[Kevin Haskins]

I enjoy working on unique projects.   This is one of them and I'm pleased to announce some of the details on the Exodus Tweeter. 
  
First off, here are the design goals.  I approached this project looking to achieve some goals, which I'll outline here.  This is not going to be an inexpensive product.   I basically approached this with a cost is not an object perspective.

•    Smooth on/off-axis response through the usable bandwidth.   This is a no-brainer.   You do the same thing with any driver you pick.  If it has response issues, you have to make sure they are correctable with the crossover.   Since passive crossovers are pretty much limited to shaping in about 1/3 octave ranges, you cannot have any high-Q resonances that become difficult to tame with traditional (non-DSP) tools.   These narrow high Q resonances tend to be difficult to shape and they tend to spawn non-linear distortion problems that cannot be cured with the crossover.   Our design had to be free of response problems both in terms of the on/off-axis FR and CSD. 
•    Low Distortion throughout usable range with most of the focus on the low-end.    Distortion on the low-end of a typical tweeter sets how low you can cross it over.    Why is that important?   Good question, there are plenty of woofers/midwoofer that will play high.   The problem is, that higher crossover points leads to not only a narrowing dispersion pattern of the midwoofer (common knowledge) but it leads to larger off-axis nulls on the vertical axis compared to crossing over lower and keeping driver-driver spacing narrow.   Horizontally, you can usually hide these aberrations and that is the plane that is most important.   Vertical performance is part of the power response though and it affects the on-axis perceived sound quality of the system.    The lower you can crossover and the closer spacing you can use, the better the response.
  
•    Time Alignment: To be honest, I've never considered it a huge factor.  I've done some testing to see how audible the couple ms of delay is compared to perfectly time aligned components and my usual response is that there are larger problems to be concerned with and the tradeoffs to achieve this are not without their own cost.   A sloped baffle to align the acoustic centers has you using the driver’s off-axis, and complicates optimization of the response curve.   In active setups you can use delay, but that is another can of worms because most audiophiles won't go for active designs and the analog methods necessitate use of extra components.    One of the advantages of the wave-guide though, is that we get pretty darn close to a perfect alignment without any downside.   While I don't consider it critical, it is a freebie benefit.   
• Cosmetics:   Lets face it, we are all vain.   I like a well made part that looks cool much better than a cheap piece of plastic from the Dollar Store.   It also shapes what we perceive in terms of subjective sound quality.   As much as I don't like that fact, people's judgment of sound quality is shaped by how something is made and looks.   Plus, if I was going to spend this much money (I have a boat-load dumped into this project) I wanted something to be proud of.   I threw price out the door as a consideration.   Price just didn't matter, quality did.   
  

The first two items where really my guiding principles.   The on/off-axis response is hugely important.   Not just low THD but low high-order distortion products are critical to sound quality.  It isn't just the total THD, but the spectrum.    I also wanted something that could crossover low.  Part of the reason the Kepler sounds good is because of the quality of the tweeter and the fact that it can cross so low, minimizing off-axis response issues.   Having a device that could crossover lower is very attractive to me.   

I am not dogmatic about design principles.   I used a horn/wave guide for reasons other than are currently popular.   I looked at the horn as a way to achieve some extra low-end distortion headroom, allowing me to use the device lower than I could otherwise.   My goals for off-axis performance of a finished system are that they are smooth, without large changes in the +/- 45 deg. horizontal window of the listening axis.   I'm aiming for a response window that is constant over that range without large nulls.

Anyway, I finally have samples that look to be 99% of the way there.   They are based upon a traditional 26mm (1") dome format with the following "extras".

• Beryllium/Copper Matrix dome.   Beryllium is all the rage for good reasons, it is light and stiff and you typically get a measurable improvement in distortion.  They are so light and stiff that the dome stays pistonic at higher frequencies than most other materials. Most Be domes are made on an aluminum/Be matrix with only about 1% Be content.   Our dome is a CuBe with about 97% copper, 3% Be.   There isn't any 100% Be domes.  Our CuBe matrix is better than a standard BeAl dome for a couple reasons.  It has better damping properties that results in less break-up.   The break-up of most Be domes is way above 20K but high Q break-up has some relation to distortion products.   If you excite a break-up, the distortion products are excited to some degree also.   Having well behaved break-up, is always better than having high Q resonances.   
• XBL^2 motor technology.   Typically, people think of this as a method for increasing stroke.   It allows us to make the BL field of a transducer more linear with stroke, which tends to have a big impact on woofers and midwoofers.   Tweeters don't see large excursions but they do benefit from a linear BL field so the approach is not without merit, especially when you are pushing the envelop for a device lower in frequency.   The other feature of XBL^2 designs that is beneficial, is low native inductance.   That IS important to tweeter performance and there really was no downside to using an XBL^2 motor, only an upside in terms of lower distortion at the bottom-end and lower native inductance.
  
• Suspension:   The suspension termination of tweeters often has profound influence over the response.  This is always a balancing act and there are some real tweaker engineers who focus on soft-part development.  It is part materials science, part acoustics and partially an art of trail-error and experience.    With a high-stroke tweeter we had to have a suspension that not only provided good termination for FR issues, we had to have plenty of travel (at least from a tweeter perspective).   
• Horn Design:   The primary maxim here was that of the medical profession, "first, do no harm".   That meant no response issues or stored energy issues.    The reason for the horn, was to gain more headroom on the bottom-end and have the benefit of some directivity control elsewhere.   We achieve that and kept the form-factor in a usable range (5" overall).
  
• Phase Plug:  That thing in the middle is there for more than just keeping your kids fingers off the dome (although that is nice also).   It provides some shaping of the response that was impossible to achieve without it.   Some people might worry about early diffraction, it is close enough that it's in the near-field pressure range of the device so its affects are all on the initial wave front.   Speaking of diffraction, this design also removes any of the early baffle related diffraction issues you typically see on-axis with a traditional tweeter.   
  

This is getting long.   Once I officially finalize the design I'll post some measurements of our final design.   For now, I figured I'd get a post started on the guiding principles and I'll post a PDF showing the horn design.   It is  pretty cool, you can download the PDF and rotate the part in 3-D.    It gives you a general idea of how the device will appear cosmetically.    These are being tooled in aluminum, both the horn and the rear-cup of the tweeter.    A couple interesting things to note, the mounting to the baffle is a traditional front-mount arrangement.   It is an overall 5.25" device.   The relief around the perimeter on the rear is for a rubber o-ring that seals the cabinet.   This is pretty beefy, about 0.25" thick and the rear cup (the rest of the tweeter) attaches via adhesives.  There is a critical dimension setting the distance from the dome, to the phase plug so that has to be carefully controlled in production.   

I've not shown the rear cup, which sets the rear chamber dimensions and we have a couple other cool design features that I'll show in the final design.   Overall response is currently +/- 2dB from 1K-22K and the device is easily usable down to that lower number.   We designed it such that the lower-end doesn't have a large frequency response peak in the horn's bandwidth.   It is flat, without crossover components across its bandwidth.    The device ends up being around 93dB/1W/1M, which was set by the top-end where there is no horn loading.   

That is it for now.  I hope you enjoyed my rambling.     


 www.diycable.com/main/pdf/ExodusWG.pdf

[sbrtoy]

Kevin,

Sounds like a very useful driver that fills a niche with few options now...hope it does well for you.

Sam

[Vapor Audio]

Hey Kevin, is this tweeter something you'll offer as a raw component or only in finished designs? 

[Bemopti123]

Just reading it makes me think you have something or a world class beater here.  I am sort of afraid to find out what it will cost, but seeing your offerings, cost no object to my mind would be something mildly expensive and an outright bargain in comparison to what the big boys offer.  Cheers.

[Kevin Haskins]

Kevin,

Sounds like a very useful driver that fills a niche with few options now...hope it does well for you.

Sam

Thank you Sam.   I hope it does too.   

There are many drivers out on the market.   Not any 1" format tweeters with this kind of usable bandwidth.   You need to move to a compression driver on a deeper horn (narrower bandwidth & more horn related problems) to get distortion levels and headroom comparable or better to what this will deliver.    Nothing is perfect for every application but I really aimed for a device that fits MY applications related to 2-channel home audio & home theater applications.   

[Kevin Haskins]

Hey Kevin, is this tweeter something you'll offer as a raw component or only in finished designs? 

My focus is using it in my completed speaker designs.   I picked design principles to fit my needs for products that I'm developing.  (Notice a lot of personal pronouns in there)   

I'll offer the raw drivers but that wasn't really the focus.   

[Kevin Haskins]

Just reading it makes me think you have something or a world class beater here.  I am sort of afraid to find out what it will cost, but seeing your offerings, cost no object to my mind would be something mildly expensive and an outright bargain in comparison to what the big boys offer.  Cheers.

If you consider the things that we measure, distortion curve, bandwidth & on/off-axis performance it should be top-drawer.   I'm sure there will be devices that can match or beat it in certain categories, but not all of them.    Also, different engineers prioritize different attributes.   Some will say, distortion doesn't matter, or horns sound bad, or dome tweeters suck, (I'm being a little sarcastic), or pattern control isn't as good as this other device.   You can never please everyone but I just focused upon things I wanted in the device and that is what drove the final design.     I know what things I prioritize and why.   Other people are more difficult to figure out.

By expensive I'm estimating a retail price of around $175-$200 per tweeter.   I consider that expensive and they could be much cheaper if they sold in Walmart in the millions.   That isn't the case though, we have to amortize the tooling/development cost over fewer units.   

[dyohn]

Excellent write up and that tweet seems to be a genuinely useful and unique design.  I can't wait to see your finished measurements and to place my order for a pair!

BTW, my company uses springs made from Cu-Be material.  It has outstanding shape retention and self-recovery properties, which in a tweeter dome might help it resist physical distortion like wrinkling or dimpling, or if it does it should return to its primary shape very naturally.

[Vapor Audio]

My focus is using it in my completed speaker designs.   I picked design principles to fit my needs for products that I'm developing.  (Notice a lot of personal pronouns in there)   

I'll offer the raw drivers but that wasn't really the focus.   

Well, your tweeter might have a larger potential customer base than you think.  I've just started working on a tweeter myself, one that in many ways would be identical to your - Integrated waveguide around 6" total diameter and depth to time align with 7" woofers, AlBeMet 140 dome, long throw and usable bandwidth close to 1000hz.

So if you're going to offer them for sale, I think you can count me in for a stack of 'em.  And I bet you'd be surprised at how many DIY'ers are wanting a tweeter exactly like you describe! 

Best of luck with it, I hope performance meets your targets.

[jr1414]

Kevin, glad to hear you're at it again.  I'd love to hear if I could do a re-cab and new crossovers for my Keplers with these tweeters.  Is that something you'd consider engineering?

[Kevin Haskins]

Excellent write up and that tweet seems to be a genuinely useful and unique design.  I can't wait to see your finished measurements and to place my order for a pair!

BTW, my company uses springs made from Cu-Be material.  It has outstanding shape retention and self-recovery properties, which in a tweeter dome might help it resist physical distortion like wrinkling or dimpling, or if it does it should return to its primary shape very naturally.

I've not tried the kid test on them yet.   

[sbrtoy]

Excellent write up and that tweet seems to be a genuinely useful and unique design.  I can't wait to see your finished measurements and to place my order for a pair!

BTW, my company uses springs made from Cu-Be material.  It has outstanding shape retention and self-recovery properties, which in a tweeter dome might help it resist physical distortion like wrinkling or dimpling, or if it does it should return to its primary shape very naturally.

I've not tried the kid test on them yet.   

I hate hexagrids but they can save some $ in the long run!

[Kevin Haskins]

Kevin, glad to hear you're at it again.  I'd love to hear if I could do a re-cab and new crossovers for my Keplers with these tweeters.  Is that something you'd consider engineering?

I'll probably not re-design the Kepler.   What I will offer is large discounts for current customers who want to upgrade to the new speakers.  I'd offer something like 50% discounts for Kepler customers, which would be more cost effective than trying to upgrade the existing cabinet.   

[Kevin Haskins]

Excellent write up and that tweet seems to be a genuinely useful and unique design.  I can't wait to see your finished measurements and to place my order for a pair!

BTW, my company uses springs made from Cu-Be material.  It has outstanding shape retention and self-recovery properties, which in a tweeter dome might help it resist physical distortion like wrinkling or dimpling, or if it does it should return to its primary shape very naturally.

I've not tried the kid test on them yet.   

I hate hexagrids but they can save some $ in the long run!

That phase plug is good for any big fingers.   A kid with a pencil, Tinker Toy or just little fingers could probably still poke them.   I'm working on a nice integrated speaker cloth grill that doesn't cause the typical problems of standard grills.   Something with a stretchable draw-string that fits in a groove around the perimeter.   For people with kids, it may be smart to cover them but they are really fairly well protected with the phase plug and if I can get the grill done right, it should really cause virtually no acoustic problems for general use. 

[JoshK]

Look really cool, Kevin. 

I imagine that there is a market for tweeters like this even though your intent was finished product.  If they hold their own about stuff like the aircirc, which they should given the physical benefits, then they don't seem so bad price-wise.

[jr1414]

Kevin,

Sounds good enough for me.  Although getting there was half the fun with the Keplers!  Keep me posted.

[TerryO]

Excellent write up and that tweet seems to be a genuinely useful and unique design.  I can't wait to see your finished measurements and to place my order for a pair!

BTW, my company uses springs made from Cu-Be material.  It has outstanding shape retention and self-recovery properties, which in a tweeter dome might help it resist physical distortion like wrinkling or dimpling, or if it does it should return to its primary shape very naturally.

I've not tried the kid test on them yet.   

I hate hexagrids but they can save some $ in the long run!

That phase plug is good for any big fingers.   A kid with a pencil, Tinker Toy or just little fingers could probably still poke them.   I'm working on a nice integrated speaker cloth grill that doesn't cause the typical problems of standard grills.   Something with a stretchable draw-string that fits in a groove around the perimeter.   For people with kids, it may be smart to cover them but they are really fairly well protected with the phase plug and if I can get the grill done right, it should really cause virtually no acoustic problems for general use. 

Kevin,
If you field test them with your kids, they should be just about safe from anything short of a Thermonuclear Event.

Before anyone thinks that I'm slamming Kevin's children, I'll say that they're all very nice, and as well behaved as you can expect from any  mentally and physically active children. Now, that seems to be an apt description of their old man as well...Hmmm

[Edit: The more I thought about it, I'd have to say that the kids are *MUCH* better behaved than their Dad ]

Oh BTW: nice tweeter Kevin, although I can't see how they could improve on the sound of the Keplers, you've proved me wrong before.

Best Regards,
TerryO

[Kevin Haskins]

Excellent write up and that tweet seems to be a genuinely useful and unique design.  I can't wait to see your finished measurements and to place my order for a pair!

BTW, my company uses springs made from Cu-Be material.  It has outstanding shape retention and self-recovery properties, which in a tweeter dome might help it resist physical distortion like wrinkling or dimpling, or if it does it should return to its primary shape very naturally.

I've not tried the kid test on them yet.   

I hate hexagrids but they can save some $ in the long run!

That phase plug is good for any big fingers.   A kid with a pencil, Tinker Toy or just little fingers could probably still poke them.   I'm working on a nice integrated speaker cloth grill that doesn't cause the typical problems of standard grills.   Something with a stretchable draw-string that fits in a groove around the perimeter.   For people with kids, it may be smart to cover them but they are really fairly well protected with the phase plug and if I can get the grill done right, it should really cause virtually no acoustic problems for general use. 

Kevin,
If you field test them with your kids, they should be just about safe from anything short of a Thermonuclear Event.

Little Abel is my ultimate test.   I put them in a room with him, some Hot Wheels, a plastic sword and some Transformers and let Abel have his way with the speaker.   If they come out unscathed after 60 minutes, then they pass the kid test.

Before anyone thinks that I'm slamming Kevin's children, I'll say that they're all very nice, and as well behaved as you can expect from any  mentally and physically active children. Now, that seems to be an apt description of their old man as well...Hmmm

[Edit: The more I thought about it, I'd have to say that the kids are *MUCH* better behaved than their Dad ]

Oh BTW: nice tweeter Kevin, although I can't see how they could improve on the sound of the Keplers, you've proved me wrong before.

Best Regards,
TerryO

Oh come on Terry, you don't trust that better measurements equates to better sound?       The Kepler is good but nothing is perfect.   Part of the sickness of this industry is that you can never be satisfied with something.   It goes for consumers and for most of us designing stuff.   Once I get done with a project I'm scratching my head trying to figure out ways to improve it.  It is a never ending cycle.   

[TerryO]

Oh come on Terry, you don't trust that better measurements equates to better sound?       The Kepler is good but nothing is perfect.   Part of the sickness of this industry is that you can never be satisfied with something.   It goes for consumers and for most of us designing stuff.   Once I get done with a project I'm scratching my head trying to figure out ways to improve it.  It is a never ending cycle.   

Kevin,

I think you know that some measurements have meaning and others are without a lot of merit.

Amplifier THD and damping factor measurements are two that really don't have much meaning beyond a certain point. Julian Hirsch (and others) promoted standardized THD measurements years ago, as the industry used all kinds of bogus and semi-bogus numbers that told you next to nothing about the performance of the amplifiers. Way back when, in the early days of SS gear, an amp would have, for instance, "Music Power" specs. An amp would advertise 100 watts of Music Power, which might equate to 15 to 20 watts of usable, undistorted power with the rest being distortion and ultimately oscillation. "Music Power" was, IMO, the total power measured .05 nanoseconds before the caps blasted through the ceiling and the transformer became a grenade. Ok, that's perhaps a little exaggerated, but it was usually the last measurement that could be taken before the amp expired.

As time passed, THD became a standard measurement that established a data point that was reliable. Where the problem lies, is that it really is a composite figure, that tells you very little about the sound. Even order harmonics sound OK and odd order harmonics are nasty sounding, so any combination of harmonics could be totaled to equal the same THD figure. That was the point that the "underground" press harped on, that the subjective sound should be a factor in the equation.

Damping factor is another spec that has little relevance, again IMO.
Back in the early days of tubes, it wasn't uncommon to have a DF of below 20 or so. The Single Ended Triodes were even less generally. Damping Factor is, in part, a result of negative feedback, which triodes being perhaps the most linear device ever invented, didn't really need. With efficient speakers the Triodes could operate in their most linear region, which was at the low end of their output and as power increased, the levels of distortion also increased, although the 2nd Harmonic especially and even order harmonics generally, were the dominate forms of distortion.

Solid State (especially bipolar transistors) are essentially nonlinear by comparison and require copious amounts of negative feedback to correct the nonlinear behavior. The more feedback, the better the measurements of THD. Damping Factor figures also tend to rise, and it wasn't long before somebody got the idea that the numerical increase in DF specs was a good thing, allowing the amp to "control the woofer with an iron hand." The fact that an SE Triode, played within it's rated power, could achieve greater bass extension than SS at the same level, tended to be overlooked. The introduction of low efficiency speakers (AR 3a, as an example) pretty much eliminated the SET as a viable choice, as they ran out of juice just at the point that bass from these speakers needed even more to have any real "grunt."

You can check this out for yourself by using a power resistor in series between the amplifier and your speaker. The result is that the Damping Factor as seen at the speaker is severely reduced and you can often achieve an extra half octave of bass extension.

Three years ago I built a design that a friend (a Boeing Engineer) had come up with, for an event in Canada. It featured a second order series crossover that had an "optional" resistor. As I was going to an event that is nearly all tubes, and thinking that the option was for use with a SS amp, I left it out. The various posts resulting from the meet were "very" complimentary towards the speaker, and one post in particular mentioned that I had left a resistor out of the crossover. My friend saw that post and emailed me to find out what in the world I had done with his design. When I mentioned that I had left out the optional resistor for SS amplifiers, he was amazed. He stated that he had intended the "optional" resistor to pad down the tweeter a bit, if it was necessary. Anyway, we talked about it for some time and about a week later he called to tell me that he had run a number of simulations and found that the resistor didn't actually do anything to tame the treble, but it did extend the bass shelf about a half octave with his SS amp. The subjective results of this change of balance would actually sound as if the treble was slightly reduced.

Another rumor in the Myth of Damping Factor is the often cited "fact" that tubes (SETs in particular) have muddy bass, while SS with their high DF are able to produce solid, clean bass. This is partly true in many instances if the impedance curve isn't fairly flat. If the curve "is" (hypothetically) flat, then the superiority of the SS amp's DF is rendered moot and, in fact, the lower DF SET amp may very well be able to produce more extended and powerful bass, within it's specs.

This has actually turned out much longer than I had intended, but I can certainly see the advantage of an XBL2 driver, with it's controlled impedance curve. Mated to a tube amp, especially a SET (if the driver is efficient enough to offset the generally low output of most triods) it might end up being the speaker designer's delight.

Best Regards,
TerryO

[Kevin Haskins]

Wow... you must have spent a lot of time pecking at the keyboard with that response!   

For those who don't know, Terry is one of my local buddies and I happen to know that his typing skills are limited to "hunt & peck".

Measurements matter but knowing which ones and giving them proper weighting is the key.   Like anything, you have to know what is important and what isn't.    Distortion @ the lower crossover point of a tweeter is important because it determines where you can use it.    Pretty much everyone, without exception, considers the total frequency response on/off-axis important.   

In terms of this device, the measurements that I'm mainly talking about are the distortion performance, with a focus on the low-frequency side of the device.   That and the fact that using a lower crossover point makes lobing issues on the vertical axis less of an issue.   We should also get very smooth off-axis behavior on the horizontal axis.    All of those features are measurement based and do not allow for subjective differences.    How do they translate to subjective preference?   That is a difficult question because the tweeter is just one ingredient in the dish.   But, if you improve your ingredients typically the dish improves with it.