Home » Sponsored » Pi Speakers » H290C Horn/Waveguide (Upgrade for the obsolete Eminence H290)
Re: H290C Horn/Waveguide [message #76444 is a reply to message #76441] Tue, 23 April 2013 10:28 Go to previous messageGo to next message
Wayne Parham is currently offline  Wayne Parham
Messages: 18793
Registered: January 2001
Illuminati (33rd Degree)

zheka wrote on Tue, 23 April 2013 09:59
I just recently noticed that H290C directivity sonograms are plotted over linear frequency scale. Do you have versions of the same plots but with logarithmic scale?

I don't, Жека, sorry. But I'll try and get some for you, as time permits. I was really more focused on the directivity midband up, because down in the crossover region, the midwoofer and tweeter patterns blend anyway. In the crossover overlap band, tweeter directivity is shadowed by the midwoofer.

That's also why I didn't consider a secondary flare to combat waistbanding. I wanted the flare to be purely OSEC, to make the wavefront propogation stay perpendicular to the walls as much as possible.

Re: H290C Horn/Waveguide [message #76446 is a reply to message #76444] Tue, 23 April 2013 11:03 Go to previous messageGo to next message
zheka is currently offline  zheka
Messages: 80
Registered: June 2012
Location: Chicago burbs
Viscount

Wayne Parham wrote on Tue, 23 April 2013 10:28
I don't, Жека, sorry. But I'll try and get some for you, as time permits. I was really more focused on the directivity midband up, because down in the crossover region, the midwoofer and tweeter patterns blend anyway. In the crossover overlap band, tweeter directivity is shadowed by the midwoofer.

That's also why I didn't consider a secondary flare to combat waistbanding. I wanted the flare to be purely OSEC, to make the wavefront propogation stay perpendicular to the walls as much as possible.


I am not sure I can tell from the sonograms where the waistbanding takes place. The normalized ones seem perfectly uniform all the way down to 1K.

http://www.pispeakers.com/Measurements/H290C_Horizontal_Contour_Normalized_6dB_strata.jpg

is that because of the scale?
Re: H290C Horn/Waveguide [message #76447 is a reply to message #76446] Tue, 23 April 2013 11:15 Go to previous messageGo to next message
Wayne Parham is currently offline  Wayne Parham
Messages: 18793
Registered: January 2001
Illuminati (33rd Degree)

Yes, it really is. More so the algorithm that creates the stratified sonogram, I think. It appears to try and fit each curve into 6dB slots. In that case, I think the individual curves probably jump from 5dB to 7dB, so the 6dB bin isn't filled until the curves that reach 7dB, something like that. Because you can see the waistbanding in the bottom octave in the 12dB strata, but not so much in the 6dB strata on that sonogram. You can see it a little better on the unstratified charts and also on the chart that isn't normalized. But as I said, it's not something I am really concerned about, since the midwoofer and waveguide patterns blend in that region.

Re: H290C Horn/Waveguide [message #76448 is a reply to message #71929] Tue, 23 April 2013 11:20 Go to previous messageGo to next message
zheka is currently offline  zheka
Messages: 80
Registered: June 2012
Location: Chicago burbs
Viscount
got it, thank you.
Re: H290C Horn/Waveguide [message #76563 is a reply to message #76448] Tue, 30 April 2013 13:15 Go to previous messageGo to next message
Wayne Parham is currently offline  Wayne Parham
Messages: 18793
Registered: January 2001
Illuminati (33rd Degree)

In the chart above, you can only really see evidence of waistbanding in the 12dB strata. As I said, I think this is largely caused by the renderer, which seems to fill the strata with any values less than the stratification level without any interpolation or curve fiting. The waistbanding level is less than 6dB lower than midband, so it just doesn't show up.

Also, the scale is logarithmic which reduces the size of the waistbanding region, making it more difficult to see.

To help better illustrate waistbanding in the H290C, we can change the scale of the sonogram to logarithmic. That stretches the area where waistbanding occurs. See the charts below:

https://www.pispeakers.com/Measurements/H290C_Horizontal_Contour_Log_Scale.jpg
H290C horizontal sonogram using logarithmic scale


https://www.pispeakers.com/Measurements/H290C_Vertical_Contour_Log_Scale.jpg
H290C vertical sonogram using logarithmic scale


These are sonograms made for the H290C using a different renderer and plotted on a logarithmic scale. The measurements were made in a different environment too, so there are slight differences in the data but it shows general agreement with the other measurements. The biggest difference is in the rendering and the logarithmic scale, where here we can sort of put the waistbanding region under a magnifying glass.

See the color scale at the right side of each chart. This legend shows us a color gradient mapped to SPL.

You will notice that horizontal waistbanding centers around 1.8kHz, and its effects are about 2dB reduction of sound at 45° compared to the midband levels. That's what waistbanding does. It reduces output slightly at the edge of the beam.

As a sanity check, we can calculate waveguide beamwidth using formulas from Keele's paper, "What's so Sacred About Exponential Horns?"

Pattern Control Lower Limit = (106 / Mouth Width * Wall Angle)
Waistbanding Frequency = 1.5 * Pattern Control Lower Limit
Waistbanding Pinch Angle = 2 * Pattern Control Lower Limit / 3

Therefore, the H290C has properties somewhere in this range:

106 / 11" * 85° = 1070Hz (lower limit of pattern control)
1070 * 1.5 = 1.6kHz (waistbanding center frequency)
2 * 85 / 3 = 60° (waistbanding minimum beamwidth angle)

Earlier measurements showed waistbanding centered around 1.4kHz. The measurements above show it to be centered around 1.8kHz. Both are reasonable, given the different environments they were taken in, and both match our expectations of 1.6kHz fairly well. One is a little higher than predicted, the other a little lower, but both are within 10% of what Keele's formula predicts. In each case, you can look at the charts and see less output at the waistbanding minimum frequency than midband. It falls a couple decibels more in the waistbanding region than it does at higher frequencies.

In my opinion, the 2dB loss at 45° from waistbanding is acceptable, especially since the woofer and tweeter are blended in the crossover region anyway. Even if they weren't, as is increasingly the case with the larger midwoofers, what we're really talking about here is a slight decrease in output at wide off-axis angles. By slight, I mean practically nothing. The 2dB drop at wide off-axis angles is completely inaudible.

Waistbanding is more damaging in a prosound implementation. In that application, the problem is not so much the slight squeeze of the pattern as it is the secondary lobe, which affects arrayability. But in this case, we're not concerned with that. A two-way or three-way speaker using an H290C waveguide has no other sources with radiation at angles that would create an interference pattern with the secondary lobe.

The thrust of the H290C design approach was to shift all anomalous behavior as low in frequency as possible. The idea was to sacrifice a little bit of waistbanding at the bottom end for smoothness from midband up. Crossover occurs between 1.2kHz and 1.8kHz, with that region blending woofer and tweeter output together. So by optimizing tweeter performance above that point, we are optimizing where tweeter fidelity is needed the most.

What we gain in the trade is a flare profile that keeps the wavefront propogation perfectly perpendicular to the wall angle all the way through the horn. We also gain better acoustic loading, which is important in a conical horn or waveguide, because they are characteristically weak in this regard. They need all the help they can get. Improved acoustic loading provides smoother response, greater efficiency and lower distortion.

We haven't talked much about distortion, but consider that for a moment. By increasing efficiency, you reduce drive requirements. So not only does the improved acoustic load reduce diaphragm excursion, but it also increases efficiency, which reduces the drive requirements as well. This, in turn, reduces excursion even further.

A 3dB increase in sensitivity means 1/2 power is required to obtain the same SPL. So the distortion reduction is improved by two mechanisms, one being the reduced excursion from loading, the other being reduced excursion from reduced drive signal. The improvements from horn loading are cumulative where distortion reduction is concerned. This is true in all horns, basshorns and midhorns, but even more true in compression drivers, because they are designed for use where loading is good. They do not have much excursion capability, because they're designed to be used on a horn, not on a baffle.

Re: H290C Horn/Waveguide [message #77006 is a reply to message #76563] Wed, 12 June 2013 10:45 Go to previous messageGo to next message
Wayne Parham is currently offline  Wayne Parham
Messages: 18793
Registered: January 2001
Illuminati (33rd Degree)

More information in the thread below, including implementation considerations and comparisons with other designs:
Re: H290C Horn/Waveguide [message #78533 is a reply to message #77006] Tue, 03 December 2013 16:32 Go to previous messageGo to next message
Wayne Parham is currently offline  Wayne Parham
Messages: 18793
Registered: January 2001
Illuminati (33rd Degree)

There were several influences that drove my decision to create this horn. The most important one was an observation that radial horns having relatively constant horizontal directivity sounded best to me. Objectively, they had smooth response and uniform horizontal directivity. Their physical features were much like modern waveguides, having no sharp edges in the flare.

The H290C is very much like a radial horn of similar size, except that its OSEC flare profile is defined as the hyperbola created from a line drawn tangent to an elliptic cylinder or oblate spheroid:

../misc/OS_and_EC_flare_profile.gif


When I first heard about this flare profile, I began to study it and found this shape was used in other disciplines as an antenna, horn, lense or other waveform radiator. This is an important "litmus test" for me, because the most successful ideas tend to be used in all engineering fields. Where I see things in just one field (and especially in audio), I am skeptical. Audio is saturated with fads and misguided pseudo-science, some even coming from big names in the industry. But this isn't one of them, and devices created using this profile provide fairly clear advantages.

However, the advantages are only found if (and this is a big "if") the waveguide/horn is properly built. Just having this basic shape isn't enough. Just like you can make a good exponential horn or a bad one, or even more so a good conical horn or a bad one, this flare profile has a few "sweet spots" and a bunch that aren't so hot. One is well advised to model the horn, optimizing it for best response and beamwidth/coverage. It isn't a "given" that this particular profile can be used no matter what exit angle is desired. Some work well, but some, not so much.

That was the basis for the last post, just above. It gives a feel for a well designed horn, and one that isn't optimized as well. Both have the same coverage pattern, but one has significantly more ripple.

More information:

Re: H290C Horn/Waveguide - closest equiv. in 1.5" or 2"? [message #92335 is a reply to message #78533] Thu, 17 September 2020 14:41 Go to previous messageGo to next message
tubino is currently offline  tubino
Messages: 13
Registered: May 2009
Chancellor
It has been almost seven years since Wayne posted this amazing analysis and explication of waveguide optimization for room use, and a compelling case for the H290C. I bought my pair and I'm convinced.

For a different system though I want to use a larger compression driver, with 1.5" or 2" exit. Wayne, are you aware of a large format waveguide that is consistent with the design principles in the H290C? There are a lot of constant directivity 90x40 waveguides out there and I'm pretty lost.
Re: H290C Horn/Waveguide - closest equiv. in 1.5" or 2"? [message #92337 is a reply to message #92335] Thu, 17 September 2020 23:09 Go to previous messageGo to next message
Wayne Parham is currently offline  Wayne Parham
Messages: 18793
Registered: January 2001
Illuminati (33rd Degree)

I put all of my focus on the H290C, and it's predecessor, the H390C. That was a wood horn/waveguide with the same profile but with much more mouth radius because it was designed to be free-standing and therefore have no baffle loading. But both are for 1" exit compression drivers.

I was able to do most of the acoustic development work back when I was making the H390C because it was cut on a CNC machine. So I was able to cut a physical model, test it and make incremental improvements.

Sometimes, lessons learned with one size driver don't transfer to another. But I find this more true of cone drivers than of compression drivers. So some of what I learned while optimizing the 1" drivers might transfer to larger drivers.

Here are a few main points:

1. The basic flare starts with an oblate spheroidal profile and then transitions to a flare based on elliptic cylindrical geometry. Said more simply, the throat is round and the mouth is rectangular. The same OS expansion equation is used to plot each axis, but the asymptotic angle is different in order to get the 90° x 40° shape.

A similar profile can be created with a OS to PS transition. That's oblate spheroidal to prolate spheroidal. It makes a very similar horn/waveguide but the mouth is oval-shaped instead of rectangular.

Each of these three coordinate systems are related: OS, PS and EC. They all make very similar flare profiles, just with different shaped mouths. Oblate spheroidal horn/waveguides are round, prolate spheroids are oval and elliptic cylinders are rectangular.

2. I find that acoustic loading is important. Waveguides are closely related to conical horns, which don't load well at low frequencies. But they still provide some acoustic loading and I think this is important to sound quality. So do not discount the quality of acoustic loading.

3. Radiusing the mouth edge, gradually opening up wider than the OS/EC profile, helps reduce waistbanding at the lower frequency range. With the right amount of radius area, you can completely eliminate waistbanding. But the radiused area does not contribute any acoustic loading. So it's almost like the horn/waveguide length ends at the point where the OS profile ends and the radiused profile begins.

4. Another thing that affects acoustic loading is the expansion rate. A horn/waveguide that has less expansion loads better. But this necessarily limits the overall coverage angle. I found empirically that a 90° x 40° waveguide is about as wide as can provide enough acoustic loading to offer smooth response. Any wider and the quarter-wave modes become pronounced, especially if a lot of the profile is sacrificed for waistband-reducing radius. For example, the H290C is actually a smidge under 90° in the horizontal, and it uses baffle loading rather than a lot of radiusing at the mouth.

5. If large size is not a problem, then you can have enough mouth radius to completely eliminate waistbanding and still have enough length to provide good acoustic loading. This makes a pretty large device though, and the extra mouth area prevents close vertical spacing between the waveguide and the adjacent driver in the system. This can make it hard to get the forward lobe tall enough - The vertical nulls might narrow too far. But with a low enough crossover point, the system can have wider vertical spacing between subsystems and still have a nice tall forward lobe with widely spaced vertical nulls.
Re: H290C Horn/Waveguide - closest equiv. in 1.5" or 2"? [message #92338 is a reply to message #92337] Fri, 18 September 2020 06:40 Go to previous messageGo to previous message
tubino is currently offline  tubino
Messages: 13
Registered: May 2009
Chancellor
Thanks, Wayne. That will take me a while to digest, but it confirms my suspicion that the common PA options are limited in several ways...

BTW, so you have any of the wood 390C available to sell?
Previous Topic: 4pi flat pack, Rhino Line
Next Topic: 3Pi and 4Pi Plans Request
Goto Forum:
  


Current Time: Fri Dec 13 05:48:53 CST 2024

Sponsoring Organizations

DIY Audio Projects
DIY Audio Projects
OddWatt Audio
OddWatt Audio
Pi Speakers
Pi Speakers
Prosound Shootout
Prosound Shootout
Miller Audio
Miller Audio
Tubes For Amps
TubesForAmps.com

Lone Star Audiofest