at a given frequency.  There is a difference between the level read on an A or C weighted SPL meter and the output potential at any given frequency.  This doesn't necessarily mean a sine wave.  The low B on a 5 or 6 string bass is 30.87Hz.  This note being played, or a low note on a keyboard for all practical purposes though does approximate a sine wave.  If you want to be able to play this note at a given level, without compression due to driver suspension and BL losses, it's required excursion can be approximated with the formulas.  That doesn't mean that your overall cumulative output levels are limited to that SPL.  You can have the lowest frequencies compressed by 6-10dB and still see higher cumulative SPL, but your response curve now greatly changes as the lowest frequencies can't increase in level anymore. 

The other important issue again is to notice again is that the driver excursion pushes it farther away from center, the loss of BL and the suspension's increased stiffness will greatly raise the driver Qts.  If you look at what happens to the Alpha15 due to it's BL loss and increase in suspension stiffness at Xmax, you can see why it doesn't appear to move much.  First consider again that at Xmax, Bl drops for 7.7 to 5.39.  Suspension stiffness goes from Cms of about .25 to .1.  This means that the apparent Fs goes up to about 65Hz and Q goes to over 3.0.  Quite simply the driver doesn't appear to move a lot because it's suspension and motor won't allow it to.  Take the Alpha15 in your baffle and run a sweep at 1W,  4W, 16W, 64W, and 128W.  As you get higher in level you will see the low end response greatly begin to compress and taper off.   This is of course with sweeps, which are not an approximation of music as you mention.  Another way to simulate this effect which you might find more practical is to use pink noise and RTA.  This would more closely approximate music vs individual sine waves.  Measure with the RTA at 1W, 4W, etc up to 128W.  Again as you go up in level you'll see the energy at the lowest frequencies stop increasing while the higher frequencies continue to increase in level.  This is essentially what is happening to your music program material as well. 

John

[Graham Maynard]

  Another nice thing is that in the active system you can passively flatten the impedance curve first.  Then work with your xover and EQ.  You can end up with a nearly resistive load to each channel of amplification. 

Hi John,

My grey cells also lit up when I read that; but I let it ride.

Your links and explanation of this in post #48 above are good, and illustrate an aspect often forgotten by those who use passive crossovers. 

What caught my attention however, was your mention of this in the same paragraph as 'active' filters with implied direct LS to amplifier connection, where, if the amplifier is good and a normal SS type, those impedance correction circuits could not have any influence upon the drive to the LS itself within its own circuit, no matter how accurately the amplifier is being loaded.

I much appreciate reading your explanations though.

Cheers ...... Graham.

[johnk...]

Just a quick comment of excursion requirements. My experience indicates that (for music) what I call the volume-separation product, VSP, should be a minimum of 33,000 cm^4 for a dipole type woofer system.

VSP = N x Sd x Xmax X D

where N is the number of drivers, Sd and Xmax are in cm, and D is the effective separation (Length of an H frame, effective radius of a baffle, etc. Xmax is the linear, one way (not peak to peak) max excursion.

Based on this, a 15" driver with Sd about 900 cm^2 in 41 cm (16") H frame would need to have about Xmax = 9 mm (one way, not peak to peak).


John, I don't know where you got the formula for SPL but I can not be correct because it does not include the dipole separation and must vary as F^3 for a dipole. The equation I come up with is

SPL = 40 + 20 Log (VSP x (F/130)^3) 

ALso, are you saying the IB 15 has an Xmax of 18 mm peak to peak? I had interpreted that as one way (i.e.36 mm p-p).

[MJK]

Just a quick comment of excursion requirements. My experience indicates that (for music) what I call the volume-separation product, VSP, should be a minimum of 33,000 cm^4 for a dipole type woofer system.

VSP = N x Sd x Xmax X D

where N is the number of drivers, Sd and Xmax are in cm, and D is the effective separation (Length of an H frame, effective radius of a baffle, etc. Xmax is the linear, one way (not peak to peak) max excursion.

Based on this, a 15" driver with Sd about 900 cm^2 in 41 cm (16") H frame would need to have about Xmax = 9 mm (one way, not peak to peak).

johnk...

I am not sure what to make of this formula and your calculation for the 15" driver. What do you achieve by meeting this requirement. what assumptions went into the derivation of the relationship? I am having a hard time with the 9 mm requirement for Xmax, I have never seen my drivers move anywhere near that amount even on bass heavy material at significant volume levels.

Martin

[Graham Maynard]

Maybe john's figure relates to 20Hz where much more extreme excursions become necessary, and often set the SPL limit.

Which SPL is that John ?

Cheers ...... Graham.

[johnk...]

The number (33,000) is based on what I consider acceptable SPL capability at 30 Hz. If you plug 33000 for VSP and 30 Hz into the SPL formula you get an SPL of 92dB. Add 6dB if the woofer is near the floor: 98 db, and, assuming stereo woofer, add another 6dB >>> 104 dB at 30 Hz.

My single woofer NaO has a VSP = 38000 and when listening to pieces like The Rite of Spring, The Firebird, Nona Hendrix, Skin Diver, and others, the woofers are definitely "moving". SL's Orion has an estimated VSP of about 33800.


As for the formula, I derived it a long time ago and I can't find the details but it is just a dipole correction to the basic monopole SPL, similar to what John gave. If you like, Monopole SPL goes like

 20 Log(Vd x F^2) .

Dipole = monopole at Feq = c/6D.

Therefore, dipole goes like

20 Log (Vd x F^2 x (F/Feq)) = 20 Log (Vd x F^3 x D) + const.... Vd = Sd x Xmax, Vd x D = Sd x Xmax x D = VSP.

[John_E_Janowitz]

John, I don't know where you got the formula for SPL but I can not be correct because it does not include the dipole separation and must vary as F^3 for a dipole. The equation I come up with is

The formula is strictly the amount of air you need to move for a given frequency, back calculated from a Linkwitz formula.  This says nothing about the type of baffle, enclosure, etc that it is on.  It is for a half space measurement only.  You can take that measurement though, simulate your baffle rolloff and use the two together.  If the formula calculates 120dB at 30hz, and your baffle rolloff is -10db at 30hz, you are limited to 110dB.

ALso, are you saying the IB 15 has an Xmax of 18 mm peak to peak? I had interpreted that as one way (i.e.36 mm p-p).

The IB15 has Xmax of 18.5mm one way.

John

[johnk...]

Ok John, I agree. The formula you presented is for a simple source radiating into 2 Pi space.That carries with it that the source is a monopole.

[AK]

what about mechanical noise? I have dayton HF subs and they have quite a bit of noise when cone starts to visibly move.
IMO for dipole you can't take advantage oh high excursion because of suspension noise. I think surface area is more important, 21in woofers is the way to go