"Unitized" Image Control Waveguide

What do you think of the Sausalito data of the Homepod, any insights that can be applied to a less compromised form factor?

The performance is spectacular. Particularly the fact that it's generating output up to 16khz. That's really hard to do with full-range drivers, and doing it with an ARRAY of full-range drivers is practically impossible.

NSW2-freq-imp.jpg

For instance, if you look at the spec sheet of the Aurasound 2", which is a very nice full-range, it has very little output above 15khz. And once you array a driver like that, the situation gets worse, because of destructive interference.

In some ways, the Homepod reminds me of an idea I had five years ago, which was to array a series of full range drivers in a ring, to make a "poor mans" JBL D2430K. (JBL M2 for The Poors)

c0143750_1221353.jpg

Cutaway of JBL's dual diaphragm ring radiators. (D2430K and D2415)

2430k-1.jpg

My goofy idea of how you could emulate the JBL using a ring of full-range drivers

large-17.jpg

Cutaway of the Apple HomePod
 
Here's a status report on the project. This post is intended to cover the following:

1) How does the loudspeaker perform?

2) What are the specs?

3) What is a recommended crossover?

axMd9CW.png

Here's what the loudspeaker looks like, at the moment. It measures 8.5" wide, 6.5" tall, and 3.5" deep. (21.6cm x 16.5cm x 8.9cm)

The waveguide "loads" the tweeter down to approximately 964Hz.

The waveguide weighs approximately 363 grams (12.8oz).

An externally hosted image should be here but it was not working when we last tested it.


The waveguide uses the Gento SP99023A, which is readily available on eBay. I printed half a dozen waveguides, and if you purchase one of my extras, I will include the midranges. Four in total.

Af2fkIS.png

Here's the measured polar response of the loudspeaker, using the recommended crossover and equalization. The hallmark of a Unity horn is that the midranges and the tweeter blend seamlessly; when executed correctly, a Unity waveguide measures like a single driver on a waveguide, and I believe this measurement illustrates how the Unity waveguides are special. In particular, note that the polar response of the waveguide doesn't exhibit any peaks or dips at the 1500Hz crossover point, on OR off axis.

RNEnrxy.png

Here's the recommended crossover. There's a total of three components:

1) There is a 2nd order highpass on the tweeter. The capacitor is 5.1uf and the inductor is 0.51mh. (You can round down to 5.0uf and 0.5mh.) There's a 2ohm resistor in series with the tweeter. The crossover components are not included with the waveguide, and can be purchased from Parts Express, Madisound, Meniscus, Solen or eBay.

2) I did something unconventional for the woofer crossover; I am implementing a low pass filter using polyester fiberfill. Basically I am stuffing the midrange taps with polyester fiberfill. I tried doing the lowpass crossover using an inductor, but the fiberfill option measured better. It had lower distortion and improved polar response. This is because the polyester fiberfill prevents the tweeter from diffracting off of the midrange taps.

Note that by design, this loudspeaker is designed to be listened to off-axis. The optimum curve is 22 degrees off axis; that's why the top end of the loudspeaker is 'tipped up' when listened to ON axis. The overall response of the loudspeaker is also using a 'house curve' that slowly rolls off the highs, similar to this:

519b00_3aa097c2593f4315af97a5a05e32690d~mv2.webp


If you prefer a response curve that's flat from the lows to the highs, that can be accomplished via MiniDSP or the DSP of your choice.

If anyone has questions about the loudspeaker performance or specs, let me know.
 
As mentioned in earlier threads, I have a bunch of spare waveguides left over.

A number of people have emailed me, and asked what I'd sell them for.

I have been stumped, because the quality of the waveguides is all over the map.

I gave it some thought, and here's what I came up with:

If anyone would like to purchase a set of waveguides, they are welcome to. I will have three tiers of pricing:

Tier One: this is the top tier, the best of the best 3D prints. A set of these prints will sell for $199. Included with the price is shipping to anywhere in the 48 states. If you're overseas, I can provide you a shipping quote. Along with the waveguides, I will include four of the midranges.

The waveguides in tier one aren't perfect but they are the best that I can print with my cheap printer. If you want a nearly flawless print, I have uploaded the model to Thingiverse and it can be printed at Shapeways. Last I looked, that was something like $1600 a pair, so this is still a dramatic discount over commercial printers. If you want to print them using a local Makerspace or the like, please do. I am not printing these to make a profit, I am printing these because I <3 Unity horns and I wanted to provide a kit that would be easy to build and affordable.

Here's an example of a "tier one" waveguide; note that there's a bit of warping and some stringing. It's not perfect, but it's as good as I can print:

sCezLrs.png


w3GEwLI.png


Tier Two: These are basically the "B Stock" of my waveguides. They are not as good as "Tier One." Tier Two waveguides suffer from some noticeable splitting and warping. Please note that my own speaker is using a Tier Two waveguide. All of the measurements that I've posted are using a waveguide that isn't the best. I've printed about a dozen waveguides, and it took me a few tries to get the process down. So the very speakers that I will be personally listening to are NOT "Tier One." So please don't consider "Tier Two" defective, they're not. They're just not printed as well as Tier One. Most of their issues are cosmetic. I'll sell you a set of "Tier Two" waveguides for $99. Midranges and shipping are included in the lower 48.

vqXZOr7.png


2uwezlH.png


a4zhDK4.png


The pics above demonstrate the difference between "Tier One" and "Tier Two." The grey waveguide is "Tier Two." It suffers from noticeable splitting and warping.

"Tier Three" is my last option. If you'd like a "Tier Three" waveguide, I'll give it to you for free. (The waveguide only.) The only caveat is that you have to meet me in person. If you want me to ship you a "Tier Three" waveguide, it's $40 for a set, which covers the cost of postage and four midranges. The "Tier Three" waveguides suffer from serious splitting, and in some cases, need to be glued together to function. This tier is basically for people who want to tinker with Unity waveguides. The cost is basically nil. As noted before, this isn't a business, this is a hobby, I just want people to hear how great Unity waveguides can sound. I'd rather give you a waveguide for free than throw it in the trash.

axMd9CW.png


Here is my own personal speaker, the one I've posted measurements from. It's not as good as Tier One, or as poor as Tier Three. It suffers from noticeable splitting and warping, but it didn't crack in half entirely. The reason that the splitting isn't audible or measurable is because I've sealed off the back of the waveguide. I sealed it with Mortite, which deadens the waveguide and makes it airtight. Mortite is readily available at most hardware stores for five dollars.
 
Last edited:
My waveguides had some splitting too (particularly when I tightened them down into an enclosure). For those considering the Level 2 or maybe 3 guides, pits and cracks in PLA waveguides can be fixed by rubbing in some JB Weld 2-part black Epoxy and letting it cure (use rubber gloves, hard to get off of hands!). Then hit it with a plastic primer (Rustoleum) and spray paint to the color you like. They can be made to look pretty decent -- but try to get the epoxy flat or at least not sticking out before you try to sand it down, as the JB Weld doesn't sand very easily and PLA doesn't take well to power sanding, it prefers to ...melt.
 
Status update:

For the past week, I have been pulling my hair out, trying to come up with a box design for this speaker.

rcRe0VB.png


My original design used a pair of MCM 55-1870 woofers in a D'Appolito configuration, along with a sonotube enclosure that's lightweight and rigid.

From an engineering standpoint, that seemed clever, but I didn't like how it looks. I know that I am *terrible* at making pretty loudspeakers, so I really wanted to make an effort to make the speaker look good, and that sonotube doesn't cut it.

Aladdin+Pics+003+cropped.JPG


I came up with a design that is similar to this speaker from Dave Pellegrene. My design uses the QSC 8NX, which appears to be the exact same driver as the B&C 8NDL51 but available for a discount.

The design was solid, but then I had a new issue, which is that the 8NDL51 is complete overkill. It can handle 400 watts while my tweeter and midrange array can handle about 50 watts.

So that seemed a bit silly; yes the 8NDL51 is very nice but it's probably unnecessarily expensive for this project. (If you guys disagree, let me know. This is a community effort.)

alpine-sws-10d2-10-subwoofer-2-2-500rms-hertz-morel-D_NQ_NP_770111-MLB20474763893_112015-F.webp


In a second attempt at making a complete speaker, I came up with a box that uses the Alpine SWS-10D4 for a woofer.

That came out nice: though it's power handling is very high, it's efficiency is low, so you wind up with a nice speaker. Basically the loudspeaker is power hungry, but the box is small. The Alpine provides a 2ohm load. So the Alpine absorbs a bunch of your amplifiers power, while the midrange and the tweeter do not.

The Alpine SWS-10D4 is available globally for about $100, so availability is no issue.

But one little problem...

The SWS-10D4 gave me an F3 of 25hz.

I wasn't planning on creating a full range speaker; my intention was to basically make a studio monitor that could compete with the likes of Behringer, JBL, Mackie and Genelec. 25Hz is a lower cutoff than I need, and the box size grew to three cubic feet.

Right now, I am thinking about going back to the MCM 55-1870 in D'Appolito, in a fairly conventional box. It would wind up looking a lot like an Aria 5:

Focal Aria 5 loudspeaker kit | Stereophile.com

tower2.jpg


1090FocA5.2.jpg


Right now, this is the plan, but if you guys have a better idea, I'm all ears! I'm fairly decent at making Unity horns, but I'm terrible at making pretty loudsepaker boxes, finishing a project, or coming up with a cohesive full-range loudspeaker.
 
Status update:

The design was solid, but then I had a new issue, which is that the 8NDL51 is complete overkill. It can handle 400 watts while my tweeter and midrange array can handle about 50 watts.

So that seemed a bit silly; yes the 8NDL51 is very nice but it's probably unnecessarily expensive for this project. (If you guys disagree, let me know. This is a community effort.)

Right now, this is the plan, but if you guys have a better idea, I'm all ears! I'm fairly decent at making Unity horns, but I'm terrible at making pretty loudsepaker boxes, finishing a project, or coming up with a cohesive full-range loudspeaker.


Even if the 8NDL51 lists it's AES power handling as 400 watts, long term compression effects ( VC heating ) will start to set in around 1/10th that power.



AES power handling is a calculated figure that uses filtered pink noise and a calculated power based on the lowest impedance.


From B&C's web pages, for cone drivers they use filtered pink noise in the region Fs > 10Fs, with the driver operating in free-air. [ 66hz - 660 hz in the case of this driver ]


Power is calculated on the lowest impedance.



400 w into 6.6 ohms is 51v. 51v into the driver at the roughly ~48 ohm impedance maxima is only 54 watts.


Also of note, once the driver is warm, the DC Re will climb from it's static room temperature value.



If you read the B&C's FAQ, they also indicate that 'Continuous power handling' is specified as double nominal power handling.



B&C Speakers


The 8NDL51 may be less overkill than you think.
 

This is looking like a winner.

Here's what I see when I model the three that I've looked at, so far:

1) the 8NDL51 works perfectly well, but it's relatively expensive and it is absolute loafing in this design. What I mean by that, is that the 8NDL51 is rated for 400 watts, but the midrange array in this loudspeaker design can only handle about 40 watts. So you wind up in a situation where the midrange array is limiting the performance overall, but the woofer is only using a fraction of it's capabilities. Again, it could work very well for YOUR design, but it seems to be "overkill" here. As diyuser2010 noted, there's definitely a valid argument for using it. The B&C is around $90-$120 depending on where you purchase it.

2) The MCM 55-1870 is compelling. It is very very cheap, performs nicely, is efficient. It's achilles heel is very high qts and low excursion. When you put these two things together, you wind up with a loudspeaker that requires EQ for flat response and limited output. Two of them cost $25.

3) The Exodus Anarchy 5.5" has 3X as much xmax as the MCM. It has a motor that reduces distortion. It's big brother has been put on the Klippel by Vance Dickason, and it's measured performance is very close to it's advertised performance, which I like. The Anarchy 5.5" cost $65.

DMVsrrW.png


Here's the response of the MCM 55-1870 vs the Anarchy 5.5".

Both are in a MLTL. The Anarchy is good for 109dB before it runs out of xmax. The MCM is good for 100dB. That big low frequency peak of the MCM has to be EQ'd away. It's inherent due to it's high QTS.

The MCM hits xmax with 25 watts into 8 ohms, the Anarchy does it at 75 watts into 8ohms.

gaspode-whole.jpg


The MCM makes a lot of sense in an array. It hits xmax in a hurry, but arrayed, the efficiency goes up and the limited xmax is addressed.

It would actually be interesting to make a "high efficiency" version of this Unity speaker using two or four of the MCMs, along with a "low efficiency" version of this speaker using a single Anarchy. (picture courtesy of Neil J Mackie: Paraglow II page)

I'm going for the latter because I use solid state amplfication, but the MCM might be an interesting option for the tube crowd.

An externally hosted image should be here but it was not working when we last tested it.


If things go according to plan, the final loudspeaker will wind up looking a lot like the Kef Reference 201. The cabinet on this won't be big at all. My speakers should be quite a bit more efficient; the KEF was 86dB, mine should be around 90dB-ish.

708KEFfig3.jpg


Here's the crossover for the KEF

708KEFfig4.jpg


Here's the overall response

Af2fkIS.jpg


708KEFfig5.jpg


Though I wasn't trying to copy the Kef design, I gotta admit they are really really similar. Similar polar response, similar crossover points. Kef data courtesy of Stereophile : KEF Reference 201/2 loudspeaker Measurements | Stereophile.com
 
I hate to flip-flop, but there's another option that looks compelling:

traveling-while-audiophile-fig14.jpg


If I arrange four of the MCM midranges in an array like the DNA Sequence speaker, I should be able to extend the beamwidth control down to about 750Hz. In the measurement of my waveguide you can see that the beamwidth begins to broaden in the midrange, due to the small size of the waveguide. (That was intentional, small waveguides "seem" to sound better than big waveguides, but the jury is out on that.)

SjKSj1g.jpg


Four of the MCMs raises the output to about 113dB on the low end, about 3dB more than my target. So that's good. They also due it with less power, due to higher efficiency. They're also an easier load; if any of you guys are running tube amps, the MCM array will be preferable to the Anarchy. The MCM array has a higher impedance and requires less power, but also needs a bigger box.

Nd6R2Rz.png

Here's the response shape with a filter in place. The filter smooths out the high frequency peak of the MCM, doubles it's power handling, and crosses us over to the midranges.
 
The 5.5-inch Anarchy is actually 55$ I think (not sure if that includes shipping). The 6.5-incher is the one that's $65. But you're right, which option you go with depends on box size and beamwidth control. Of course the array design is much more exotic and reminds me a bit of Follgott's pseudo-coaxial.
 
Did you know about this DIY Loudspeakers Kimmo Saunisto.

Kimmo has designed and produced several different speaker constructions so there is much info to be found. He has also produced VituixCAD a "free engineering and simulation software for passive and active multi-way / multi-driver loudspeakers".

The program is full of features so just take a look. The documentation is also extensive with 53 pages help pdf file plus youtube videos etc.

One cool feature that can be mentioned is the "SPL Trace tool that can capture frequency and impedance responses from bitmap loaded from file or pasted from the clipboard.

Software
 
Hi Patrick,
Are there any other midrange options that would be more capable than the 2” drivers you are using?

Tymphany sells a series of drivers named "TC6" that will drop right in. Their frame is basically identical to the Gento. I can't tell if Gento copied Tymphany or vice versa.

The Tymphany starts around $8.

There are some TC6 models with lower inductance, but I'm not sure if that's worth the additional cost since we're crossing over at 1500hz.

At some point I should probably see if the Tymphany performs better, but I haven't had time to. The Gento performs ridiculously well for the cost. My reference speakers, which Bill Waslo built, use the Gento. I've been listening to them daily for months now. I can't complain about a thing.
 
Did you know about this DIY Loudspeakers Kimmo Saunisto.

Kimmo has designed and produced several different speaker constructions so there is much info to be found. He has also produced VituixCAD a "free engineering and simulation software for passive and active multi-way / multi-driver loudspeakers".

The program is full of features so just take a look. The documentation is also extensive with 53 pages help pdf file plus youtube videos etc.

One cool feature that can be mentioned is the "SPL Trace tool that can capture frequency and impedance responses from bitmap loaded from file or pasted from the clipboard.

Software

I've been trying to learn it :( :(

I basically blew my entire Christmas break learning ABEC in 2018, so I have to be careful about investing *too* much time in learning software, since that is basically my day job already. (I'm a software consultant.)
 
Still confused on the diff between Unity and Synergy horn.

Did find this explanation but it can hardly be so simple.

"The Synergy horn is like a grown-up version of the Unity"

First off, this is not legal advice, and if you are interested in finding out the difference between a Unity horn and a Synergy horn, you might consider consulting a lawyer.

Here is my impression, based on a hobbyist's observation of the two devices. These observations are based on my examination of a Lambda Unity Horn and a Danley Sound Labs SH-50:

The Unity Horn is a multi-entrant horn designed by Tom Danley for Sound Physics Labs. The patent largely covers the geometry of the horn. The patent is expired.

The Synergy Horn is a multi-entrant horn designed by Tom Danley for Danley Sound Labs. The patent covers the geometry of the horn, which has some important differences from the Unity Horn. The design is covered by a valid patent.

I can't recall if the crossover is mentioned in either patent. They are different.
 
Last edited: