Dipole Design Questions?

[mayhem13]

Can an Dipole design use 2 identical drivers in a Push-Push configuration in a sealed box or is an open back the only way to go? Are there tradeoffs to this and if so, what might they be?

[dumaresq]

if you have two identical drivers in push-push isn't that a Bipole no a Dipole?

[mayhem13]

heh heh-Don't know either-hence the question. I'm feeling pretty ignorant right now.

[ThomasW]

Can an Dipole design use 2 identical drivers in a Push-Push configuration in a sealed box

Yes

or is an open back the only way to go? Are there tradeoffs to this and if so, what might they be?

You don't get the benefits of the dipole cancellation (this is important IMO), but you do get more bass output with less EQ since the drivers are thermal not excursion limited.

[mayhem13]

Are there benefits for a mid-bass driver in this configuration, say from 200-2.5khz?

[Dennis H]

I'm not sure what you mean by push push. If you have 2 drivers mounted on the front and back of a box, it will be a bipole if the drivers are wired in phase and a dipole if they are wired out of phase.

You might want to build a dipole that way with two drivers if you wanted the driver to play high enough that the frequency response from the front was way different than the back. No way to predict how high that might be without measuring.

http://www.linkwitzlab.com/models.htm

[mayhem13]

One driver front, one back, out of phase then? when wired out of phase, what is the effect on spl-same as single driver or slightly higher given a 88db single driver efficiency ?

[Davey]

A second dipole driver (baffled into the same box) doesn't get you any increase in SPL. You could think of them as one driver.

A dipole will have some frequency (Fequal) at which point its SPL is the same as a closed box configuration. Below that frequency SPL will fall off at a 6db/octave rate. Take a look at section "A" of the link posted by Dennis.

A symmetrical dipole configuration for the midrange.....like example "d" is intriguing as a possible DIY effort.

Dave.

[JasonB]

A symmetrical dipole configuration for the midrange.....like example "d" is intriguing as a possible DIY effort.

Dave.

Dave,

I can see there might be one advantage and that is that the rear driver is not obstructed by its own basket and magnet. Is there something else I am missing?

[Dennis H]

One driver front, one back, out of phase then? when wired out of phase, what is the effect on spl-same as single driver or slightly higher given a 88db single driver efficiency ?

As Dave said, the SPL will be about the same as a single open-baffle driver if D is the same. Compared to a single driver in a sealed box, SPL will be much less at low frequencies because the - back wave cancels the + front wave. It's just the nature of dipoles, whether you build them with one driver or two front-back. That's why most dipoles cross to a monopole sub at some frequency; it just takes too many dipole drivers to get loud SPL all the way down to 20 Hz. You need to move 8x the air every time you go an octave lower with a dipole.

[Davey]

Well, depending upon what your objective is the obstruction of the magnet/basket might be a good thing. If the objective is to be more directional as you go up in frequency then a smooth(er) transition might be achieved by making use of that "obstruction."

The totally symmetrical setup "d" is....symmetrical. Unfortunately, it creates a cavity (or box) and the distance between two back-to-back drivers can be considerable. This yields something that might be pretty big. I've talked to Siegfried about this on a couple of occasions and (I think) he believes the trade-offs are not worth it. Might be something interesting to try though.

Cheers,

Dave.

[JasonB]

Well, depending upon what your objective is the obstruction of the magnet/basket might be a good thing. If the objective is to be more directional as you go up in frequency then a smooth(er) transition might be achieved by making use of that "obstruction."

Dave,

I have wondered how much of an effect the basket and magnet on a dipole midrange driver has. If you ever get the change to shoot some front vs. rear on and off axis response plots for the mids in your dipole setup I'd love to see the results.

Thanks,

Jason

[Davey]

Jason,

I didn't go all the way around from 0 to 180, but I did go from 0 to 90 degrees. Installed on the baffle of course.

I'm moving at a turtles pace on this project anyways, but I will post something in the future.

As you might expect the basket/magnet form a crude low-pass filter. Getting the mid-baffle dimension/construction correct is the key to this type of speaker system. Too wide and the polar response widens because of the off-axis angle addition and the reverberant field is too much. You can't equalize that so you have to fiddle with a different baffle size.

Cheers,

Dave.

[Saurav]

Too wide and the polar response widens because of the off-axis angle addition and the reverberant field is too much.

Could you explain that a bit more? Is this a different effect from the baffle width affecting the dipole rolloff frequency? Is it because of the low-pass effect of the basket/magnet? And how do you determine what's too wide? Is this something that can be simulated/calculated, or do you just measure the polar response? Or use some rules of thumb?

[dlneubec]

Could you explain that a bit more? Is this a different effect from the baffle width affecting the dipole rolloff frequency? Is it because of the low-pass effect of the basket/magnet? And how do you determine what's too wide? Is this something that can be simulated/calculated, or do you just measure the polar response? Or use some rules of thumb?

Saurav, you might want to look at ABCDipole from John Kreskovsky. You can model the response of the driver on and off axis (by what ever amount you want to enter) with different baffle widths, flat U or H baffle, etc. It is very easy and quick to do. The Egde can also do on axis open baffle dipole modeling to help determine optimum width

[hobby1]

hi
i'm working on similar design,here some observations
first,Qts of the speaker is important,with a hi Q speaker it have some gain at low frequency,in other word, no rolloff,possible more bass without equalization
with a low Q speaker the effect is rather negatif,the cause seems to be the 'infinite load" see by the speaker (the nul presure on the back of the 2 speaker)
second,there is a positif effect on nonlinear distortion ,when speakers are mounted back to back with magnet in touch
third,this design have a negatif effect in the mid frequency ,interference and resonnance,that can be damped but not easily
fourth,it is clearly have a better polar response then a "classic dipole" because of the perfect symmetry between front to back radiation .
excuse my poor english if necessary i'll try to give better explanation ,but this dipole design is interesting ,for exemple a bass dipole with this design and hiQ woofer can give a good result and be small!

[Saurav]

Saurav, you might want to look at ABCDipole from John Kreskovsky. You can model the response of the driver on and off axis (by what ever amount you want to enter) with different baffle widths, flat U or H baffle, etc. It is very easy and quick to do. The Egde can also do on axis open baffle dipole modeling to help determine optimum width

I have both programs, and have spent some time with them. Narrowing the baffle seemed to reduce the Q of the dipole peak, and shift it higher in frequency. I'm trying to correlate that with Davey's comments on widening the polar response. Maybe we're talking about the same thing, but I wanted to be sure I understood what 'widening the polar response' meant.

[hobby1]

here is a application http://www.perfect8.com/t_radiation.htm

[mayhem13]

Just a little food for thought, the design doesn't include an actual baffle but both drivers mounted opposing in a sphere, mechanically coupled. I'm not looking for anything below 150hz from this, just smoothest midrange response to 2.5khz.

[Davey]

I have both programs too, but I'm more of empirical data guy.

"Widening" maybe wasn't a good word. How 'bout "narrowing less quickly than desired"

The polar response of the mid/baffle helps to shape the overall balance of the system. The rear wave comes around to meet the front at off-axis angles so the reverberant field can have too much energy relative to the direct response....which might be flat. The objective is to create a driver/baffle combination that can be equalized to be level for on-axis measurements, but not have that equalization damage the polar response objective of the system.....which should still be uniform.
Too narrow a baffle and the dipole cancellation kills you, and with too wide a baffle you start running into these unequalizable polar response issues......and the thing becomes visually ugly (IMHO.)

Even with a symmetrical configuration you could still run into this problem. And in that case building more experimental baffles to address it is much more labor.

I think a person can get pretty darn good results with just a simple flat baffle for the midrange and appropriate equalization. John K's NaO is a perfect example. But, I've seen a lot of hair-brained ideas too.

Hope that helps.

Cheers,

Dave.

[Saurav]

Thanks.

How 'bout "narrowing less quickly than desired"

You're talking with respect to the polar angle, right, not as you go up or down in frequency. So at any given frequency, you'd get a 'fatter' figure of 8 with a wider baffle?

If I got that right, then I think I understand what you're saying. For some reason I was thinking in terms of the polar response widening as you went up in frequency, or something like that, and couldn't understand why.

ABCDipole recommends a baffle width that's roughly 2x the driver diameter. I'll probably start with something like that, though I plan to have tapered baffles so I'll shoot for ~2x the diameter at the height of the driver and see how that looks.

[dlneubec]

Saurav,

If it is any consolation, I'm doing a tapered open baffle midrange and the driver has an effective diamater of about 5.1" (132cm2 effective area). It is a flat baffle. I passed my design past John K. and he confirmed it was fine at 9.75" wide at the driver height. He did not see any reason that the tapered baffle would be a problem.

[Saurav]

That's good to know.

Are you doing anything specific in your modeling to handle the fact that you have a box under the baffle? My last speakers were a "box with OB", and I was never sure if I should include the box's front surface area in the OB model or not. It didn't make sense to model with just the upper baffle, since it wasn't open at the bottom so the sound wouldn't wrap around that edge. In fact, for something sitting on the floor, the floor edge of the baffle is effectively an infinite distance away, right, since the sound will never wrap around that edge.

Ultimately I ended up just using the full front baffle height in EDGE, and it seemed to work OK compared to my final results. The bottom edge is far enough away that I'm not sure it makes much difference to EDGE's calculations. And this is probably getting into a level of detail that the sims can't handle.

Also, if I remember from your drawings, you're going for symmetrically placed drivers, right?

[dlneubec]

I have modeled the open baffle section by itself (in the Edge and ABCDipole) and the entire front baffle to the floor in the Edge. The results in the Edge using the entire front baffle is not much different than just the mid baffle. The dipole peak is taller and less broad on the mid baffle and smoother. Below are the two results with both open baffle and without open baffle (showing diffraction) checked.

My woofer and mid are symetrical, but the tweeter is offset by 3/4". Only the mid is open baffle/dipole in my case, however. The pdf attached shows the most current plans that go with the Edge simulations.

[Saurav]

The results in the Edge using the entire front baffle is not much different than just the mid baffle.

That matches what I remember from when I tried this (been a few years).

The dipole peak is taller and less broad on the mid baffle and smoother.

My guess is that this is because it's calculating cancellation around the bottom edge of the baffle. But unless the OB baffle is raised above the woofer cabinet (like in the NaO or Davey's design), you won't really get that cancellation with your speakers, so I'd expect the 'full baffle' sims will end up being closer to reality. You'll only know once you build and measure