The Jenzen speakers
Copyright 2012-15 © Troels Gravesen Gravesen

Why do the Jenzen speakers have a stepped front panel? Why do all feature 2nd order Linkwitz-Riley filters and why do they have the particular front panel designs? Below you will find some of the rationales behind this speaker range.

The Jenzen series of speakers are meant to offer high-end sound for low-end price, even for the constructions with expensive drivers. As discussed in Speakers' Corner; what diy people can do speaker manufacturers cannot, is this: Spoil our drivers with the best of cabinet work. MDF is cheap and it's no problem making 30-50 mm panels and adding solid bracing to produce low-resonance enclosure that is often a significant part of high-priced speakers. Cabinet resonances add more to the sound than we may think.
We may think commercial speakers are overprized - and some are - but making a living from manufacturing speakers is not an easy task despite having drivers and crossover components at less than half price compared to what we pay retail. As diy'ers we only have ourselves on the payroll, no packing and shipping, no dealers' profit, no advertising, etc. If production price is 10-20% of retail price we may have a viable business, hence the price of a single super cap or added bracing may be omitted to stay competitive. As diy'ers we don't have to take these precautions.

THE CROSSOVER
Experiments with different crossover designs led to a closer investigation of the various topologies' impact on sound, in particular the ability to reproduce vocals, and some of the results can be found here: http://www.troelsgravesen.dk/vocals.htm.
1st order filters and time-aligned drivers can sound wonderful - and it did - but price to pay is limited power handling. 1st order 2-ways soon sound harsh as we crank up volume, i.e. the tweeter really can't handle playing some midrange and increasing the point of crossover may get the middriver yelling due to beaming. Usually crossovers get extremely complex if we want a middriver doing a 1st order roll-off some two octaves above point of crossover. Some manufacturers may claim 1st order filters and they may be so - on the electrical side - but acoustically we often find the drivers rolling off by 2nd or 3rd order slopes. Think of a 1" dome crossed at 2 kHz/1st order. It would have to do -12 dB at 500 Hz. It can't. 1st order filters apparently have some magic to certain consumers and we never really see measurements of these construction based on a "coil and a cap". 

The question is how wide the overlap has to be, and from my experience some two octaves below and above point of crossover. As soon as we are some 12+ dB below summed response we can have the drivers' roll-off as steep as it may be. To realise a 1st order filter I would pick at least four drivers. And again, I have to point to Vandersteen, who probably do the right thing in making an extremely broad-banded 4" upper-mid driver providing a huge overlap to lower-mid (8" driver) and tweeter. Finding a 3-4" driver that can do this is actually not that difficult, e.g. the ScanSpeak Illuminator 12MU is an example.

Now, having four drivers all providing 1st order roll-off two octaves above and below points of crossover - and all being connected with positive polarity - takes serious equalisation and the crossover becomes complex due to e.g. all-pass filters - and a lot of energy is getting lost. The result is an appallingly low efficiency. Vandersteen's top model doesn't make more than 83.5 dB/2.8V and needs a lot of power to come alive, thus a 400 subwoofer in-built subwoofer. It may sound wonderful - I never heard it - but I guess this is as far away from a 100dB 15" driver in an e.g. 220 liter Onken cabinet - or open baffle - as can be.
We rarely discuss the trade-offs of loudspeaker design as people soon find a religion and stick to it as a matter of specific performance preferences. Some build humongous 4-5 way horns and surely they'll deliver the dynamic aspects of music like no other system can do. The average living room does not allow such systems and neither does the average consumer want such a system. We tend to settle with 10-50 liters, 85-88 dB sensitivity and trade in dynamics and maybe low-end extension.

Another thing key issue with 1st order crossover is the endless search for drivers that will be make a good blend of sound as the drivers over a wide frequency range will do the same thing - but their way. 2 kHz from a 1" dome and a 4" mid-driver just does not sound the same.
The classic 3rd order Butterworth filter is out of the question as I just don't think it sounds good. I like the 4th order (LR4) filter with its dry monitor-type of sound due to the lack of driver response overlap and usually even power response, which usually can be tuned to a very precise target frequency response due to the large number of components. However, the LR4 does not - to my ears - deliver the life-like presentation of overtones compared to 1st and 2nd order filters, but I surely value the presentation of a well-constructed LR4 filter for certain applications.

This leaves the 2nd order filter (LR2) as the best compromise when it comes to sonics and power handling. Only thing is that we need to time-align the mid and tweeter to make a text-book LR2 filter work properly. Sometimes we can get away with a tilted front panel like the Illimina 66 and simple crossover tricks to make a flat frequency response and a decent phase integration - or use an all-pass filter to delay tweeter response. The latter makes a rather complex crossover although it can work very well with proper capacitors. Proper phase integration is often grossly neglected in commercial designs and the question is: How good does phase integration really has to be - to be good enough? Because it can never be perfect as phase integration depends on where we position ourselves with regard to the speakers. Common practice in testing good phase integration is to set up the microphone between the two drivers and reverse polarity of one of the drivers. This should produce a significant dip in frequency response from a well constructed crossover. As we could see from the Chario Sonnet speaker, phase integration was anything but good, but yet it has a pleasant sound. Chario here claims that phase behavior of the two drivers does not have to be coincident, rather parallel over a wide frequency range and that placing the point of crossover in the 1-2 kHz range, where our ability to locate sound is poor, does the trick. Right or wrong, this speaker performs very well indeed. To my experience two drivers displaying poor phase integration - but having parallel phase slopes - can sound very good indeed, but the the more out of phase the two drivers are, the more we create some "phasiness" in the overall sound, which may be perceived as enhanced three-dimensionality, sense of depth. Cheap trick? 

Placing a point of crossover at 200 Hz where we have a wavelength of 34400/200 = 172 cm, does not necessitate time-aligning drivers. It will have no significant impact on sound and as long amplitude is fairly flat, we're well off.
2-3 kHz is another matter. Wavelength at 2.5 kHz is 34400/2500 = 14 cm. Placing a middriver and tweeter on a plane baffle may mean the tweeter is some 25-35 mm in front of the middriver based on the "acoustic depth" of the middriver. This may mean some 90^o phase difference and this is significant when it comes to implementing 2nd order crossovers. To overcome this phase difference we usually do two things: We tune our (2nd order) crossover to flat response and disregard the phase error, or - we use 2nd order filter to the middriver and 3rd order filter for the tweeter - or visa versa. This way we can usually manage flat amplitude and proper phase integration. The latter I have used numerous times and it's not bad at all, but - it can be better.

To find out the acoustic delay of the drivers we can do this: We place the microphone at 1 meter distance between mid and tweeter and record amplitude and minimum phase for both drivers and finally the summed response (impedance data not needed here). It is important the microphone is not moved during these measurements. Next we take the data files to the LspCAD6 software and simulate a 3-way crossover with tweeter, mid and summed response as the three drivers. The measured summed response is disabled in the SPL/Xfer function, thus does not add to the total summed response. This will produce 4 graphs, tweeter, mid, mid+tweeter and summed response of mid+tweeter. Next we change dZ for middriver until summed response coincides with mid+tweeter response at suggested point of crossover, because what we also find is that the dZ is not consistent with frequency. Usually a middriver will change its phase behavior because not all parts of the cone may produce the same frequencies. At higher frequencies it may only be the center of the speaker, e.g. dust-cap, that does what it should do.
Once done we can start simulating our crossover with the correct dZ. Now, this applies to a single spot with regard to height and distance to speaker and it can be useful to repeat the session with the microphone in different vertical positions to see how it impacts dZ. This goes hand in hand with the set-up of test crossovers to see how reality fits measurements and simulation and new measurements may be needed from time to time. All in all a rather tedious procedure that has to be evaluated sonically with various points of crossover. Sometimes a less than "perfect" crossover may give the best results sonically regardless of poor correlation to the perfectly simulated crossover.

We will find dZ for most 6" drivers to be around 25-35 mm relative to a standard 1" dome. With the middriver on top and suggested listening position at tweeter height the middriver will be a little further away from the listening point and we may get away with the middriver being some 18-25 mm in front of the tweeter as can be seen from the various Jenzen mid-cabs. The ScanSpeak Illuminator is a particular "deep" driver, thus some 27-29 mm is recommended here. Plus/minus 1-2 mm doesn't change a thing so don't take this too literally. There are a lot of things to dynamic drivers that aren't perfect so this not is the place to be nitpicking although high-end manufacturers will tell you another story.

THE BAFFLE
The response from a given driver highly depends on the baffle on which is it mounted. Some diy retailers gladly simulate crossovers from infinite baffle measurements found in driver manufacturers' data sheets. Forget it, it does not work.
Numerous experiments were done to optimise front panel geometry before launching the first Jenzen speaker and you can find some measurements on mid-drivers here. A little to my surprise the same basic baffle made all tested speakers perform very well. And I was really surprised to find the ScanSpeak 18WU performing exceptionally well compared to the infinite baffle response found at ScanSpeak's data sheet:

  
Left: 18WU/8741T00 on Jenzen baffle. Right: Same driver from ScanSpeak data sheet (infinite baffle).
I know others have looked into this strange 4-6 kHz peak by removing the dust-cap but without resolving the issue.

SUMMARY:
Based on sonics, the LR2 filter has been chosen for these speakers to avoid the large phase shifts of 3rd and 4th order filters and at the same time provide decent power handling. 
To summarise correctly the mid and tweeter need time-alignment, thus the stepped baffle. For bass-mid this is not important due to wavelength at 200 Hz.
The transmission line cabinet is another key ingredient in improving sound although the bass driver is the driver benefiting the most from this. I truly believe the bass quality is superior to the traditional vented solution.
Size matters and membrane area is a key parametre in getting distortion down as membranes shouldn't move much. The Jenzen speakers are not particularly small, but even low-cost drivers like those used in Jenzen SEAS ER will from its hard-cone mid-driver deliver great detail and transparency. A pair of Jenzens will set you back some 1,000 - 6,500 USD + materials for cabinets depending on design and quality of crossover components and as always we're into the area of diminishing return for our investment the higher we go.

Jenzen SEAS ER, SEAS CA25RFX, CA18RNY and T25C003/T25CF001

Jenzen NEXT, SEAS W26FX002, W18NX001 and T25C003/T29CF002

Jenzen Accu, Audio Technology 10C77-25-10-KAP, Accuton C173-6-191E and Accuton C30-06-024

Jenzen D mkII, Audio Technology 10C77-25-10-KAP, Accuton C173-6-191E and Jantzen Audio diamond dome JDT1024

Jenzen Illuminator, ScanSpeak W26/8861T00, 18WU/8741-T00 and D3004/660000

Also check out how a range of 6" drivers perform on the Jenzen front panel.