Flint Acoustics: Rocketman Loudspeaker (one off build)

[Dentman]

Today I finished wiring the drivers up and the speakers are as close to be done as I can get right now. The midrange drivers are brand new, so I decided it would be fun to measure them, break them in, then measure them again to see how break-in time affects the performance of a new high end driver.

Well. I measured their electrical and acoustical performance then ran quite loud varying test tones through the midrange as well music, news radio, and did a conference call where they served as the speakers in my speakerphone. Then I measured them again and compared the to sets of measurements.

And, the verdict is.....


wait for it....

There is absolutely no difference between the first measurements and those made after 10 hours of burn-in. Now, someone in the audiophile snake-oil world of imagined audio "facts" will tell me that 10 hours wasn't enough and that 48 hours is necessary. Well, if that were true, would there be at least some change between freshly out of the factory packaging and after 10 hours of pretty aggressive usage? In fact, I had mentally prepared myself, should there had been a measurable difference, to continue the burn in period until the last measurements were identical to the previous ones - suggesting no more change was expected.

But, alas, this greatly convinces me that me previous experiences with high end drivers were not unique and that a well engineered, properly manufactured driver won't require a burn in period. I have, however, seen mass produced cheap-ass drivers shift as much as 10% from the initial measurements and those taken after a burn-in period.

I would say you really didn't wait long enough for the break in period. It's a proven scientific fact that speakers do not break in until after your 90 day return policy for most credit cards so 91 days of use is the correct break in time.

 

[Flint]

I would say you really didn't wait long enough for the break in period. It's a proven scientific fact that speakers do not break in until after your 90 day return policy for most credit cards so 91 days of use is the correct break in time.

That's the audiophile spirit!

 

[Flint]

About the trapezoidal baffle on the top cabinet. These are not new by any means and dozens of companies have made speakers with cabinets or baffles using this shape since the 1970s when research into baffle acoustics got real. The became more critical when dome tweeters took over the industry in the 1980s as domes radiate sound in all directions forward, especially at the crossover frequency range where they need to pair with a midrange which might be more directional at the same frequency range.

Over the years we have studied, refined, and experimented with what baffle shape produces the most pleasing stereo imaging in average homes and many great designs have emerged, from the high end B&W speakers with spherical midrange enclosures to the huge trapezoidal slabs of granite from Meadowlark to the small egg-shaped enclosures from KEF. I happened to fall love the approach I first saw from Dahlquist which was to design the baffle independently from the enclosures necessary to the best bass performance from the mids and woofers. That just makes sense to me as it simplifies the design process - build the baffle to address the issues associated with how sound propagates into the room, then stick the required enclosure onto the rear of said baffle and enjoy the resulting sonic magic.

The mainstream players didn't ignore all of this, but they also knew the importance of keeping total costs of delivering a product as low as possible while delivering acceptable sounding speakers. If you look at damn near all of the major players, they address edge diffraction on their boxy narrow speakers in different ways, some with horns or waveguides on the tweeters to prevent their radiation patterns from projecting to the side at 90 degrees, or chamfering the edges off, or putting a speaker grill frame onto the speaker with its own waveguide, or even putting sound absorbing materials, like felt or foam, around the edges of the tweeters to absorb the sideways radiating sound. But they did address the problem in some shape or form.

Then they knew they had to address the baffle size issue, so they either used crossover filters to fix the baffle step issue in a simple 2 way speaker (at a 6dB cost to sensitivity and power handling), or they added a second identical mid/woofer crossed over at the baffle step frequency and created the now common 2.5 way speaker, or they went with a 3 way design where the crossover from the mid driver(s) to the woofer(s) is set to the baffle step frequency.

Ultimately, they were able to minimize the issues of baffle design without dramatically increasing the manufacturing costs or complicating the warehousing and shipping. For instance, the benefits of off-set drivers is well known, but that would require a speaker be designated "left" or "right" and then it would require they be shipped and sold in pairs and warehouses and dealers would have to keep track of the pairs so they didn't find during their annual inventory process that they had four "left" speakers for every one "right" speaker due to sales morons carrying out the wrong boxes that were sold.

Since those speakers are so ubiquitous, most people never see a great set of ideally designed speakers properly placed in an attractive room, so when told to look at the more ideal designs they cannot envision how they could make that work in their homes.

I have an old friend who came back into my life recently and bragged about his super high end home audio system and used words like "high resolution" and "high fidelity" to describe his interests. He saw my posts about speakers and wanted to come hear them, which he did. Before listening, we had a drink and caught up on life and during the conversation he kept talking about how incredible his system is. When he finally got around to telling me what he owned I was shocked to hear it was a 7.1 surround system with B&W 700 series front speakers, a matching "larger option" 10 inch subwoofer, and B&W in ceiling surround speakers from a completely line of products. He thought this was an amazing setup. I was polite and didn't say anything to point out his narrow-mindedness. Basically, he did his research the best way he could, and that was to visit the local audio stores and there was one that sold B&W, and he heard their big Nautilus speakers and got the next best thing which he could afford. They don't suck, don't get me wrong. But this is what the world of audio is today. High end to one person is the speaker system which is next to the top of the line at the store. Basically the best is defined by the most expensive the store has to offer. In follow-up conversations he has told me he can never hear the dialog when watching movies and sometimes has to turn up the center channel by as much as 16 dB in order to hear what they are saying. Now, I have no idea how he calibrated his system, but that's crazy to me. He brags about how amazing his system is, but he cannot hear the dialog? He's also complained that the surrounds are often too bright and piercing and he has to turn them down to where he cannot really hear them and it frustrates him. Meanwhile, he has spent a fortune on SACDs, Hi-Rex downloads, and a Hi-Rez streaming player for those downloads. He says he can hear the difference, but in my listening room he cannot hear the difference even when the mixes are obviously different to me. So, that's where we are. I am polite, he isn't into this hobby in the way I am, so I am not going to embarrass him, but he could benefit from some education.

By point? Well, the look of a speaker, in my opinion, should start with great physics and physio-psychology on how we hear sound and stereo imaging. Then it should be practical to build and setup. Then it should be made as attractive as the owner desires within the owner's budget.

 

[TitaniumTroy]

Just asking, why is it that that the Big Wilson’s do not sound as good as the smaller models? Too much bass or boomy bass, plus the bigger ones are harder to dial in with that adjustable time alignment. Would be my guess, I have heard this at shows and show rooms.

I did like the Watt Puppies I heard, at Batt’s place but the they are small compared to the ones that look like Transformer robots.

 

[Flint]

I don't know why one might have sounded better than the other. But, I do know that the bigger the speaker, the more crucial it fits into the room.

 

[TitaniumTroy]

My first exposure to a trapezoid type speaker, blew me away in the imaging and soundstage department.

The Spica TC-50, this came out the 80s and I would still put its imaging up against many of today’s speakers.

 

[Flint]

Today I made measurements of the individual drivers mounted in the system to get an idea of how they perform with the baffle/enclosure acoustics. Then I hooked up a MiniDSP 2x4HD digital crossover to each channel of speakers to make a "tri-amped" active stereo system and tuned the crossover filters on each DSP to have simple crossover filters on each driver to the calculated ideal crossover points, using 48dB/Octave Linkwitz-Riley filters. This was to just get an idea of how the three different drivers would sound together. I adjusted the individual levels to get a relatively flat response at three meters, then sat down and listened.

And I was impressed. I am intimately familiar with the resolution and smooth clarity of the dome tweeters and the power and clean bass of the woofers, but the midrange was new to me. It was delicate, detailed, and resolved. Very nice.

The room is less than ideal, so I am still getting a grip on the stereo imaging and the bass tuning could use some manipulation. But overall the sound was more impressive than I anticipated.

Here's a photo of the setup.

 

[Flint]

Alright, folks.... there's been lots of interest in charts, so here's the initial chart showing the basic frequency response of these speakers in the testing room with the microphone at 1M at tweeter height. Above about 2,000Hz the measurement is gated, so the room is not interfering with the levels, but below 2,000Hz the room plays a role in the response being measured. I also chose to roll off the bass during the initial crossover tuning.


Chart 1: Initial FR of the Rocketman speaker system (1/6 Oct Smoothing)

To my pleasure, the only programs being used in the DSP to get this level of performance are crossover filters. There is no EQ, phase compensation, or even delays. I could improve the performance by adding those, but right now I am just getting a feel for how well these drivers all work together. This is pretty good considering I am only using basic crossover filters.

As I tweak and adjust I'll share more. I also think I need to move them into a more controlled room. They are currently in my large, open front room which is completely open to my entryway and stairs as well as the hallway and dining room area. It's a great room in which to work on stuff because it is open and there is plenty of space. But it does not make for a great listening space for quality speakers.

 

[Flint]

Alright... I spent about a hour adjusting the angle of the top baffle to time align the drivers to a person sitting 9 feet away with their ears at 38 inches above the floor. This eliminates the need for a delay on the tweeter and serendipitously it time-aligned the midrange to the woofer, so without even planning it, these speakers are now perfectly time aligned with a perfect impulse response if the listener is sitting exactly 108 inches from the tweeter with their ears exactly 38 inches above the floor. Sometimes I get lucky!

Next I recalibrated the crossover to get as smooth a response as I could through the two octaves around each of the crossover regions, which was from about 200 to 800Hz and from about 1,400Hz to 5,600Hz.

I then put the mic exactly 1M from the tweeter perfectly inline between the tweeter and the ideal position for the listener's ear for each channel, and captured these response curves:


Chart 2: Left & Right speaker responses from 1M with no smoothing (software captures as 1/128th octave)

Take a look at that... those are completely unsmoothed response curves from loudspeakers in a lively room!!! Above about 2,000Hz they are as smooth as can be had!!!! That demonstrates the power of a good baffle design, tweeter placement on said baffle, and proper countersinking of the tweeter's faceplate into the baffle face. WOW!

No, seriously, look at that again... This is what makes a well designed baffle superior to all the common and more acceptable boxy rectangle speaker cabinets in every possible way!!! Damn!!! I am bragging about this one for some time, folks.

Anyway, here's the same data smoothed with a 1/3 octave filter:

Chart 3: Left & Right speaker response curves with 1/3 octave smoothing from 1M

This is what manufacturers would use to market their products (if they ever show a response curve at all). The only real difference is these datasets were created in a less than ideal room, so the bass is clearly not consistent between the two. There's a slight dip in the 180 - 310Hz range, but otherwise they are very smooth and balanced.

More importantly, the response from about 60Hz to over 15,000Hz is smooth and flat. I am always ranting on and on about the importance of accurate and clear midrange, and one way to help get accurate midrange is to have a flat response throughout that range. These speakers are doing it well, I'd say.

Bass extension of the 11" woofers in this room is very good as well. The left speaker in this case is 4.5 feet from the side wall and 5 feet from the rear wall and has a IEC bass limit of 20Hz while the right speaker is away from all side walls and is 5 feet from the rear wall has a limit of about 23Hz. That's pretty impressive!

I then put the microphone exactly in the center behind the ideal listening position by about 6 inches so the direct line of sight between the mic and the tweeter would pass through the earlobes of a seated listener and made a single combined measurement to get an overall in-room response.


Chart 4: Stereo pair frequency response from listening position, 1/3 smoothing

This was very surprising measurement as I usually get very bad comb filtering from mic not being exactly the same distance from both speakers, but in this case all my care to place it perfectly paid off and I got a flat response almost devoid any comb filtering.

From this response I could argue these speakers are capable of a solid response of 24 - 18,000Hz +/-3dB with a bass limit of 19Hz.


Later I will make dynamic measurements to see how loud they should be able to safely and comfortably play. Capturing the impulse response is difficult to do as the software I use doesn't do decent snapshots, but I'll try to get that as well.

I am currently letting good music play through them for a bit then I'll go do some critical listening with the "Flint's Super-Spectacular Loudspeaker Auditioning CDs" to see if I want to adjust them any more.

 

[team_tsp]

Alright... I spent about a hour adjusting the angle of the top baffle to time align the drivers to a person sitting 9 feet away with their ears at 38 inches above the floor. This eliminates the need for a delay on the tweeter and serendipitously it time-aligned the midrange to the woofer, so without even planning it, these speakers are now perfectly time aligned with a perfect impulse response if the listener is sitting exactly 108 inches from the tweeter with their ears exactly 38 inches above the floor. Sometimes I get lucky!

Next I recalibrated the crossover to get as smooth a response as I could through the two octaves around each of the crossover regions, which was from about 200 to 800Hz and from about 1,400Hz to 5,600Hz.

I then put the mic exactly 1M from the tweeter perfectly inline between the tweeter and the ideal position for the listener's ear for each channel, and captured these response curves:

View attachment 8627
Chart 2: Left & Right speaker responses from 1M with no smoothing (software captures as 1/128th octave)

Take a look at that... those are completely unsmoothed response curves from loudspeakers in a lively room!!! Above about 2,000Hz they are as smooth as can be had!!!! That demonstrates the power of a good baffle design, tweeter placement on said baffle, and proper countersinking of the tweeter's faceplate into the baffle face. WOW!

No, seriously, look at that again... This is what makes a well designed baffle superior to all the common and more acceptable boxy rectangle speaker cabinets in every possible way!!! Damn!!! I am bragging about this one for some time, folks.

Anyway, here's the same data smoothed with a 1/3 octave filter:
View attachment 8628
Chart 3: Left & Right speaker response curves with 1/3 octave smoothing from 1M

This is what manufacturers would use to market their products (if they ever show a response curve at all). The only real difference is these datasets were created in a less than ideal room, so the bass is clearly not consistent between the two. There's a slight dip in the 180 - 310Hz range, but otherwise they are very smooth and balanced.

More importantly, the response from about 60Hz to over 15,000Hz is smooth and flat. I am always ranting on and on about the importance of accurate and clear midrange, and one way to help get accurate midrange is to have a flat response throughout that range. These speakers are doing it well, I'd say.

Bass extension of the 11" woofers in this room is very good as well. The left speaker in this case is 4.5 feet from the side wall and 5 feet from the rear wall and has a IEC bass limit of 20Hz while the right speaker is away from all side walls and is 5 feet from the rear wall has a limit of about 23Hz. That's pretty impressive!

I then put the microphone exactly in the center behind the ideal listening position by about 6 inches so the direct line of sight between the mic and the tweeter would pass through the earlobes of a seated listener and made a single combined measurement to get an overall in-room response.

View attachment 8629
Chart 4: Stereo pair frequency response from listening position, 1/3 smoothing

This was very surprising measurement as I usually get very bad comb filtering from mic not being exactly the same distance from both speakers, but in this case all my care to place it perfectly paid off and I got a flat response almost devoid any comb filtering.

From this response I could argue these speakers are capable of a solid response of 24 - 18,000Hz +/-3dB with a bass limit of 19Hz.


Later I will make dynamic measurements to see how loud they should be able to safely and comfortably play. Capturing the impulse response is difficult to do as the software I use doesn't do decent snapshots, but I'll try to get that as well.

I am currently letting good music play through them for a bit then I'll go do some critical listening with the "Flint's Super-Spectacular Loudspeaker Auditioning CDs" to see if I want to adjust them any more.

Amazing job as always!

 

[Flint]

Amazing job as always!

Thanks!

You want to buy some speakers? I have a few sets.

 

[Haywood]

In some future year when I'm not about to drop $5k on medical bills for my wife, you and I definitely need to chat. Your speakers look amazing.

 

[Flint]

In some future year when I'm not about to drop $5k on medical bills for my wife, you and I definitely need to chat. Your speakers look amazing.

Wow! Thanks!

I really do appreciate that.

 

[TitaniumTroy]

I bet some Deep Purple sounds really awesome on HeMan's system.

 

[team_tsp]

Thanks!

You want to buy some speakers? I have a few sets.

It's really tempting lol! If only you had started building speakers (to sell) before I put together my new system last year. But similar to what Haywood mentioned, I'll definitely be interested in something sometime in the future!

 

[Flint]

To better understand how these speakers perform for off-axis output, I decided to properly and precisely measure the response at 5 degree increments at tweeter height both from -90 degree to +90 degrees (inside to outside, since the tweeter and midrange are offset on the baffle).

The resulting data is then used to make a polar plot:


The scale on the left shows the angle off of dead center, the negative numbers being to the inside and the positive numbers to the outside. The heat map scale is on the right, red being the highest levels and purple being to lowest levels. The scale on the bottom is for frequency. I chose to limit the measurements to frequencies above 905Hz as that was the lowest I could reliably capture a gated sound where all room reflections are filtered out. The chart shows absolute SPL by frequency and angle off axis on the horizontal plane.

That chart is pretty, but a better chart would be if I normalize all the measurements to the 0 degree measurement so I can see how the sound changes as I move to the sides.


By normalizing to the perfect on axis response, the chart now shows how the sound changes from on axis as you move to the sides. That's why is the second polar plot the middle horizontal line is solid red and it moves to yellow and green and you look above and below the middle.

As for my impressions of the chart, the roll off of the treble as you move to the sides is outstandingly even, except for the slight issue at the range just above 5kHz. But, those anomalies are minor. In fact, the off axis performance is damn near ideal from 900Hz to 5kHz, which means the mating of the tweeter to the midrange is perfect. Both drivers are demonstrating the same dispersion characteristics at the crossover frequency, which is in the upper 2,000Hz range.

What's also interesting is the since I offset the drivers on the baffle, the anomalies to the inside are at slightly different frequencies and in different shapes than the anomalies to the outside. This balances the issues and makes for a more natural in-room power response.

Also, the dome tweeter has a very uniform dispersion in the range above 10kHz and is nearly perfect within a 30 degree window, so aiming directly at the listener is not absolutely required. In general, if the speaker is aimed within 15 degrees of the listener's ears, they will sound balanced and natural and provide all the top end treble one might need.

I am extremely pleased with the dispersion characteristics and this reinforces the excellent stereo imaging I have already experienced. Also, the smoothed and balanced off axis performance means these speakers will sound good in a highly reflective room or any installation where the side walls cannot be treated to the absolute ideal. A perfect room is still preferred for the best performance, but if these are not in a perfect room, they will sound better than most speakers out there which suffer from unbalanced off axis performance. This is one area where I don't like mating compression horn drivers or ribbon tweeters to cone midrange drivers. Metal dome tweeters have similar off axis issues, especially at the higher frequencies.

I am very pleased with these results.

I don't know if I explained this well enough. Maybe I'll measure some commercial speakers to show the difference with a normal speaker.

 

[Flint]

I am going to talk some more about edge diffraction using the polar response plot which is normalized to the on axis response:


The least amount of variation from the on axis output is the red area, as you get as much as 3dB below the on axis response you see orange, 6 dB is yellow, and 9 dB starts looking green.

Since I went to so much trouble to reduce all edge diffraction by sinking the tweeter and midrange into the baffle so their front faceplate (tweeter) and frame (midrange) are flush with the baffle face. I also used a very large 1.125" roundover for the edge of the baffle which reduces the edge diffraction to nearly zero. Still, the tweeter has six screws/bolts sticking out on the fronts of the faceplate. The midrange also has four screws on it's frame. Although the junction between the edge of the faceplate and the baffle is extremely good, it isn't absolutely perfectly aligned without any edges or corners. Also, what little edge diffraction might still occur from the baffle edge can impact the sound a small amount.

So, you will see some cancelations in the off axis response as the diffraction which does occur will vary in intensity and frequency as the mic is moved off axis.

Looking at the upper half (inside angles) of the plot above, you can see traces of yellow in the region just above 2,000Hz which is not present in the lower half (outside angles) of the graph. The next area you see yellow is in the upper 3,000Hz range, but the shape of those slightly yellow areas are quite different on the upper (inside) region than the lower (outside) region. This goes on as you go higher in frequency. If you look at the shapes of the yellow areas, they are often at different frequencies and intensities between the inside and the outside measurements (upper and lower, respectively).

Basically, all my effort to address edge diffraction by either reducing the intensity (rounded edges, trapezoidal shape) or varying the frequency (offset drivers and top distance from tweeter) have smoothed the problems so they are very much inaudible in practice.

I am very pleased, indeed.

 

[Flint]

I also made a cumulative spectral decay plot of the speakers with the mic at 0.5M to reduce room reflections. I limited the measurement to 5mS to reduce room issues as well.



It would help if I had identical measurements from other common speakers (which I can make later and share), but I am VERY pleased with how there are almost not serious resonances or ringing issues with these drivers or their baffle reflections. There is almost zero ringing with an amplitude of more than -30dB beyond the 2mS range. Of course, these are some the finest tweeters money can buy, so I am not surprised by them, but even the midrange shows good characteristics.

This fast decay helps lead to significantly better clarity and resolution during playback as there is less noise. I will make the same measurement with my SVS MBS-01 speakers later so you can see the difference.

In addition to using very good drivers, this aspect of performance is improved by using an active crossover, which I speak to in depth in the thread I created on the evils of passive crossovers.

Good happy times.

 

[PaulyT]

Awesome. Interesting info, thanks for all the detail.

 

[PaulyT]

Yeah it'd be cool to compare to some other speakers, like the C1 or something (though if you did exactly those @heeman would probably have a coronary worrying about it ).