DIY point source horn

Update:

DIY point source kit

This point source horn combines a unique mix of features that make it very appealing. It has the extremely high efficiency of horns and the associated dynamics and clarity, but without the lack of coherence due to the coaxial design. It also features constant directivity. Here is my first prototype:



A compression driver is loaded into a conical horn at the apex and four 5" closed back mids are loaded into the same horn via ports. The mids operate in a bandpass configuration, where the air trapped under the cone acts as a compression chamber and provide acoustic filtering. This means harmonic distortion is also filtered, resulting in very low harmonic distortion.

S1 - prototype 1

My first prototype was a simple 60 x 60 degree conical horn with four 5" closed back mids, a compression driver at the throat and an 18" pro woofer. It performed very well, confirming the simulation was correct. It gave me an appreciation of what a point source horn can do in a small room. I enjoyed the effortless dynamics, unlimited clean output and midrange detail.

S2 - prototype 2

Now more confident that I have a good grasp of how this works, I started a more ambitious version with 90 x 45 degrees coverage and the 6 drivers necessary to retain the bottom end extension. This version has a second flare angle for an improved polar response and was much more difficult to build.

Drivers

Driver selection >

Tweeter:      B&C DE250 1" compression driver (1 - 18k)
Midrange:   Pyle PDMR5 5" closed back (230 - 1k)
Woofer:       Eminence Magnum 18LF (35 - 230)

(Subs: Rythmik Servo tapped horns - future project)

Simulations

Hornresp has proven more effective than expected for the mid drivers.

Sims vs measurements >
Hornresp lessons >

Construction

Build photos - part 1 >
Build photos - prototype complete >

Miniwaves foam mod

 

My Miniwaves are a budget test of the waveguide approach to speakers, including a 6" Vifa midrange and B&C compression driver loaded into an oblate spheroid waveguide. Previously I tried a foam mod with dacron, with little success.

More about the Miniwaves

In a recent waveguide shootout, they had the least amount of horn harshness.

I've found that reticulated foam takes waveguides to another level, hence this mod shown here.

First I created a template. I then cut the foam using the soldering iron. The result is shown below.

You can see I've tried cutting the foam differently on the outside.

Where to get reticulated foam in Australia

Bigger bass traps

More recently I have purchased some rigid fibreglass for bass trap testing. I now have 3 very large traps, two of them 1.2m wide and a third is 1.5m wide, each are thick as well.

 

What you can see here is that the modes are now very well damped. This is  an exceptional measurement that shows modes that have been well controlled.

Waveguide foam mod

Do you have horn or waveguide based speakers?

I'm sure you've noticed how they can at times hit you with a harsh sound. It isn't related to over-driving them, but this problem occurs almost randomly, even at levels well below their maximum output. Dr Earl Geddes has put forward that this effect can be described as higher order modes, or HOMs. They are reflections within a horn or waveguide. The solution he has proposed involves filling the device with reticulated foam, often called "filter foam." It differs from the more common open cell foam in that it has larger pores so there is less attenuation, but the concept is to selectively attenuate the HOMs without too much loss of efficiency. Having now tried this modification to waveguides I can say quite emphatically that it works. Once modified in this way, a waveguide with an obnoxious sting is tamed to the extent that it rivals a dome tweeter in smoothness and lack of listening fatigue. However, the efficiency is about 20 db greater and this combination can achieve far better controlled dispersion characteristics.

Three steps are involved:

1. Determine the horn/waveguide shape.
2. Cut the foam to shape.
3. Measure the response and modify the crossover to compensate for the loss

How to cut the foam

Various methods have been proposed, however the most effective appears to be using an electric hot knife to sculpt the foam into shape. Axi-symmetric units are the simplest, as they can be placed on a turntable. In my first attempt, I derived the contour with a process of measuring, trial and error. I ended up with a template for the contour. I have used a soldering iron set to about 420 degrees. First, I cut the ellipse for the end that is seen and that must be neat. I had a template made to ensure it was neat, then traced with the iron around the curve. It was a slow process and since I don't plan to make a large number of these, I have used patience and a certain amount of skill rather than building a cutter. Most of the foam will not be seen.

 

Here the foam is inserted into an 18 Sound waveguide in what is a deluxe version Econowave. 

 

Front view

 

Rear view seen by the compression driver.

 

Confused about the foam? What is the difference between open, closed and articulated foam? There is a good description here:

Stephenson & Law foam

Where to get reticulated foam in Australia

Cutting reticulated foam

I'm currently working on finding a way to cut reticulated foam to improve waveguides.

Here you can see an old Birko 120w soldring iron. I flattened the tip, then drilled a small hole to fit the copper wire.

The wire is shaped to the profile of the waveguide.

Unfortunately only the first 12mm or so of the wire was hot enough to cut the foam. It isn't likely that it will get hot enough and stay hot enough for the entire wire to do the job. I will need to either use just the tip, or investigate other methods. I will be trying a regular hot knife.

Waveguide shootout 2 - conclusions

We conducted these two events to learn about different waveguide and compression driver combinations. Both events proved very enjoyable and revealing and in coming together there is a unique opportunity to condense the learning.

 

Above: the JBL clone loaded with DE250 sits on top of the TD10M - one of the best measuring and sounding options we tried.

Some of my conclusions include:

• The Selenium driver is not a good choice for a home system due to sound quality issues
• The JBL clone waveguide is an excellent budget conscious waveguide
• Apart from the Selenium, the other compression drivers all had very good performance

Econowave comments

It is unfortunate that the Selenium driver that is used in the Econowave is not the best unit. Our listening tests have indicated that titanium isn't the best material for a diaphragm and this agrees with the experience of many. The Selenium is easy to use because it has a screw on connection which suits many waveguides. Many of the better compression drivers are bolt-on types. One solution is to remove the thread and create a mounting plate from an acrylic sheet. Another is to use an adaptor. Either way it's important to ensure a smooth transition. Some of the cheaper B&C drivers may be worth considering, although where budget allows, the DE250 is highly recommended. If you already have the Selenium driver, the good news is the sound can be improved a great deal. 

Where does it go from here?

I'm currently listening to some of the combinations that we didn't get to hear on the day. Reports to follow soon. 

We will also be investigating foam mods to the waveguides where we fill them with reticulated foam. 

Waveguide shootout 2 - measurements

See introduction to this event

If you aren't sure what these charts mean, you might like to read about how to read them. They aren't hard to understand once you know what you are looking for. Reading directivity plots >

XT1068 + Faital Pro

We measured the waveguide only, but you can read about this speaker that I have written about previously. It was built in one weekend. Econowave deluxe >

You can see here that the axial response is not flat and that this combination isn't offering constant directivity.

Normalised. Once the axial response is corrected, it turns out that this combination is better than it appeared. At both ends, the response narrows. Compare to our previous measurement:

This indicates that our measurements aren't totally consistent. 

I have spent about a week with this speaker, and I can say that the Faital compression driver sounds better than I recall during our first session. It may be that some driver break-in has improved the sound, or it may be that normal listening is different to our test in which we put the speakers through some torture tests, playing music quite loud that would expose any harshness. Further, this time a high quality mid driver was used and the speaker was placed in an acoustically treated room with bass traps and servo subs. All this suggests it should sound better than in a test designed to show up flaws. 

Overall this is a very nice speaker with a clean sound and great dynamics. Compared to my Miniwaves, the midrange clarity is apparent but the top end is quite similar.

The performance with the DE250 is noticably better. 

Normalised you can see that while the waveguide narrows response at the top, this driver performs better on the bottom end. 

Dayton 10" and DE250

The response here is not as good as what was achieved previously with the Selenium.

The performance is fair. In the normalised plot below, you can see the response gradually narrows with increasing frequency, but from 2.7 - 5k that smooth transition is interrupted. 

Dayton 12" and DE250

Normalised:

Very similar to the 10" version.

Pyle YD-L033 + DE250

Some of the best results of the day were achieved with this combination.

Normalised:

For most of the range, the response is about as good as it gets. 

The Pyle YD-L033 along with bolt/screw adaptors are available locally at DJ city. 

This waveguide accepts screw type drivers, so we had to use an adaptor, which needed to be modified. The hole needed to be made larger and the screw holes were not big enough. Not exactly a nice neat solution out of the box. 

Acapella Plasma tweeter

This is an unusual and exotic tweeter that requires it's own power supply and has a crossover built in. It is a super tweeter with a very extended top end response as shown below.

The directivity control in the top octave is very good, but lower down it varies considerably. 

Continue to conclusions >

BMS 4550 compression driver

Out of curiosity at my last Waveguide shoot out, we took apart a BMS compression driver that was featured. Above you can see the magnet and below the diaphragm.

This is described by BMS as an annular diaphragm. You can see the voice coil and it's former. The diaphragm is the clear plastic ring to which the voice coil is attached. Beneath it you can just make out some radial slots. The sound enters these which form the beginning of the phase plug. The sound waves move downwards in this picture, which is actually towards the rear of the compression driver. They then come out the middle inside the black tube, so the sound waves have travelled through a 180 degree rotation. This is similar but different to the Faital compression driver (HF10TX) in which the sound waves move forward the entire time. The exit tube appears to have a straight tube, followed by a conical section. The transition is not as smooth as it could be, but overall the exit tube is more smooth than the B&C DE250.

Efficiency Bandwidth Product (EBP)

Efficiency Bandwidth Product (EBP) is a number which shows the trade-off between efficiency and bandwidth of a driver. It is useful in determining if a driver is suited for a sealed or vented box and is also used to determine suitability for horn loading.

EBP = fs / Qes

As a general rule:

EBP < 50 - use only for a sealed box
EBP 50 - 100 - can be used in either
EBP > 100 - vented box only

Efficiency and bandwidth

These two parameters are inversely related. For maximum possibly efficiency, the bandwidth must be narrow. For maximum possible bandwidth, the efficiency then becomes less. Due to the assistance of the vent, a vented driver can have higher efficiency. Many pro woofers will reach 40 Hz with a vented box with around 95 db 1w1m. To achieve this with a sealed box requires an fs one octave lower (20 Hz), and in a 12" size this would be a subwoofer driver with about 85 db 1w1m.

One example:

Beyma SM 212 is a pro 12" woofer with an EBP of 102. Placed in a 110L vented box it achieves a -3 db point around 45 Hz. Efficiency is 96 db 1w1m.

Exodus Audio Shiva X2 is a home theatre 12" subwoofer with an EBP of 47, suggesting a sealed box. A sealed box with 45L volume has the same cut off but the efficiency is 85 db 1w1m.

The two illustrate two very different ways of getting a result. The Shiva has high excursion and needs about 1kw to achieve it's full potential output of 112 db. The Beyma can achieve 118 db with only 150 w of input, and higher output can be achieved if tuned a little higher. The Shiva, however, has more output below 30 Hz as expected, being a home theatre sub driver.


What happens if we break the rules?

Since these two drivers are not far from the 50 - 100 range, we can stretch things a little. The Shiva does just fine in a vented box and gains more LFE output. The Beyma doesn't do quite as well because it starts rolling off at 80 Hz. We could still make it work if it had sufficient excursion, but that isn't the case. In other words, it fails to work well for other reasons.

Waveguide GTG #2

Following on from our previous event, we wanted to investigate compression drivers and waveguides. Previously we heard differences and we also measured them, but we were comparing multiple changes at a time. This meant we couldn't be sure what accounted for the changes. 

There is a thread about this even on StereoNET

Our measurement setup included a laptop with Arta and speaker and mic elevated outdoors. We even used some acoustic treatment to damp the first reflection off a wall (not shown). 

We measured:

1. XT1086 WG + Faital Pro HF10TX

2. XT1086 WG + B&C DE250
3. Dayton 10" WG + B&C DE250
4. Dayton 12" WG + B&C DE250
5. Econowave (JBL Clone) + B&C DE250

XT1086 is a waveguide by 18 Sound. The Dayton 10" and 12" waveguides are a low cost plastic moulded unit.

JBL clone and Dayton 10" and 12" waveguides.

BMS 4550 on the left and Selenium D220Ti.

B&C DE250

Faital pro HF10TX

We had some problems with the wind and with the turntable setup, unfortunately meaning the results are not as accurate as last time.

After measuring outdoors, we set up a range of combinations indoors with Behringer DCX acting as an active crossover. The room was set up with 3 very large bass traps, two of them shown below. Those are 1.2m wide 50mm thick rigid fibreglass sheets with about 300mm of foam behind.

Always some last minute jerry-rigging! We had to drill some holes larger for an adaptor to get the bolt on drivers to work on screw type waveguides. 

Removing DE250 from my Miniwaves:

Continue to measurements & impressions

Concluding remarks

DIY Unity horn

Coming soon: I will soon commence a Unity horn project. 

 

The basic concept is a Unity horn built into bass traps. This makes it a very unobtrusive version since I was already  going to build the bass traps. There is considerable volume behind the traps, since the space behind the 150mm panel does not need to be filled in. 

Details

The horn will actually be a waveguide (oblate spheroid) 60 degrees vertical x 90 degrees horizontal dispersion. It will feature a B&C DE250 compression driver, mids are yet to be decided. It will be a 3 way active design with 18" woofers. I plan to run a horn sub below the floor based on 3 x Rythmik servo kits clustered together in a tapped horn arrangement. They will achieve maximum efficiency that way.

The tricky part is to build arrange the woofer boxes so that they don't become a boundary that effectively means the bass trap is no longer very deep. Hence the bass must be free to move around them. If they filled the corner completely, then the traps would have an effective depth that would make them midrange absorbers and not bass traps. I have them in two boxes joined at the middle to avoid difficult cuts in a design that has the Unity and woofers isolated.

Bass trap density

When it comes to bass traps, density matters. Ethan Winer did a test of various trap materials to find out how they compare.

This shows some highlights, with low density 24 kg/m3 vs high density 96 kg/m3.

Slower version shown.

The test is shown here:
http://www.ethanwiner.com/density.html