DIY dipole to replace Apogee Diva?

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

I've had planars/ribbons for years (Magnepan 1.6s, rebuilt Apogee Duettas, rebuilt Apogee Divas), and with the Apogees have been multiamping with crossovers generated by Acourate software.

I like the big line-source ribbon sound, but the Divas have obvious weaknesses in sensitivity, dynamic ability, bass, etc. It is possible to obtain a new, 'modern' Apogee but the cost is *very* high... and I have never trusted the design of the planar bass panel.

The obvious thing to do would be to build a linesource dipole speaker using (just for example) Revelator or Accuton ~6" mid drivers and (again for example) Raal tweeters. The slightly less obvious thing to do would be to build a Horbach Keele dipole.

I am not constrained by crossovers (Acourate can generate anything I could really want) or by amplification. By DIY standards cost is also not a major constraint. I do have a small room now (12 x 17 x 9, roughly) but I've heard IRS Vs sound great in a room only slightly larger, so I am not giving up hope just yet.

Has anyone here been in a similar position and found a good solution? I'd also be interested to know about professional audio engineers who might take on a side project.

all best,

Brandt
 
I don't know of a true ribbon dipole driver that is efficient and is also a *long* line source - that is commercially obtainable.

The resident expert on DIY long ribbon drivers would be Paul W. (..but he has personally moved on to large horns).


These people appear to do custom orders on longer line ribbons, but list neither the cost nor the measurements:

Transmission Audio Inc.


Because you have an active (digital) solution for crossover, the things you need to be more concerned with are dispersion patterns horizontally (and vertically if you plan on stacking an array), linear decay, non-linear distortion, and basic operational limits.

IMO ribbons are best suited to higher freq.s and in particular long very narrow line sources for very wide horizontal dispersion. IF you build yourself such a driver, or have one made, then you need to consider other drivers better suited to different bandwidths.

Now consider the midrange panel. Electrostatic panels, line source planars, line array planars, and line array cone drivers.

Hint: Think line array planar..

BG corp. just released their "10" version which is suitable as a midrange driver in an array. See measurements here:

Zaph|Audio

In an array of 8, and depending on the size of the baffle and the order of crossover, you could use these to around 150 Hz.

(Note that you could also use an array of the Neo 8's for a treble line array as well, but they might not have quite the detail of a true ribbon, nor would they have a horizontal dispersion pattern as wide and stable a more narrow line source.)


Finally there are the lower freq.s to consider. There are planar base panels available:

INFRAPLANAR : haut parleur isodynamique plan pour grave

..but not exactly cheap.

IMO a line array of bass drivers should do the trick, and in particular these drivers and their amplifiers:

SW-12-16FR - SW-12-16FR

Again, an array of 8 per loudspeaker. 2 "stacks" of 4 wired in parallel with one of these amplifiers per stack:

H600PEQ - H600PEQ
 
Last edited:
Let me back up. Ignoring for a moment (?) the specifics of the drivers, is there a dipole line source configuration superior to that of the Infinity IRS V? I.e., something better than a curved, wide baffle for the mid-tweet array? I wouldn't be able to make the baffle quite as wide as the IRSV baffle (room too small for that to be practical) but let that go for the moment.

If I can determine the overall configuration of the speaker, then I can move on to the drivers.

thanks,

Brandt
 
Let me back up. Ignoring for a moment (?) the specifics of the drivers, is there a dipole line source configuration superior to that of the Infinity IRS V? I.e., something better than a curved, wide baffle for the mid-tweet array? I wouldn't be able to make the baffle quite as wide as the IRSV baffle (room too small for that to be practical) but let that go for the moment.

If I can determine the overall configuration of the speaker, then I can move on to the drivers.

thanks,

Brandt

Most dipole drivers will have some inherent advantages over drivers run front and back out of phase.. like the IRS V.

Frankly the IRS V also had over-damped drivers (closed-back planars), and the surface area was inadequate for the midrange panels. Additionally, a full line source for the treble will not have combing effects.

As far as a basic design.. of a large baffle for a dipole line source being better (or not), that really depends on the drivers.

The more surface area AND displacement you have - the lower/louder you can go. The larger baffle is there to partition the rear out-of-phase signal with the frontal in-phase signal. Less baffle requires more surface area and greater excursion.

You can play around with free-ware apps like the Edge and the Baffle Diffraction Simulator to see the effects of spl-loss from a reduction in baffle area and dipole cancellation.

Home of the Edge

FRD Consortium


One thing to note: most designs do not have a true dipole radiation pattern. The larger the baffle the *less* the pattern resembles a dipole. (..that cancellation that is removed by a large baffle provides those side nulls in the pattern that are part of a dipole.)

IF you are more interested in a *mostly* dipole pattern then look to JohnK's Nao Note. Also see cubono's design.

http://www.diyaudio.com/forums/multi-way/163072-nao-note-preview.html

http://www.diyaudio.com/forums/multi-way/161768-violet-dsp-evolution-open-baffle-project.html



You *could* go with an array of cone drivers (midrange portion) with linear excursions significantly exceeding +/- 1mm, and "boost" the lower freq. response for use with a much less wide baffle. Perfect 8 does this (as does the ribbon manufacturer I provided a link to). The problem with this is the mass of the drivers in relation to the treble unit, and the differences in excursion between the two drivers. (..the treble unit is hundreds of times less mass than the cone driver array and the excursion levels are also radically different.) These differences tend to result in dissimilar sound between the two lines. This becomes less important as you go lower in freq.. BUT for some it's still a problem.
 
Ok, thanks -- that's helpful (sorry to be asking such noob questions...).

I take your point regarding the baffle width and making the polar response 'less like a dipole.' I also recognize that a true dipole driver may be better than pairs of closed drivers run antiphase (as in the IRSV).

So how do I know if a midrange driver has enough displacement?

And have you seen any true dipole tweeters suitable for an array? Raal makes a dipole tweeter, but the shape of the faceplate makes it unsuitable for array use. Mundorf has an AMT suitable for line source use 'in development.' Beyma's AMT looks a bit rough from a fabrication point of view, and has a somewhat large faceplate, but can be run as a dipole... have you seen anything 'real world' that looked good?

cheers,

Brandt
 
Ok, thanks -- that's helpful (sorry to be asking such noob questions...).

I take your point regarding the baffle width and making the polar response 'less like a dipole.' I also recognize that a true dipole driver may be better than pairs of closed drivers run antiphase (as in the IRSV).

So how do I know if a midrange driver has enough displacement?

And have you seen any true dipole tweeters suitable for an array? Raal makes a dipole tweeter, but the shape of the faceplate makes it unsuitable for array use. Mundorf has an AMT suitable for line source use 'in development.' Beyma's AMT looks a bit rough from a fabrication point of view, and has a somewhat large faceplate, but can be run as a dipole... have you seen anything 'real world' that looked good?

cheers,

Brandt


You can play with this calculator:

Piston Excursion calculator

Note it is an "infinite baffle" and you'll have to adjust for surface area.

Again, the BG Neo 8 is suitable in an array format. Note however that NONE are really optimal. The problem is that they aren't narrow enough, nor are they full lines. (..ideally it should be half an inch wide and about 5+ feet long.)

Again, I don't know of any manufacturer of dipole ribbon (..or even a planar) that meets those requirements (or is even efficient).

Even if Mundorf produces a dipole driver that has amazing off-axis character, it almost certainly won't be over 3 foot in length. :(
 
Would a 0.5" ribbon be able to get down to ~200Hz and move enough air to be useful? That seems awfully narrow... I understand that it would have great horizontal dispersion but at a cost, no?

The BG Neo 10 looks like an interesting driver.

No - a ribbon like that should be good for 1 kHz.. maybe a little lower. Below that it should transition to Neo 10's, or other panel or driver array. That should push the array low enough in freq. that combing shouldn't be a problem, nor directivity in general. Sure there is a cost, but it isn't performance, and it *shouldn't* be one of value (particularly if you DIY the ribbon).

The Neo 8 can be used down to about 400 Hz in a line array (with an appropriate filter and baffle). You can also achieve better high freq. dispersion via the Neo 8 PDR version (..with only a marginal loss in efficiency.) (..while I know budget is not as much of an issue for you, this is pretty much the least complicated and most value-oriented way to achieve a good "dipole'esq" open baffle implementation.)
 
Last edited:
btw, Scott, do you have listening experience with 'truer' dipoles vs wide-baffle dipoles? Curious what the tradeoffs are. I suppose I've never heard a 'true' dipole with a very narrow baffle.

Yes. Less room noise from side-walls AND baffle (..resulting in increased clarity), more non-linear distortion from "gain" to offset the high-pass character of the cancellation depending on the design (resulting in increased coloration and loudness). Off-axis response is still pretty wide (..around +/- 40 degrees). (This was using Neo 8's, but only one per side.)

Realistically it *may* offer enough clarity to favor the design over a narrow ribbon with a large baffle. ;) (..I'd estimate that a long 8 driver line (Neo 8 PDR) would be good to about 1000-900Hz used like this while still providing good headroom and very low non-linear distortion, provided the high-pass filter is a higher order design.) You would need to separate the Neo8 PDR line from the Mid driver line - say a 4 inch wide panel for the tweeter line with 3 inches of space before transitioning to the next panel. Also, you should probably continue the theme with the Neo 10's operating down to 300 Hz, and then transitioning to the next line of midbass drivers. (i.e. tweeter line/mid line/midbass line.. then further off to the side: subwoofer line).
 
I am trying to chase down Mundorf re AMTs designed for line arrays. Their site shows this:
http://www.mundorf.com/downloads/MUNDORF_AMT_Pressetext(de-en).pdf

And they say they have an AMT for arrays 'in development.' I realize there would be some comb filtering with a driver like this, but unless I can find someone to make a ca. 6' ribbon tweeter for me, I think this may be the best bet. (?) Is there a straightforward way to calculate how many tweeter drivers would match the output of a given number of midrange drivers?

Re woofers, what are the tradeoffs of a servo design vs non? If I weren't going to use the drivers and dedicated plate amps you mentioned, I'd probably use an AE Speakers dipole driver (12 or 15").
 
Hi seaspeak,

I think you would get the best sound if you DIY built a midrange and a tweeter ribbon and used 15" dipole speakers like the AES IB15 for the bass. All the knowledge for DIY constructing a "potentially superior" modern ribbon dipole design has been published in this forum. Stacking several production ribbon tweeters together will create comb filtering.

With modern NdFeB magnets with N45 - N50 strength midrange and tweeter ribbons can be constructed with ~95db/watt efficiency. $3K per pair for magnets and steel is required. A 0.5" wide tweeter ribbon, plus a 2" to 3" wide midrange ribbon based the desired crossover frequencies.

For the bass, companies like InfraSub have proved that powerful deep bass is possible from a large panel. Another alternative is to build a dipole bass line array out of 3 or 4 large woofers like the Ascoustic Elegance IB15. Four IB15s can be arranged for 95 - 98 db/watt efficiency.

Simulation will show that the Apogee trapazoid shape is one of the best monolithic baffle solutions. Some newer designs at companies like Transmission Audio put the midrange and tweeter ribbon in one narrow block with just half-rounds on each end for diffraction control. The tweeter ribbon is on the inside of the soundstage, similar to the Diva, to minimize dipole baffle nulls. Separating just the ribbons puts the baffle width closer to the dipole frequency of the midrange. Putting the bass line array in a separate frame isolations vibration. Wide mid+tweet baffles like the Infinity or Genesis produce many nulls above the dipole frequencies, which can be modeled.

I think you will need to build DIY midrange and tweeter ribbons to beat the Diva in sound quality with high efficiency.

=================

Acoustic Elegance Speakers: IB15 are 8ohm drivers. Running a pair in parallel at 4ohm off a modest pro audio amplifier will easily push these drivers to high levels. They have a full copper sleeve on the pole to lower inductance and distortion. Due to the low inductance they also have very good high frequency extension which could be beneficial in dipole applications.


Fs: 16Hz
Qms: 6.8
Vas: 439L
Cms: .45mm/N
Mms: 220g
Rms: 3.239
Xmax: 18.5mm
Xmech: 25mm
Sd: 830sqcm
Vd: 3.07L (p-p)
Qes: .78
Re: 5.5ohm
Le: .33mH
Bl:12.49
Pe: 500W
Qts: .7
1WSPL: 86dB
2.83V: 87.3dB

Physical dimensions:
Cutout hole: 14"
Outer diameter: 15.5"
Mounting Depth: 7.5"
Overall Depth: 8"
Weight: 19lbs

$150 each or 4 for $500
 

Attachments

  • 300_SB_300_Sample_444_1_.jpg
    300_SB_300_Sample_444_1_.jpg
    18.1 KB · Views: 342
I am trying to chase down Mundorf re AMTs designed for line arrays. Their site shows this:
http://www.mundorf.com/downloads/MUNDORF_AMT_Pressetext(de-en).pdf

And they say they have an AMT for arrays 'in development.' I realize there would be some comb filtering with a driver like this, but unless I can find someone to make a ca. 6' ribbon tweeter for me, I think this may be the best bet. (?) Is there a straightforward way to calculate how many tweeter drivers would match the output of a given number of midrange drivers?

Re woofers, what are the tradeoffs of a servo design vs non? If I weren't going to use the drivers and dedicated plate amps you mentioned, I'd probably use an AE Speakers dipole driver (12 or 15").


Interesting. Must be very limited vertically if it's designed for an array - and that means little combing. (..though you would need to confirm this.) At higher freq.s the rear radiation will be "disturbed" due to the cavity resonance of the driver.

As far as gain..

double surface area +3db; parallel connect is another +3db; series connect -3db. (..the parallel/series connection gain/loss is dependent on the amplifier.) With 4 drivers you normally series-parallel connect to end-up at the same nominal impedance as one driver with a net +3db of gain.

If it has little vertical output then you don't need to calculate any "array-gain".

A midrange array is far more complicated to figure because it includes "array-gain". (..basically you need to do some measurements.) HOWEVER, I think the Edge might show array-gain. (..you can also add-in a floor "mirrored" line to more accurately model the gain with respect to the floor boundary.)


As far as servo vs. non-servo..

Rythmik Audio • Direct Servo - how it works

The GR drivers are less expensive and have greater xmax.

On the other hand with the AE dipole drivers you can use both coils and drive each separately.. and perhaps with an amplifier more to your liking. (i.e. two amplifiers per subwoofer line - each driving the entire line.) Since you are "active" anyway - I'd look to NewClass D. (..good power and supposedly excellent sound.)

Which is better for this application? I have no idea. Certainly if *output* is your priority then the GR solution is better (..because of increased excursion). Also *gain* would be better with the GR solution because of the 16 ohm drivers all parallel (netting + 9 db per stack, or net +12 db with both stacks for the line.) Other than that I'd think that both solutions are excellent, but the AE/NewClassD solution will be more expensive.

For something less expensive and a bit smaller:

For either line you could go with 2*3 drivers parallel and wire each parallel grouping in series - but that gives you considerably less gain (and ultimately a bit less power). (..if you do this for the GR line then "switch" to the 8 ohm drivers instead and use only one amp.) Gain and Power is less of a problem with the AE solution in that you can still power each coil (..again, using an amplifier for each coil).


EDIT: I think LineSource's ideas are good.

DIY'ing a ribbon that meets your needs is possible, or perhaps you can find someone to fabricate it for you?

The IB15 is also a good idea, and less costly. it will necessarily be bigger of course (or at least wider depending on the number of drivers you use in a standard vertical orientation).
 
Last edited:
hi LineSource,

Thanks for the suggestions. Your avatar would suggest you know what I'm looking for (and I've looked at some of your archived posts as well)...now that you've voted with Scott for a single long tweeter ribbon, I'm trying to think about how to accomplish it.

This doesn't seem like a trivial process, as I'd need to:
1 - acquire wisdom (which in my experience tends to take a lot of time and work)
2 - develop CAD skills along with dipole modeling skills
3 - source raw materials
4 - find various service providers (to machine magnets & steel, potentially cut and corrugate ribbon, etc.)

I realize that's what DIY is ultimately about, but I don't have time to do all of the above to the standard I'd like. Are there any ways to compress the process somewhat? Any books that would be a good substitute for many many hours spent surfing the forum? Any enthusiasts or engineers who might trade some wisdom and/or design skills for cash or wine or something? I don't need to have every last thing handed to me on a platter but I also won't be able to do this from scratch given my life constraints.

Ok. Please let me know what you think -- where should I start?

thank you,

Brandt

ps maybe I could build a Neo 10 and AMT line source first, and work up to a dual-ribbon design...
 
Last edited:
linesource - I'm contemplating separate woofer arrays, have you heard such a setup, and are there soundstaging issues with physically separating the woofers laterally from the mid/tweets?

seaspeak,

I have heard just the midrange and tweeter ribbons of the Full Range Apogee set up right next to a line array of four Lambda TD15_dipoles (99db/watt) crossed LR4 at 120Hz. I thought it sounded superior to the standard Apogee flat bass panel. Much more dynamic. Far lower distortion on large transients like drums. The Apogee Full Range is separted into two pieces... ribbons and bass panel. The ribbons are only 86 db/watt efficient. NdFeB magnets can build 95-98db/watt ribbons.

From what I learned, I believe it is important to:
1) "dipole optimized" woofers provide superior bass to every planar bass panel I have heard. Run some sims with four AES IB15 woofers in parallel.
2) separate the bass array from the mid-tweet ribbons to isolate vibration. You can butt them right next to eachother. You can pivot the bass line source slightly toward the walls to help control room modes. You can play with the baffle shape - trapezoid, rectangle, wave curvey edge, etc..
3) use (digital) tme delay crossover to compensate for the woofer vs. ribbon physical offset
4) put the tweeter ribbon on the inside of the soundstage with just a half-round along the edge for baffle diffraction to get some dipole frequency imaging pushed into the center soundstage.
5) From my experience, a single baffle like the Diva puts an accurate soundstage between the two speakers. Safe for ribbon+bass butted together. A two piece (ribbons + bass) design where the bass array is significantly physically separated from the ribbons like the Infinity IRS V or Genesis extends the soundstage beyond the area between the ribbons, but the size and shape of the soundstage sounds incorrect. Unless the midrange ribbon is capable of 60Hz, I think the bass and ribbons much be very close together.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.