Unity Horn taming a Lambda Acoustic Unity

I purchased a pair of Unity Horn kits from Lambda Acoustics nearly 10 years ago (hard to believe) that got put aside in favor of planar, dipole designs that I have been using for some time. This summer after reading William Cowan experiences with both Planar systems and Unity systems, I decided it was time to complete my Unity project. I had already purchased a used but just re-coned pair of TAD 1603 woofers to mate to the Unity Horn. Learning from Williams past experiences, I decided to built the system as a corner system. Not as elegant as his built in the wall system, but trying to achieve many of the same benefits of corner placement.

Attached it a picture of the rough (no finish yet) speaker in my living room.

I will use the Unity horn by using a DEQX digital processor for crossover and EQ. As soon as I hooked it up and ran some DEQX curves, I found some nasty things going on that were unexpected. The DEQX kept coming back with correction curves with a huge peak at 4.5K. If I by passed the correction curve portion and used it as a straight crossover, things started to sound promising, but I wanted to find the source of the issues with DEQXs correction curves.

I did some measurements. I combined some of the measurements on the attached graphs. I discovered the source of the 4.5K problem with DEQX. On axis, the Unity from Lambda has a sizable dip at that frequency which the DEQX tries to correct. The problem, is that the dip in only on axis and goes away off axis. I did a little web search and found references to the dip being caused by the mid-range holes in the Unity Horn, and a hint that the implementation of the Unity Horn by Lambda, was not exactly what Thomas Danley did on the original. I could not find any definitive references to what was different, but did see other DIYers efforts to experiment with the same issue with other Unity clones. I have also noticed that the radius of the curvature in the “seams” of the horn seem much greater on the Lambda version compared to sharper radius on other I have seen photos of.

Measurements. Attached graphs high frequency and Mid range portions measured separately. FYI they are all 1/10 octave smoothed (to leave detail), and scaled up and down to allow several on a page. Outputs are not absolute.

On the first graph, (Unity DE25.jpg), the Red Line is the stock finished Unity response from the DE25 (image of stock unit holes attached). The 4.5K dip is clearly visible. (I did do off axis measurements that showed that this largely went away off axis but for this discussion, I have not shown them). Next I covered the mid-range ports with metal foil tape. The gray line shows this change. The 4.5K dip is gone and the whole response gets soother. So, I know the ports are the problem. Next I poked holes into the foil - roughly 1/4 in diameter (image attached) and measure again - this is the Green curve. Most of the improvements are maintained. The last curve was to replace the thin foil with a thicker rubber washer with a 1/4 hole drilled in it. There is also a picture of that and the result is the blue curve - which is nearly identical to the hole in foil curve.

So far so good. Now to see what happens to the mid range drivers. The second graphs (Unity Mid.jpg) shows the curves for the mid range. First the red line is the stock Lambda ports. What the red graphs shows is that the stock Unity has elevated response at 300 Hz of nearly 10 dB above the level at around 650 Hz.
I did not do a gray graph with the ports covered as no sound comes out.
The green graph shows the response with the foil with 1/4 holes. The graph is scaled down, but in absolute output, it is within about a dB in the middle of the pass band, and the 300 Hz peak is suppressed a little and moved lower in frequency.
The blue graphs shows the same thing with the rubber washers and again is nearly identical. While the foil and washers now seem on the right track to work, I find both unacceptable in appearance.

So my next step is to find a more refined and permanent way to do this. So I took a piece of birch plywood, drilled holes in it the diameter of those in the unity horn (3/4”)and then filled one with Bondo and the other with Durham’s Rock Hard Water Putty. Then I drilled a 1/4 hole all the way through them, and finally drilled a 5/8” hole from the back side to within about a 1/8” all the way through. This can be seen in the attached photos. So it appears that either of the two material will work, I think I like the Bondo best.

Questions to anyone with knowledge of Unity or Synergy Horns.
1. Does anyone have any knowledge of the preferred geometry of the holes for the mid-range drivers for Unity?
2. Will the 1/4” holes cause any distortion problems (bearing in mind that this is a home installation - not a PA app)? I have no way to measure distortion.
3. Can anyone explain the tuning of the ports? The mid system appears to have double resonant humps at 300 and 1000 with a valley in between. Any suggestions on tuning these - more cavity in the port or less, taper vs cylindrical, etc. I did try filling the original holes with damping material and the pass band response dropped significantly, so I am not inclined to use damping.

Bwaslo,

Thanks for your reply. I agree that the 4.5kHz hole is very narrow and since most listening is not done on axis, it is not very audible. The problem with the dip is how it is handled by DEQX. Unlike some other speaker management systems, DEQX measures speakers and creates it's own correction curves (for both FR and Phase) which makes it unique and the results can be spectacular. The on axis measurement with the dip causes a big correction peak which sounds awful because that is not what is happening in the power response. The correction curves cannot be modified to eliminate the peak. In using DEQX you always work on finding the best measurement position. When you get there - the results are spectacular. So that is my motivation to acoustically optimize the port holes.

I tried measuring the Unity off axis and generating a curve and there is no problem with the 4.5kHz dip (and corrective peak), but because there is some HF rolloff off axis, the corrective curve is a little hot. So, I still want to acoustically tame the dip as much as possible. I think I have a good solution planned, but before I bondo up the holes and start making new holes as I described, I am hoping to hear from others about their experiences. (I was hoping that Thomas Danley might see the link and tell me what he found as optimum - from various forums discussing Unity Horns, it appears that the Lambda version may have had some significant differences from those made by Sound Physics).

Thanks, Ray

PS Am i correct in my guess that you are the BWalso of IMP and OmniMic fame? I wish I had OmniMic. My 20 year old LMS/LEAP system has served me well, but is showing it's age and the new LinearX replacements are way beyond my now current budget (I'm now retired).

Thanks again.

Thanks Bill and Coctostan, I did try an off axis, but it was probably more than 22 degrees. After I modify the mid port holes I will again remeasure things and find a good slightly off axis measure point.

Just to clarify what DEQX does (as I am convinced that what they do is unique and better than anything else out there). Nearly all other "loudspeaker management systems" have crossovers with choice of freq. and slopes, and usually several bands of parametric EQ that can be applied to each band and sometimes also some that can affect the entire band (before x-over).
DEQX has a different approach. It measures each driver in a system at close range and provides a correction curve for each driver independent from the others. The correction includes both Frequency Response and Phase response. The goal is to get each speaker producing the flattest response with uniform phase integrity. Crossover frequencies and slopes for the individual drivers are user configurable. These are far more complicated than simple parametric EQ and the correction curves cannot be adjusted after the fact. THEN, after the driver in speaker is optimized to be producing the best possible sound and you have configured the crossover, you make a measurement at the listening position in the room and to that you can apply many parametric EQ bands. I don't remember how many bands it has but it is more than I have ever needed. So, it can do as much parametric EQ as you might want, but you cannot modify the very sophisticated driver corrections because they are doing much more than just FR correction. They are the reason that when the driver corrections are executed well, they have a dramatic effect.

The reason that this is an issue with Unity is that the on a very narrow band of on axis is so different than the off axis or power response. I have use DEQX on many different types of systems and find that the DEQX process delivers an integration of sound not achieved by FR compensation only.

I am aware of someone else using DEQX with Unity, but they used a passive crossover for the Unity, and then used the DEQX crossover for a mid bass driver and a sub. I am using the three bands of DEQX crossover to use High for the DE25 compression driver, Mid for the Mid drivers in the Unity, and Low for the TAD woofer. This should optimize both frequency response and phase response for both the Unity and the whole system. The only passive components will be a capacitor on the compression driver to protect it from any low frequency components from direct hookup to the amplifier.

Next step to take everything back to the garage to disassemble the horns for port modifications, and for putting veneer on the enclosure. I have limited experience in veneering, so that should be interesting. I have read a lot about it, and know that veneering something this big will be challenging.

Coctostan,

I am sorry if I have contributed to this drifting off topic. The topic is the Unit Horn and the issues of optimizing the mid range port opening.

The DEQX was mention in regard to how it reacted to the anomalies of the Unity on Axis. I only explained how the DEQX works to explain what it does - not wanted to cast it in a bad light. I am a big DEQX supporter.

I didn't elaborate much in my first post but I chose a corner system with a controlled directivity horn for many of the same objectives to those in the article you site - namely much direct sound and little to no short time duration reflections. This does create a "window of sound" in the corners that makes the walls seem to disappear - you hear things beyond the walls. The effect is very stricking - revealing more of the decay of the sound around instruments than I have heard in other systems. I used similar principles to place dipole speakers raked in roughly 45 degrees, well into the room, so that there is no reflected from the sides of the dipole (by definition), resulting in a predominance of direct sound over any early reflections.

I am still hoping to learn more about others experiences with Unity Horns - either home design, Lambda kits, or those from Sound Physics or Danley Sound Labs.

Thanks for the link Nate. It turns out that I had come up with a solution like the Fig. 10 in the patent, and was in the process of implementing a final version of it on my Unity Horn. First I filled the existing ports with Bondo. I put tape over the inside opening to keep them flush and filled from the back. Then I finished them flat on both inside and back. Next I drilled a 5/8 in hole leaving about 3/16" material at the inside opening. Then I drilled a 1/4" hole in the center. Here are some pictures of the progress. I haven't reassembled them yet to measure but fully expect it to be as good as if not better than the washer versions or the foil with holes versions above (which were nearly identical).

I will post before and after test results after everything is back together.

Thanks again for the link - it is nice to know that Danley arrived at the same solution.

Re: distortion
I did not hear any noticeable distortion of either the foil with 1/4" holes or the washers with 1/4" holes. Also, the output level was not diminished either. Smaller holes did reduce the output, as did any form of damping in the ports. From the patent drawing, if the original port size is 3/4" the smaller opening would be about 1/4". I will be using these in a living room for music, so I will not be pushing SPL limits anyway. I wish I had an Omni-Mic to measure distortion and know the read impact of the mod. My old LMS system still does some things very well, but measuring distortion is not one of them.

Update on Unity project. Both cabinets are done, Cherry veneer is on both, (that was a learning experience - but with careful planning, reading, and make test boards first, it all went well). I have put two coats of stain and two coats of polyurethane on one side. The other side will start tomorrow. I have not modified the mid-range port holes on the other side yet so I can compare performance of modified to un-modified before I modify the second Unity Horn.

Thanks to all for input, advice, and encouragement. I still planning on doing more measurements and posting them. Here is a photo of the pair.

Here are some results (hard to say if it is ever Final) of Lambda Unity Horn project.
By modifying the port openings I eliminated the suck out interference dip at approx. 4.5kHz. The modified ports also smoothed out the mid driver as well, though the smaller holes decreased mid output several dB. This could result in high distortion (I don't have any means to measure), but I sure don't hear any distortion at any volume I can tolerate at home!

Using the DEQX HP3 configured as a three way system with Linear Phase 96 dB crossovers at 300Hz and 1kHz. Attached is a LMS measurement of the results. On the graph, the red line is the 1/3 octave smoothing of the Unity Horn with the DEQX crossover. I included the Un Smoothed lower green curve as well just to see how flat it is even without smoothing. The top blue curve is the response including the TAD woofer, measured from a point between the woofer and the Unity Horn, probably still slightly off axis of the horn. All measurements were made at 1 meter.

How does it sound? Well, I'm pretty happy. I am hearing things in recordings I have not heard before. I do (like William Cowan) find that a gentle downward roll off is preferrable, at least on some close mic-ed recordings, while others sound fine when totally flat. My wife thinks they look nice, and if I clean up the living room, I can then just enjoy.

BTW, though I am not using a passive crossover, I plan on making one as an option. I did quite a bit of modeling in LEAP which will show you a final output with the filter components you choose. It starts with accurate SPL and Impedance measured by LMS. The original Lambda circuit isn't too bad (using the data from my modified port horns), but the other "later" circuit did not model very well with my modified horns. If I optimized the later circuit big improvements could be made by some radical changes of just resistors, but more could be done by change LC values as well. Anyway, I included a screen shot (taken with my phone!! Old DOS program has no image capture) of one optimized crossover. As you can see on the lower left screen, it has lots of parts. I will do more work on this before ordering or building.

Thanks to all for the help and encouragement.

LTDO2
Re: Bass response. I hope the wink means your kidding. The bass curve is a result of in room measurement. Measured at 6", other than a natural roll off below the 40 Hz mounted resonance, the bass is much flatter and any peaks and dips are pretty much a function of mic placement, not driver. The DEQX does have (in addition to crossover and driver correction) a room correction parametric EQ which is done from the listening position. That is not part of any of these curves.

UFOKILLERZ
Re: Yorkville u215. I have not every heard or seen in person the Yorkvilles, but in doing my exploration on this project did read a bit about them, and looked at pictures. I have read good comments by posters on the Yorkvilles, so you may not need to do much except improve their WAF - ;-) Do you have some measurement tools?

It looks like they tackled the mid port issue by using elongated port holes, and there are only 3, so I am guessing that nothing needs to be done to correct for that anomaly that was present in the Lambda Unity. Areas that might be improved upon could be Cabinet - both to eliminate any panel resonances and perhaps change grille and horn mounting to eliminate diffraction. The other is crossover. Again, don't know if it needs help, but crossover components are expensive and most commercial products do the minimum for acceptable sound. That being said, modeling crossovers is difficult. Many crossover or circuit design software packages will model the network results into a static load, but not into the complex impedance of the actual loud speaker, let along let you optimized multiple drivers. LEAP from Linear X does that, but costs more than the speaker. That is why I am using my 20 year old DOS version of LEAP for crossover modeling. It still works great, but is just not user friendly.

Enjoy your new speakers.

FOH

Yes, I know and agree. I don't have a place to put it yet, but when I remove the subwoofers of the previous system from the room, then I can move it. One good thing about the Unity horn is the controlled directivity minimizes the impact of the cabinet.

Thanks for the feedback