20 Hz to 20 kHz Dipole Speaker System

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I have been busy the past couple of months designing, building, and testing a new two way dipole speaker system. The resulting speaker has an efficiency of 88 dB/W/m and covers the full spectrum from 20 Hz to 20 kHz. You can find pictures and the documentation of the design on the Projects page of my web site.
 
Hi Martin,

Apart the obvious compliments, particullary for excellent calculated Vs measured data, I have two question:
1) How is imaging with such large baffle?
2) How bad are the vibes :cannotbe: ?
Excuse for my rude english, I found your work always accurate and scientific.

Cheers,

Paolo
 
inertial said:
1) How is imaging with such large baffle?
2) How bad are the vibes?

Hi Paolo,

Imaging is very good, as good as any of the other systems that have been in my room.

I assume you are asking about cabinet vibes. To be honest I have not spent much time investigating cabinet vibes. In the future I am going to add mass using a large piece of stone iwhich should be a step in the right direction. But I am not sure it is required.
 
Hi Martin
I like your work esp. correlating theory with measurements.
I quote you from your project paper



"The use of an active crossover really makes the system work. Levels and
crossover points can be adjusted to optimize performance. Crossing low maintains the
full range driver sound while removing the demand to produce bass from the Jordan.
The system comes together really well, deep dynamic bass and clear crisp response
from approximately 20 Hz to 20 kHz. But be warned, at 88 dB/W/m it does take some
power to make them really sing their best. At the end of an evening of enjoyable listening
at decent volume levels, my amps are warm."


1) How do you choose the crossover frequency and order of the active filters? I realize this is partly based on the excursion tradeoffs of the fullrange driver and the want to cross low. But there is alot more that comes to play here... yes?

2) Can you discuss the highpass boost to the H frame woofer in more detail?

3) Are you going to publish circuitry details for 1) & 2) ?
 
infinia said:
1) How do you choose the crossover frequency and order of the active filters? I realize this is partly based on the excursion tradeoffs of the fullrange driver and the want to cross low. But there is alot more that comes to play here... yes?

2) Can you discuss the highpass boost to the H frame woofer in more detail?

3) Are you going to publish circuitry details for 1) & 2) ?

1) My first cut at crossover frequencies and slopes is done using MathCad simulations. I have found that 2nd order filters connected in phase work well for the style of dipole designs I am focusing on. I revisit these calculations when I have measured SPL response data for the individual drivers in the system. Then adjustments are made by ear, I am using a dBx Driverack PA digital crossover so you can change setting on the fly and hear the results immediately. Simulations get me most of the way and the final small tweaka are done by ear.

2) I just applied a 2.5 dB boost, I tried other values to see what sounded most balanced and it came down to between 1.5, 2.0, and 2.5 dB settings. I settled on the 2.5 dB value for now.

3) The only thing I build are the speakers. The amps and crossovers are all off the shelf purchases. I have no interest in electronics design and construction.
 
Hi,

Another excellent and informative article.

to address some points in this thread and the article :

The c/o point selection principles are in the other articles.

For all things active see http://www.linkwitzlab.com/

The sensitivity and gain adjustment indicates a bucketload of drivers
would work in this design, if they are flat or passively EQ'd flat.

The Tangband-W4-1337SA (with some form of treble disperser ?)
springs to mind for 86dB/W, just reduce the bass gain a little.

Regarding the need for or lack of a treble driver :

Off axis response does matter and is perceived at the listening
position by the ratio of reflected to direct sound, how much ?
YMMV but off axis starts going off at around 3KHz.

Adding a tweeter would bring in a lot more driver possibilities.

The Tangband W5-704D + tweeter would certainly be cost effective.

:)/sreten.
 
Administrator
Joined 2004
Paid Member
LOL! That's a funny speaker Martin! Jordan with an 18. Looks like something my crazy friend John would do. He likes running an 18" with a tweeter - and it works!

FWIW, he bought 4 of those Goldwoods - to put 2 each under his P.Audio 18" coax drivers. Says he never could get the GW-1858 sounding right on OB, tho.

So how are they treating you in the that H-Frame? With the high QTS, they should sound nice and full. Pretty efficient, too.
 
That would seem a sensible compromise, if moderators & Martin are agreeable. On the whole, I would have thought it more useful over in the FR section anyway, given that a) it's using a Jordan to cover ~99% of the audible BW, and b) people who own Jordans are likely to frequent, and see the thread, in the FR section, rather than the main loudspeaker forum.

Either way though, nice project Martin. I bet a few people are going to be enjoying this one.
 
High Martin,

Nice effort. 20 to 20k does seem to fit the frequency response, however my calculations indicate very limited max SPL at the lower end. I'm assuming the effective diameter of the 18" woofer is about 16" and the spec's give Xmax as 3.5mm. I'm getting around 77 dB (max 2 Pi @ 20 Hz) in a 16" H frame. I also have a question about your quoted sensitivity, at least for the woofer section. The Parts Express page gives 94.5dB/2.83 V/M. With a 16" H frame the dipole = monopole point should be around 141 Hz. This seems in close agreement with your Figure 17 (center) where you have about 95dB at about 140 Hz. With a 100 Hz LR2 filter there would be about 10dB attenuation at 140 Hz so once the filter is applied the woofer output would be only 85 dB/2.83V input at the dipole = monopole frequency. This sensitivity seems consistent with the additional 2.5dB boost you apply to the LP to get to nominally 88dB.

Perhaps you can expand on the SPL limits and this apparent discrepancy in sensitivity.
 
Hi,

Martin, thank you for a nice OB article. With biamping and active crossovers there's much more room to play, and we can have a broader choice of speakers. Through experimenting, and with a big thanks to Nelson Pass and to You, I noticed that some crossover settings work very well, even though they are not typical textbook ones: Low pass 80 Hz 2nd order L-R, and High-Pass 240 Hz 2nd order L-R. I tried this one on my OB's (Eminence Beta 15+ Visaton B200) and it seems to work very well. Simple baffle, 50x120 cm.

It's a pity that you don't build active crossovers yourself. I don't favor the use of opamps (they are usually used in most active crossovers-a lot of them) as, in my opinion, they degrade the sound quality. No flames here, but I just opted to use Nelson Pass buffer B1 as a basis and build an active crossover using only J-fets (Nelson Pass will present this in B4).

Best Regards,

Vix

p.s I also agree that this thread belongs to the fullrange forum:cool:
 
I usually avoid OPs in the HP section and get away with simple passive RC.

By chosing the proper "R" associated with output impedance of the upstream and input impedance of the downstream, I found it works pretty good at preserving the sensitive and fragile mid-high details.

Putting those sound-contaminating OP buffer stages and gain contols on the LP section can save you a lot of troubles. OPs still show their shortness but much less harmful on the lower frequencies.
 
Martin, I'm two days away from ordering the 1858s (gotta make sure the money's in the account). They first caught my eye for similar reasons that you chose them- cheap, cheap, cheap, high Q, and huge. I already have a set of the Goldwood 8003s (the price makes for easy experimentation if we were comparing to the Jordan or any other quality FR, which we aren't) mounted on a 24" x 40" baffle, 1 1/2" thick. I thought the sheer mass of the contertop material would be of huge benefit, but it has quite a bit of vibration. Sound is good so far (needs an tiny tweet and potentially a notch filter), my question is am I wasting my time trying to make the 18s work with no H-frame and non ideal baffle material? I don't expect a miracle, just satisfactory sound to get used to before I plunk down $600 or more for the real deal done right. Worth my time to even cut the holes or should I go for the H or U frame?

Thanks to you and all others for the help

Jason
 
Hi John K,

Nice effort.

Thanks for the positive feedback, it is appreciated.

20 to 20k does seem to fit the frequency response, however my calculations indicate very limited max SPL at the lower end. I'm assuming the effective diameter of the 18" woofer is about 16" and the spec's give Xmax as 3.5mm. I'm getting around 77 dB (max 2 Pi @ 20 Hz) in a 16" H frame. I also have a question about your quoted sensitivity, at least for the woofer section. The Parts Express page gives 94.5dB/2.83 V/M. With a 16" H frame the dipole = monopole point should be around 141 Hz. This seems in close agreement with your Figure 17 (center) where you have about 95dB at about 140 Hz. With a 100 Hz LR2 filter there would be about 10dB attenuation at 140 Hz so once the filter is applied the woofer output would be only 85 dB/2.83V input at the dipole = monopole frequency. This sensitivity seems consistent with the additional 2.5dB boost you apply to the LP to get to nominally 88dB.

Perhaps you can expand on the SPL limits and this apparent discrepancy in sensitivity.

OK, I am not sure I completely understand your chain of efficiencies so let me step through it from simple to complex and then you can let me know if we are in agreement. These are all calculated values from my MathCad models.

1. From Sd I get an effective diameter of 15.4" for the Goldwood GW-1858.

2. Infinite baffle SPL measured on axis = 94.4 dB/2.83 V/m, it peaks at 95.9 dB at 40 Hz. I believe Table 2 in my write up quotes efficiency for 1 watt into the actual driver Re, I copied the values from the Praxis output table.

3. H Frame in free space measured on axis = 84.5 dB/2.93 V/m at 100 Hz, 81.8 dB at 30 Hz, and 76.7 dB at 20 Hz

4. H Frame sitting on the floor measured on axis = 88.6 dB/2.83 V/m at 100 Hz, 87.2 dB at 30 Hz, and 82.4 dB at 20 Hz

5. H Frame sitting on the floor and measures at 1 m on the Jordan's axis = 85.4 dB/2.83 V/1.14 m , 84.3 dB at 30 Hz, and 79.5 dB at 20 Hz

By moving the mic to be on the axis of the Jordan driver, I loose just over 2 dB due to the increased distance and the off axis position. Also note that the floor reinforcement adds almost 6 dB to the low end SPL values, this should not be neglected in the design of dipoles.

As far as Xmax being a limitation. I plot the driver displacement as a function of frequency for 2.83 V and it would indicate that a lot of the driver linear range is being used up. But this is only true if we are listening to sine waves. I believe that this is an overly conservative measure of a speakers dynamic capability once a whole collection of sine waves with different phases are applied as in a musical signal. I was listening to a bass heavy recording last night and checked to see how much the woofer was moving to produce a significant volume level, it was really not moving very much and required touch to determine it was actually moving, no big cone jumps back and forth. That is my subjective insight on Xmax.

Does that help?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.