[Kevin Haskins]

Just wanted to post some quick design notes on the new loudspeakers.   I'm naming the next generation of loudspeakers after historically important scientist.   Its a departure from our old naming scheme.   

The Kepler is the first design.  Its a traditional two-way design that fits a lot of applications.    I have done design work around three of the Parts Express finished cabinets.   The smallest design is a 0.38ft^3 which will be a sealed design.   The 0.5ft^3 & 0.75ft^3 enclosures are both ported designs.    This gives us a nice range of cabinet sizes & levels of low-freqency extension/output.   

My design priorities are fairly simple and there is nothing earth shattering in my methodology.   I pick transducers based upon several properties, linear/non-linear performance, output, suitability of T/S parameters for desired enclosure size and dependability of the vendor and production Q/C.    For all Exodus designs I've removed cost as a design parameter.   It isn't that we have infinite resources, we don't, but when deciding on how to build a driver we optimized them based upon performance first and let the final cost fall where it may. 

I'm lucky in that I have the resources and expertise to draw upon to design my own transducers when needed.   We designed our own 6.5" format driver, the EX-6.5.   Not that there is a shortage of 6.5" off-the-shelf drivers.   There are plenty of good units on the market but we wanted to implement the XBL^2 motor technology, I wanted something that had class leading performance in terms of non-linear distortion, particularly odd order products and I needed something that would work across all my intended applications. 

I don't have the time/space in this area to completely describe our design process for transducers.   We use Klippel measurements to look at the fundamental causes of distortion.   We use FEA tools and Maxwell to design motors.   We optimize for flat BL curves, low inductance and symmetric Cms curves (suspension compliance) while trying to keep our original design parameters in place.   Its always a balancing act and there is no such thing as perfection.   You have to keep in mind the tradeoffs your making for any given area of performance.   I heavily weigh non-linear distortion since its fundamental to how the transducer sounds in a finished design and you cannot "fix" it with crossover design.    In terms of non-linear distortion I will take higher even order products in order to reduce any odd-order products.    Its a simple choice, even order products are less offensive.   

In terms of final system design I look for the following:

• Smooth off-axis behavior.  I tend to design over a 35 degree horizontal axis rather than a single on-axis measurement.
• I design to limit the use of drivers to ranges where they are happy.   I use expensive drivers so that give you a lot more flexibility than using budget drivers.   Most high-end tweeters play lower with less distortion (generalization).   This gives you more flexibility in choosing how to use them based upon the acoustical measurements.
• I don't have a philosophical axe to grind concerning what crossover is best.   I use whatever the measurements suggest will work well.   I also listen to several design choices when doing a crossover.   Those include low & high order acoustical solutions.    I listen to them blind so that I don't favor one vs. the other based upon what I think should sound best.   
•   I listen in several different acoustic environments (rooms).   The largest variable I find in my loudspeaker designs is almost always the room.   How you voice a loudspeaker (by ear) is immensely influenced by the room your using.   You have to be careful not to optimize around one room configuration.   
  

Anyway... the Exodus Kepler is our first MT design with the new Exodus EX-6.5.   It features LOTS of linear excursion.   It has about twice the available output down low compared with even the top-tier 6.5" drivers on the market.    This is important for several reasons.   First, it allows us to get more output, at a given distortion figure in a smaller cabinet.   You can always double up drivers to lower non-linear distortion, but that choice is not free.   It increases the size of your final design and it introduces some polar response problems that can be problematic (when using multiple drivers vs. a single).   Everything is a tradeoff and the money spent on using the best possible transducer is a good one.   

The tweeter is one made for us by Tymphany.   Tymphany owns and produces ScanSpeak, Peerless & Vifa.   They specialize in high-end audio transducers and have a good reputation for design, top-notch Q/C of the production parts and they are a reliable long-term provider of transducers.    Our tweeter is one of the best I've ever measured in terms of low non-linear distortion and it crosses over low with ease.   Its better than the Usher we used in the first generation Exodus kits in pretty much all measured parameters.   Its also much more sensitive than the Usher, important because the EX-6.5 is an honest 88-89db part and dual woofer designs can benefit from a more sensitive tweeter. 

When doing crossover design I measure frequency response/phase at several different angles off-axis.   Horizontal off-axis is more important than vertical.   I limit my analysis of vertical off-axis behavior to +/-20 degrees vs. 60 degrees off-axis for the horizontal plane.   I don't simulate off-axis curves.   I measure them because there isn't an accurate way of simulating the response.   Designing the crossover is a time consuming exercise in optimizing one axis and then plugging the values into the off-axis simulation to see how it looks in the off-axis measurement.   In choosing final designs I simulate voicecoil heating and component tolerances.   I use LSPCAD pro which can simulate both.   I then choose a couple solutions that I can start the listening test.   

A couple general things... I design for about 3db of baffle step correction.   My measurements are done outside with the loudspeakers elevated about 8ft from the ground so I get pretty good measurements down to about 200Hz.   The baffle step occurs over a range of about 500Hz-1K, depending upon the physical size of the baffle.   In typical listening positions you get some re-enforcement of frequencies within about a 1/4 wavelength of a boundary.   Even if you pull your loudspeakers out into the room for listening you get re-enforcement at baffle step frequencies so I find 6db correction gives too much re-enforcement from 100-500hz.  Male vocals are a good choice when evaluating this range.   They sound too "chesty" and full if you dial-in too much baffle step compensation, too thin if you dial in too little.   You get some general room gain below 100Hz.   This varies a lot with the room.  If you listen in a small basement (concrete walls) you get a LOT more re-enforcement over what you would get in an upstairs stick-built house (2'x4' construction with 5/8" drywall).    Its hard to dial the low-frequency performance in with the system design.   The crossover typically doesn't give us the ability to easily adjust LF performance (passive crossovers at least) so your limited to your choice of either box tuning (in ported designs) or system Q with a sealed design.    My target low frequency curve is typically around Q = 0.7 type curves.   I find that in most rooms aiming for that kind of response matches up well with room-gain curves.   

Our active designs will offer some adjustment of these areas that just are not possible to adjust with a passive crossover design, or even with an active one without measurement capabilities.   The only accurate way to adjust for LF room issues is with a measurement of the system (loudspeaker-room interface) and adjustments provided based upon the final room measurement.   There is no way to ball-park these things with any level of accuracy.


Tomorrow I'll upload some measurements.   I'd do it now but my I don't have the password for my ftp address on this computer.       Thanks for reading my long-winded post.   

[klh]

Kevin...

It's very interesting and exciting to read about how and why a designer of transducers and loudspeakers makes critical decisions in the developmental process. Thank you for including all of us.

[Kevin Haskins]

That was one long post.   I'm amazed anyone would take the time to read it.   

Here are some measurement/simulations.   This is all raw data that is imported into LSPCad for simulation of the crossover.   I find that the simulation is typically VERY close to the actual measured result.    The off-axis simulation is done with off-axis measurements.   None of these are smoothed, they represent the raw data and all the measurements are truncated at 200Hz due to the innaccuracy of the data below that point.   



The above is about +/- 1.5db from 500Hz-20Khz.   This is not directly on-axis.  It is about 5-10 degrees off-center.    If you smoothed the data for a more accurate representation you would get a little better idea of the audible results.   Notice about a 2db baffle step, which I may adjust after some listening test.    Notice a slight 1db plateau up above 8K which I may also adjust based upon listening test.   



The reverse null for those who desire to see it.   Crossover is currently at 1.6K and a fairly steep 4th order acoustic response.



This is about 30deg off-axis.  The reason for the low crossover isn't because the woofer needs it.  Its because the lower you can crossover to the tweeter the better your off-axis results (generalization).   You have to use a tweeter that will play low and do distortion measurements to confirm you don't have any problems.   Look at that off-axis response..... perfect.   If you look at the range from 0-45 degrees off-axis this design looks very nice.    A very good place to start.



The impedance curve represents an easy load.    No crazy phase angles and a minimum impedance of about 6 ohms, maximum of 20 ohms just around 1.1K.   

Total crossover component count is nine parts.   Four inductors, four caps and a single resistor. 

[Kevin Haskins]

While I'm at it I better show the LF simulation.   We aimed for parameters that would allow use in 0.5ft^3 - 0.75ft^3 vented enclosures.   You can use them in sealed alignments but you end up with a high f3 with this low of a Qts driver.    Since we have some active EQ coming down the road I optimized the unassisted parameters for the vented alignment figuring I could always EQ whatever I wanted for sealed alignments.

Here we are in the 0.5ft^3 Parts Express box tuned to about 33Hz.   Keep in mind this is an anechoic 2Pi simulation.  You can pretty much count on at least another 10-12Hz of extension with room reenforcement.   



For a frame of reference, here is the Extremis in the same box & tuning.   This is a good comparison because both drivers are roughly designed for the same enclosure sizes, with the Extremis favoring a slightly larger enclosure.    The Extremis is the low frequency king in terms of 6.5" midbass drivers.   I don't know of another that goes deeper and is still usable as a midbass.   Notice the obvious tradeoff, lower sensitivity.   The Extremis is a good 3dB down from the EX-6.5.   After baffle step compensation you end up around 82db in a finished design.   Notice the f3 in the same box.   The Extremis is at about 43Hz and the EX-6.5 around 47Hz.    You pay for that extra 5 Hz of extension with the sensitivity difference. 

[Voodoo Rufus]

I read the entire first post as well.

Looks very nice so far. I can't wait to see your 2.5 way design results.

I better get a house so I can do some box building again. 

[kfr01]

Kevin - this sounds very exciting.  I like your idea of using the Parts Express cabinet sizes, if possible.  I don't think my wife could handle another 2641-consuming-the-garage project, so routing a couple holes and soldering a bit is ideal! 

[Kevin Haskins]

Yeah.... the PE cabinets are nice too.   

The tower 2.5-way is using the Madisound pre-made cabinet, which is also very good quality.   

I'd be working on the 2.5 way now but I'm doing SMD work on a bunch of circuit boards that where assembled wrong.        

[klh]

Any progress?

[Kevin Haskins]

Any progress?

Yes.

[JLM]

I love your design philosophies, especially, "Everything is a tradeoff and the money spent on using the best possible transducer is a good one."  If it could be researched, I'm sure we'd find audiphiles spending more on speaker finishes than the drivers that go in them.  The open baffle and single driver "nuts" would be the biggest exceptions to this premise.  My single drivers cost nearly 50% of the cost of the finished speakers.  With single driver designs there's no place for a bad driver to hide.

Your comments about the room effects makes me wonder if you would include some room recommendations for the Kepler cabinet variations when it comes time to market them.

I look very much forward to your active designs.  (Now you're cooking!)

[Kevin Haskins]

I'm really still voicing the crossover.    I want to get some feedback from a few outside sources whom I trust and then I can move on with sourcing.   I'm also concurrently working on the 2.5 Way tower, an MTM (possible center channel) and in-wall designs. 

Once I finalize a crossover its still a 4-6 week time for crossovers to be built, assembled and delivered.   We are probably looking at a October release date on the passive versions.

The active version will use our Exodus Amplifier as a basis for the electronics.   It will probably be a 4-channel UcD180HG build with the active crossover built into the Hypex based amplifier chassis.     Control over the EQ section will be via a USB cable and you will need a computer to interface with it for room measurements, & PEQ settings. 

I'm projecting that will be another 6 months after the passive version is released.   The passive version is designed to easily upgrade to the active so there is an upgrade path.  Features the active version will offer that are not available via the passive (or 99.99% of other loudspeakers on the market), will be extended bandwidth with a little EQ, room measurement tool for low frequency optimization with parametric equalization, and variable baffle-step compensation for different room placements of the loudspeaker. 

The active version won't be cheap, as the 4-channel amp alone is $1455 and the active circuits with software will probably add another $500 to the cost of that.    But.... it will be a kick in the pants to have that much control over the loudspeaker and room.  Should be lots of fun.   



 

[klh]

Kevin...

I presume the 4-channel UcD180HG will drive a pair of 2-way loudspeakers... is that correct? Or, are you referring to the 2.5 way tower? Still, that's a lot of power for a rather small speaker (especially the tweeter).

I don't think cost is as big of a factor if this is for a dedicated two channel rig. Most high end systems have separates, and couple that with the amount of control, I think this will sell well. Could you do a WWMT with a 1x400 driving the Ws and 2x180 driving the MT? This would also be a 2.5-way active loudspeaker. You could also do a midlevel version that is 2x180 (180 for the WW and 180 for the MT). That would be interesting to say the least .

[klh]

OBTW, having an easy path to upgrade from passive to active is very smart. It gives people a reason not to wait to get the speakers. And, once they have them, it gives them incentive to go active once available (as opposed to another retailer). It also breaks up the cost which makes the finalcial aspect easier for the customer. Very smart!

One more thing, we know which drivers you intend to use. Can you please go over which combinations you're thinking about implementing (MT, MTM, WMT, WWMT, etc)? Thanks .

[Kevin Haskins]

Kevin...

I presume the 4-channel UcD180HG will drive a pair of 2-way loudspeakers... is that correct? Or, are you referring to the 2.5 way tower? Still, that's a lot of power for a rather small speaker (especially the tweeter).

I don't think cost is as big of a factor if this is for a dedicated two channel rig. Most high end systems have separates, and couple that with the amount of control, I think this will sell well. Could you do a WWMT with a 1x400 driving the Ws and 2x180 driving the MT? This would also be a 2.5-way active loudspeaker. You could also do a midlevel version that is 2x180 (180 for the WW and 180 for the MT). That would be interesting to say the least .

The same set-up would be used on the MT, MMT 2.5 way, MTM and in-wall and yes.... the 4-channel amp would drive a pair of loudspeakers.   We can do about any combination we choose... it would just require more money and an extra amp here or there.   

I don't have a 3-way on the design list at this point.    Mainly because I don't have a midrange that is better than the EX-6.5.   I'm thinking ahead to a tweeter-midrange combo  (think of a tweeter + 3" dome midrange) for some 3-way designs.    To be honest, the smaller midrange is good for dispersion but not always better distortion wise than the 6.5" used in that capacity.    I'm crossing the current designs over low, down under 2K mainly because I have a good tweeter and that pushes the crossover down closer to where the woofer's dispersion is less narrow.     

I could do a 3-way with the EX-6.5 but I'm not sure there is much to gain. 

Designs coming out first:

#1.  Kepler:  2-way monitor both passive & active.   Based on PE 0.50 ft^3 cabinet.

#2.  Tower:   2.5 way tower based on the Madisound 38l cabinet.   Both passive & active

#3.  MTM:  2-way in the PE 1.0 ft^3 cabinet.

#4.  In-wall:   2-way with PE blank in-wall baffle.    Passive version in the works and I may do an active/passive combo that uses the passive network for the crossover and the active EQ network to perform PEQ, and room measurement functions.   

#5.   Dipole design with a pair of EX-6.5s, BG NEO3 & a pair of our new DPL-10s.   

My goal was to have a fully functional home theater or multi-channel audio solution.    We also are going to have lots of subwoofer choices and an amp with the room measurement tools and PEQ integrated.   

Should be a fairly comprehensive lineup.

[kfr01]

"The active version will use our Exodus Amplifier as a basis for the electronics.   It will probably be a 4-channel UcD180HG build with the active crossover built into the Hypex based amplifier chassis.     Control over the EQ section will be via a USB cable and you will need a computer to interface with it for room measurements, & PEQ settings. "

Kevin, this sounds quite exciting.  Will you be suggesting / selling some solution that integrates the Kepler with some of your new subwoofers?

I'm thinking this - sealed active Kepler, sealed eq'd 12" or 15" sub.  How do you recommend best integrating the two for music listening (and some HT use)?

[Dustin B]

So when do you think you'll be wanting guinnea pigs to test out the active systems and their tweaking capabilities in nice challenging 11'x20' theater rooms with 8' wide screens requiring tighter than optimal to wall placement of speakers? 

[Kevin Haskins]

"The active version will use our Exodus Amplifier as a basis for the electronics.   It will probably be a 4-channel UcD180HG build with the active crossover built into the Hypex based amplifier chassis.     Control over the EQ section will be via a USB cable and you will need a computer to interface with it for room measurements, & PEQ settings. "

Kevin, this sounds quite exciting.  Will you be suggesting / selling some solution that integrates the Kepler with some of your new subwoofers?

I'm thinking this - sealed active Kepler, sealed eq'd 12" or 15" sub.  How do you recommend best integrating the two for music listening (and some HT use)?

I probably won't have a fixed recommendation until I'm a little further along on the development.    The active EQ network allows a lot of things that you couldn't do otherwise though..... including running a high pass filter wherever you want it on the 2-ways in order to merge them with the sub. 

[Kevin Haskins]

So when do you think you'll be wanting guinnea pigs to test out the active systems and their tweaking capabilities in nice challenging 11'x20' theater rooms with 8' wide screens requiring tighter than optimal to wall placement of speakers? 

We are a little way off yet. 

[poseidonsvoice]

#5.   Dipole design with a pair of EX-6.5s, BG NEO3 & a pair of our new DPL-10s. 

Now that's interesting  aa,

Delete the DPL-10's and use my XJ12's in a U-frame instead  aa

Nice talking to you this afternoon ....Anand.