NEWS:
Following the concepts originally applied in the NaO Note; introducing a small upper midrange
coupler and a contoured baffle in an effort to extend the dipole operating range to higher
frequencies,
Siegfrield Linkwitz  announced that he would be introducing a new speaker, the
LX521, on October 28, 2012 at the Burning Amp Festival. Details of his new speaker are now  
available at his
web site. As a result of that announcement, Music and Design decided to present a
pre-release disclosure of the up coming revision to the NaO Note, the Note II RS, on October 20.
While the Linkwitz "
LX521" is very similar in its driver compliment to the revision of the Note, there
are several significant and important differences, as discussed below. As of November 28, 2012,
this page represents the final, to be released, version of the Note II RS. The Note II RS sets a new
standard of dipole speaker performance and at a significantly lower price point. The anticipated
release date is December 7, 2012.
The NaO Note                                                             Note II RS Prototype Revealed                                                 Linkwitz LX521
Introduced October, 2010                                    October 21, 2012 (11/14/2012 revision)
October 20, 2012

Last updated: 12/06/12
October 2010:

After several years in development,
the NaO Note is introduced. An
active/passive hybrid design using
dual 18cm lower midrange drivers,
a10cm upper midrange coupler,
and the Neo 3 dipole tweeter in a 3
1/2 way dipole panel with dual 10"
woofer in a quasi-cardioid, damped
U-frame woofer system. The upper
midrange coupler allows the dipole
polar response to be extended to
higher frequency than that of the
NaO II.
October 20, 2012:

The first showing of the NaO Note II RS. A fully active, 4 way system designed
around the
miniDSP 2x8 digital crossover platform (two miniDSP 2x4 units may
also be used). The dual lower midrange of the original NaO Note  has been
eliminated in favor of a single ScanSpeaker Discovery  22W/8534. The same
ScanSpeak Discovery 10F/4424G upper midrange coupler is retained while the
Neo3 tweeter is  replaced with front and (optional) rear 3/4" Vifa (or Peerless
Gold) OX20SC00 soft dome tweeter mounted in a wave guide. The design
eliminates the passive crossover and provides more uniform polar response to
even higher frequencies than the original NaO Note (see polar response plots
below). The dipole woofer system, which was an option for the original Note, is
implemented using Peerless SLS 830668 woofers to further reduce cost. The
woofer system has a 30 Hz, B4 cut off. The 4th order high pass response protects
the woofers from over excursion. Even with the use of the SLS woofers a stereo
pair is capable of producing 100 dB at 35 Hz (110dB at 50 Hz). The XXLS
835016 woofer used in the original Note or the XLS 830452 woofer may be
substituted with an increase in max SPL of 3dB, but with a disproportionate
increase in cost and little change in low frequency performance. The correct
equalization for each woofer is set in the miniDSP. The additional money for the
XXLS or XLS woofer may be better spend on a separate, powered subwoofer.

The cost breakdown for the Note II RS with SLS woofers is as follows:

Driver set:                                          $804
miniDSP 2x8 crossover kit :            $299
Plan set :                                            $  75
Total:                                               $1178

Additional items include wood for cabinet/baffle construction, wire and binding
posts, and 4 channels of amplification per speaker. Total cost to build the speaker
system with SLS woofers should be in the $1400 to $1500 range, excluding
amplifiers. Compare to the estimated cost of the LX521 at $2850 and  the hours
required to assemble and test the LX521 active crossover.   

The plan includes full construction documentation and the miniDSP configuration
files. Detailed instructions on laying out and cutting the main baffle are provided.

An advantage of the miniDSP crossover is that updates to the crossover are
easily made by loading a new configuration file. All crossover updates will be
supplied at no additional cost. Furthermore, use of the dsp crossover allows each
builder to add up to 5 channels of equalization for room specific corrections.
Additionally, the miniDSP is an investment that can be used with other systems as
well.
The Linkwitz LX521 is shown
above. Details are available since

October 25th, 2012
. Please check
the
Linkwitz web site. In summary,
the LX521 panel uses an 8" lower
midrange
passively crossover to
a  4" upper midrange using a 1st
order crossover. An analog active
crossover with equalization is used
between the woofer, the compound
midrange, and the tweeter. Front
and rear dome tweeters are used.
NaO Note II RS panel on axis frequency response is almost identical to that of the original NaO Note as
shown in the simulated response below. The principle difference, aside from the change in driver compliment, is that the
original Note panel was implemented using a complex passive crossover.  Active equalization and a 110 Hz active high pass
filter were then applied to the passively integrated panel. In contrast, the revised Note is a fully active, 4-way system. The
crossover frequencies for the revised Note remain 1k Hz and 6 K Hz. Initially the LR4 acoustic targets were retained.
However, configuration files using 1st and 2nd order crossovers between the upper and lower midrange drivers have also
been developed. The miniDSP configuration files for these variations are included with the plan set. The different versions are
discussed in the FAQ, below. The panel rolls off LR4 at 110 Hz.  It should be emphasized that the use of 4th order acoustic
slopes for the high pass filter at 110 and is not arbitrary. When a dipole source is equalized for flat response the driver's
excursion increases at a rate of 18dB/octave as the frequency decreases. Thus, to prevent the driver excursion from
continuing to increase below the crossover point, a minimum of a 3rd order high pass acoustic (2nd order electrical)
response must be achieved. The choice of a 4th or higher order slope ensures that not only does the driver excursion stop
increasing, but that it actually decreases below the crossover point. This is important to minimize excursion related nonlinear
distortion over the entire bandwidth. This same reasoning applies to the choice of a B4 alignment for the woofer system.
Measured Note II RS panel polar response with 4th order coupling crossover from
125 to 16k Hz. Plots are, left to right, 125-250 Hz, 250-500 Hz, 500-1k Hz, 1k-2k
Hz, 2k-4k Hz, 4k-8k Hz, 8k-16k Hz.
Premiminary data:
Measured data:

The Figure below shows the measured response of the Note II RS panel and a separate measurement of the SLS and XXLS woofer
systems. Panel measurement is the far field response merged with the near field, dipole corrected response below 300 Hz. The woofer
response was obtained from the woofer near field response corrected for dipole operation. Please note the 3dB/division scale in the
woofer plot.
Summary of the revised NaO Note II RS:

Format:
4-way, fully active speaker.

Woofer system: Two Peerless SLS 830668 coated paper cone woofers per speaker in a damped U-frame
                quasi-cardioid format.
                (Optionally two XXLS 835016 aluminum cone woofers may be substituted at increased cost.)

Low frequency cut off:  4th order, -3 dB at 30 Hz, -6dB at 26 Hz.

Dimensions: Height: 46", Width 12 1/2", Depth at base 17 1/2" (with grills on).

Main Panel:

Lower Midrange: One ScanSpeak Discovery 22W/8534 per speaker

Upper midrange: One ScanSpeak Discovery 10F/4424G per speaker

Tweeter: One Vifa OX20SC00 soft dome front tweeter per speaker. (A rear tweeter with wave guide option is included in the plans.)

Crossover: Woofer to lower mid: LR4 acoustic; Upper mid to tweeter,LR4 acoustic; Lower mid to upper mid coupler: User selectable,1st ,
2nd, 4th order electrical or 4th order acoustic, depending on the configuration file loaded to the miniDSP.

Recommended hardware: miniDSP 2x8,  or miniDSP 4x10 Hd.

Crossover frequencies: 110 Hz, 1k Hz, and 6k Hz

Crossover type: Various configurations supplied. Digital delay offset compensation. All drivers connected to amplifier in phase. Phase
inversion, if required, is performed in the miniDSP.

Projected system cost: A complete stereo pair of NaO Note II RS speakers should be able to be built for approximately $1400, less
                                amplifiers. Some variation should be expected depending on specific choice of materials and additional hardware.
FAQ:

Q1:  
Why did you abandon the hybrid approach of the original NaO Note?

A1:  The original NaO Note is a very high quality speaker and I really did not want to update it at all. However, the increasing prices of
drivers and passive crossover components lead me to the conclusion that it needed to be less expensive to build. One way to reduce cost
was the substitution of the single ScanSpeak 22W8534 for the dual SEAS ER18RNX lower midrange drivers. Substitution of the Vifa dome
tweeter for the Neo3 provides additional savings. Finally, going all active, digital, saves the cost the passive crossover parts and the cost,
labor and time required for building and testing an analog active crossover. There is no wondering if you wired things correctly. Additional,
with a digital crossover updates to the system can be made by simply loading a new dsp configuration file, at no additional cost. The down
side is the requirement of 2 additional amplifier channels per speaker compared to the original Note.

Q2:  You could have limited the Revised Note to 3 amplifier channels per speaker by following the hybrid approach, used with the NaO II
and the original Note, with a simple 1st order passive crossover between the upper and lower midrange drivers, similar to the Linkwitz
LX521. With a 1st order passive crossover between the two drivers certainly the cost of a single coil and cap would not be excessive? Why
did you not do that?

A2: It is true that the revised Note could have been limited to a 3-way active system by using a passive crossover between the upper and
lower midrange drivers. However, I did not feel that a simple cap-coil 1st order "coupling" crossover was the correct way to go for several
reasons, including driver excursion. Driver excursion below the crossover point of a 1st order crossover increases at 6dB/octave for the
driver connected to the high pass section. When dipole equalization is applied, the effective roll off of the high pass filter is reduced, leading
to the possibility of excursion increasing at up to 12/dB/octave. Even if the excursion remains below the Xmax and mechanical limitations of
the driver, the increased excursion below the crossover point may lead to higher HD and IM distortion in the intended pass band of the
driver. Thus, even if the hybrid approach was retained between the upper and lower midrange drivers, a higher order passive coupling
crossover would still be highly desirable.

To illustrate this I have prepared some simulations of the revised NaO Note using 1st and 4th order crossovers between the upper and
lower midrange. The first figure, below, shows the simulated axial response and polar radiation if the revised Note was built as a hybrid, 3
way system. Please note that the polar plots do not take into account that the system is a dipole. As can bee seen, both the version
implementing a 1st order coupling crossover and that implementing a 4th order coupler between the upper and lower midrange drivers have
almost identical response.
This next figure is somewhat complicated so please read carefully and refer to the figure. Recall that for an unfiltered driver the excursion
above resonance, for a constant level input voltage, increases at a rate of (1/f)^2 or 12dB/octave as the frequency decreases. Therefore, to
keep the excursion from increasing below the crossover point the transfer function of a crossover filter must roll off at a minimum of
12dB/octave. Now, the net transfer function of the filters applied to the upper midrange driver of the revised Note consist of the high pass
section of 1k Hz coupling filter; the active, 4th order, low pass filter for the upper midrange-tweeter crossover; the 120 Hz, 4th order, active
high pass filter for the woofer-midrange crossover; and equalization to correct for the dipole peak and roll off. In the figure below 5 filter
transfer functions are shown. All include the 6k Hz, 4th order low pass filter for the midrange-tweeter crossover. The middle transfer function
is this 6k Hz low pass filter in combination with a 2nd order, 1k Hz high pass filter. The roll off of this filter below 1k Hz provides a useful
reference as it represents the boundary from where the excursion will increase and where it will decrease. A transfer function which is above
this boundary will yield higher excursion. If, in addition,  the slope of that transfer function is less than 12dB/octave, excursion will continue to
rise as the frequency decreases. Please note that the transfer functions shown are for example only and do not represent the actual transfer
function using in the revised NaO Note.

There are four additional curves shown in the figure. The curve just above the reference curve, shown in blue, is a combination of the
mid-tweeter low pass filter, the 1k Hz,
1st order coupler, and the active 120 Hz, 4th order high pass filter. This would be the net transfer
function which would be applied to the upper midrange if there were no additional dipole equalization. Between 120 Hz and 1k Hz, the slope
of this transfer function approximately constant at 6dB/octave. As a result, the upper midrange driver excursion would
increase as 1/f in this
region (hilighted in light gray). Below 120 Hz the slope increases to higher order and the excursion will begin to reduce, but will remain
above the reference level of the 1k Hz, 2nd order filter.

The next curve to consider is upper must curve. This is the transfer function of the filter just discussed but with the effect of the required dipole
equalization included. This equalization consists of a notch filter in the region of 1.5k Hz to attenuate the dipole peak, and a 1st order
shelving filter to correct for the dipole roll off at lower frequencies. As can be seen, the result is that below about 900 Hz the transfer function
is even higher by the amount indicated in dark gray, and the slope is even shallower. This indicates that excursion will
increase at an even
faster rate
as the frequency decreases from 1k Hz to 120 Hz.

The last two traces, below the reference line, show the transfer function with and without the dipole equalization included when the 1k Hz
coupling filter on the upper midrange is 4th order. As can bee seen, even after the dipole equalization is applied, the amplitude is well below
the reference curve and the slope is much steeper,  as expected. These curves indicate that the excursion will be decreasing at a rate close
to f^2, or a little less with dipole equalization applied.
Q 3: Just how bad is this with regard to the actual excursion?

A3:  This is addressed in the next figure. Here the modeled excursion of the ScakSpeak 10F driver used in the Note is compared for four
different upper midrange filter transfer functions actually implemented in the new Note. These include midrange coupling filters with 1st, 2nd
and 4th order electrical slopes and  a 4th a case where the coupling filters yield a 4th order acoustic slope. The plots include the effect of all
the required  equalization and were generated for a power level of 60 W relative to 8 ohms. While the excess excursion when using a 1st
order coupling filter does not exceed the limitation of the 10F upper midrange driver, it is clearly far in excess of that which would occur when
using a higher order filter. The case using a 2nd order electrical HP filter shows some increase in excursion between 1k Hz and 160 Hz due
the application of the dipole equalization. However, it is significantly less that of the 1st order and the acoustics of this configuration may be
advantagous, as will is discussed below (see
Q11).
Q5: I understand what you consider the shortcoming of 1st order coupling of the upper and low midrange drivers. But I don't have the
required number of channels of amplification for a fully active 4-way system. Would it be possible to reconfigure the system as a hybrid on
special order?

A5: While the design is for a fully active system, if you wanted to build the system and did not want to invest in additional amplifiers it would
be possible to configure the system as an active 3-way hybrid. However, from consideration of the power requirements across the audio
spectrum,  it makes more sense to couple the upper midrange to the tweeter passively in a hybrid, 3-way active system, and this is the
approach I would follow.


Q6: Why did you choose a digital crossover as opposed to an analog active crossover, and in particular the miniDSP?

A6: I have considerably experience with both the design and construction of analog active crossover and the use and theory behind digital
crossovers. The simple fact is that the benefits of digital far outweigh the liabilities. The availability of analog components for typical DIY
"through hole" PCB construction is becoming more and more limited. Their cost, as well as the cost of PCBs continues to rise. Some
DIYers would prefer not to be required assemble an active crossover or do not have the skills to do so. Additionally, some would suggest
that different operational amplifiers or boutique branded capacitors would sound better than those I would specify. I chose the miniDSP
since it is a very cost effective solution for an active crossover. Being digital, it is easy to program,  allows the user to personalize the
system to meet his taste and the needs of his listening environment, and makes it easy to issue updates if and when they occur, without the
need to replace parts on an assembled PCB. Finally, the investment in a miniDSP crossover is an investment in a piece of equipment that
can be used with any system. When you invest in an analog active crossover designed for a specific speaker, it is simply part of the
speaker.
"Waterfall" presentation of polar response of revised Note from 200 to 20k Hz with
4th order coupling crossover.

It should be recognized that the off axis response contours are normalized by the axial response.
Thus, the off axis peaking around 2k Hz is, in part, a result of the on axis dip at that frequency
see axial response plot above).
Q7: The crossover from upper midrange to tweeter seems high. How does this affect the vertical polar response?

A7: The high crossover point is obviously not ideal, but after considerable testing the 6k Hz point was found to be close to the optimum with
regard to the horizontal polar response. Additionally, the center to center distance between tweeter and upper midrange in the preliminary
design is 10 cm for historic reasons which need not be discussed. This spacing has reduced to about 8 cm in the final design to improve
vertical response in the crossover region. This spacing is the minimum allowed by the size of the upper midrange driver and the wave
guide. For reference the figure below shows the effects of reduced spacing and lowing the crossover point to 4k Hz. Using the -5dB point
as a reference, with 10 cm spacing and a 6k Hz crossover the listening window extends over +/- 13 degrees which translates to +/- 0.69M
at a listening distance of 3M. Lowering the crossover point to 4k Hz and reducing the separation to 8 cm would extend the listening window
to +/- 17 degrees or +/- 0.91M at a 3 M distance. As noted, the final design retains the 6k Hz crossover with the tweeter lowered slightly.
Note: This page may not view correctly on Mac or with
some           browsers (Safari, Snow Leopard)
  I apologise for the inconvenience,
Below is the simulated response of the Note II RS including the dipole woofer response.

A 4th order Butterworth alignment is used, -3dB at 30 Hz*, -6dB at 30 Hz. The 4th order roll off protects from over excursion.

* The plot below shows the preliminary response when the woofer was set to a 35 Hz cut off.
Q8: How did you arrive at the at the crossover frequency between the upper and lower midrange?

A8: This was investigated in the development of the original Note. The polar response of the lower midrange was measured and
examined. Below are presented lower midrange polar response plots in the form of a waterfall plots and individual polar plots in 1/2 octave
increments from 1k Hz to 2.8k Hz. At 1k Hz the dipole pattern is clearly very good. At 1.4 and 2 k Hz, while the response is still reasonable
though some dipole bloom is apparent. However, at 2.8 k Hz the polar response is unacceptable. Based on these measurements, and
corresponding measurement for the upper midrange, a crossover point of 1k Hz was chose for the original NaO Note. This crossover point
has been retained in the current revision.
Polar response of lower midrange from 200 to 4k Hz.
Q9: Why did you decide to discard the U-frame woofer system and in favor of a dipole woofer, and why switch to the SLS woofers?

A9: The switch was made for a couple of reasons. First, while the U-frame performance is excellent, the correct damping is very important
to obtain the optimum performance. Since many DIY speaker builders may not have the measurement hardware required to optimally tune
the U-frame it was decided to switch to the dipole system where no damping is required. This simplifies construction and the extraction of
optimum performance at the sacrifice of between 4 and 6dB maximum SPL capability. It also makes the system a true full range dipole.

The switch to the SLS woofers was also based on several factors. I wanted to reduce the cost of the system. The four SLS woofers cost
under $300. Four XXLS 835016 woofers cost over $900. Other similar woofers cost as much as $1000 for a set of four. Additionally, the
higher Qts of the SLS woofers results in the application of less low frequency equalization and the higher impedance allows the woofers to
be connected in parallel and driven from a single amplifier channel. The lower required equalization is also advantageous with digital
crossover implementation.  If the XXLS woofer are used a series connection is recommend unless the amplifier to be used is capable of
driving a low impedance load. The trade off is that the SLS woofers can produce about 3dB less max SPL at low frequency. This is
compensated for by limiting the low frequency cut off to 30 Hz and by using a 4th order high pas alignment to protect the woofers from over
excursion. Never the less, the stereo pair of woofers using SLS drivers is capable of the equivalent of 100 dB/M at 34 Hz and 110 dB/M at
50 Hz.

However, if you should choose to use different woofers the use of a dsp crossover makes adapting the woofer equalization an easy and
straight forward task.
Note II Rs with grill.
(Camera flash reveals
grill frame.) Mouse
over shows speaker
w/o grill.
The NaO Note II RS,
The next development from the leader in extended range dipoles speakers.
Q4: It has also been suggested that a 1st order midrange coupling crossover can result in more uniform group delay across the transition
between midrange drivers. This seems logical. Is there more to it?

A4: Yes, there is considerably more to it. Certainly a true 1st order crossover with the driver connected in phase will eliminate group delay
associated with the crossover. However, this is not the case when one of the drivers is connected with inverted phase. Furthermore,
regardless of the connection, a true 1st order crossover sums in phase quadrature with asymmetric vertical polar response. If a symmetric
vertical polar response is desired the corner frequencies of the HP and LP filters can be shifted or staggered so that the filters are down
6dB at the crossover point. When one driver is then connected with inverted polarity symmetric vertical polar is restored, but at the expense
of some slight wigglers in the frequency response above the crossover point. These are actually rather minor and could be corrected with
equalization, if desired. The result is that the crossover, while having 1st order slopes, actually has characteristics very similar to a 2nd
order Linkwitz/Riley crossover (LR2)

The bigger issue reverts back to the group delay. If one compares the group delay of a 2nd order Linkwitz/Riley crossover, with that of a 3rd
order Butterworth crossover (B3) with one driver connected with inverted polarity, it is found that the group delay of the LR2 and B3 are
identical. Additionally, the DC group delay of the LR2 and B3 crossovers is actually slightly lower than that of the  staggered 1st order
crossover discussed above, and the variation of the group delay through the crossover region is also slightly less severe for the LR2 and
B3 crossovers. Given that the B3 vertical polar response is asymmetric where as that of the LR2 crossover is symmetric, it would be
apparent that from considerations of group delay and vertical polar response the LR2 crossover would be superior to either the B3 or the
staggered 1st order. But there is still more to consider. When coupling the two midrange drivers the major contribution to non-constant
group delay is not from the coupling crossover but from the 110 Hz HP filter at the low frequency cutoff of the lower midrange driver. The
only way to avoid this group delay variation is to develop a linear phase speaker system using a dsp crossover which is more advanced
than the current miniDSP crossover products.  This could be accomplished usging, for example, PC based dsp crossover tools such as the
Bodzio Ultimate Equalizer.
Q10: Your speaker is shown with a grill. Must the grill be in place to listen to the system and what is the effect of the grill in general?

A10: This is best answered by examining the axial response with and without the grill in place. The effect of the grill will be dependent on
the type of grill cloth used but for a shear cloth, intended for speaker grills, the effect is not very significant. The figure below shows the
measured response of the Note II RS form 600 HZ to 20k Hz, 1/6 octave smoothed, with and without the grill in place. Enough said!  
Q12: After consideration of the Note II RS I have one sticking point. I really don't like the idea of a digital crossover. Will there be an active
analog version using any of the couplers?

A12: There are no plans for an active analog version of the crossover. If you are reluctant to build a system with digital crossover I would
suggest that you consider the Linkwitz LX521.
Q11: Can you provide more detail on the differences between the different order midrange coupling crossover?

A11: Certainly. The figures below show the unequalized response of the combined upper and lower midrange drivers overlaid with the
individual response of each driver prior to applying equalization, with the last figure showing an overlay of the combined response for all
three cases. As can be observed, the combined response is almost identical in all three cases. Only minor adjustments required to the
equalization would be required to make these identical. The 2nd and 4th order result is so similar that no change in equalization was felt
necessary. In the case of the 1st order coupler one additional stage of equalization was applied to compensate for the slight bump in the
roll off of the lower midrange at 3K Hz. Below the frequency response comparisons is a comparison for the group delay for the coupled
midrange drivers after equalization and filtering to an LR4 acoustic band pass with 110 Hz and 6K Hz corner points. That plot should that
the 1st and 2nd order couplers yield nearly identical group delay through the pass band with slightly lower in the 250 to 500 Hz range.
However, all three couplers yield fairly flat GD through the crossover region. Recall, it is not the magnitude of the GD alone which matters, it
is the variation with frequency that leads to envelope distortion. Adding a constant delay to any of the system would not increase wave form
distortion. Thus, if a constant delay is added to the 1st and 2nd order results (approximately 0.1 msec) to raise the GD in the crossover
region to the same level as that of the 4th order, it would be apparent that all three couplers exhibit almost identical GD variation through
the crossover region. This suggest that audible differences between the different order crossovers are most likely due to the changes in
the overlap in the stop bands or are excursion related.
1st order Coupler
2nd order Coupler
4th order Coupler
Overlay of the 3 coupled responses
Group Delay comparison
Red - 1st order
Green - 2nd order
Blue - 4th order
_________________________________________________________________________________________________