www.audioexcite.com

I’ve built the two prototype cross-overs described in my last post and I’ve started the listening evaluation of the two.

Before I describe my findings from the listening tests, I thought I should describe some cross-over simulations I’ve done for the two prototypes.

Both of the below simulations shows the frequency response optimized for max linearity at the 15deg off-axis measurement. A flat frequency response like this can in some loudspeaker designs sound a bit too bright and I will in my listening evaluation test different tweeter attenuations.

When I publish the new Sequence Two – Monitor design I will suggest different tweeter attenuations options in order to satisfy personal sound preferences and HiFi equipment matching.

These above four pictures show the “5deg backwards tilted baffle” filter design. The frequency response is very linear and smooth. The phase tracking between the individual driver units is great. When connecting the tweeter without reverse polarity you get as expected a -26db suck out in the frequency response centered at the 3kHz cross-over frequency.

These above four pictures show the “ladder delay network” filter design. The frequency response is also here very linear and smooth. The phase tracking between the individual driver units is even greater. When connecting the tweeter without reverse polarity you get a very deep -35db suck out in the frequency response slightly below the 3kHz cross-over frequency.

Both of the above simulations are optimized for a >2.5m listening distance.

This above picture shows a control measurement done at a 2m distance, 15deg off-axis and is valid down to 400Hz. The blue response is the tilted baffle version and the red is the “ladder delay network” version.

The control measurement confirms the simulated ones and the difference between the two filter options is very small. The main difference is that the “ladder delay network” version have +1dB output at 4kHz.

The suck-out centered at 1250Hz is some sort of cone edge resonance for the mid-woofer and it doesn’t look nice in measurements, but doesn’t hurt sonically.

The same goes for the tweeters rising frequency response from 15kHz and up. It may not look nice in measurements, but doesn’t hurt the performance of the treble.

The million dollar question is……. Which one of the filter versions sounds the best?

I will have to get back about that, since I need more time for listening evaluation before I make a judgment.

To readers of the AudioExcite blog, there should be no surprise that I like two-way stand-mount loudspeakers and here is another one and most likely not the last one. 🙂

With the Sequence Two – Monitor we take a step up in both the price, but also in sound quality.

At this price point you should expect good measurements and consistency between the driver units. The chosen drivers for this design fulfills that more than enough.

The response dip centered at 1250Hz for the mid-woofer sure does look nasty, but despite some higher second-order distortion it doesn’t affect the sound quality of the driver and is actually only a beauty spot in the frequency measurements.

There is a very nice consistency between the two mid-woofer samples. They have a very flat frequency response from the mid-range to the upper mid-range 1.5-4kHz. The driver has a very nice off-axis behavior and starts “beaming” as high as 3.3kHz. The rising top-end is easily tamed with a response shaping circuit in the filter design and there is no severe and nasty cone-break-ups that needs to be treated.

This mid-woofer just begs for to be crossed-over high up in the mid-range and the listening test confirms it.

For further details see:
AudioTechnology 15H520613SDK

After using my DEQX system for a while in order to break-in the drivers and to get a feel of the driver units behavior and sound characteristics, I’ve decided to cross-over the drivers at a fairly high frequency, around 3kHz.

One might argue that a tweeter like the ScanSpeak AirCirc driver is over-kill for such a design and why not use a small ¾” tweeter instead, but in this design I want to use a true symmetrical LR2 cross-over slopes.

Using a true LR2 cross-over slopes requires some robust tweeters to handle the shallow frequency response slopes as well as a well-behaved mid-range without frequency anomalies above the cross-over frequency.

The AirCirc tweeter has an extended low-end frequency response and has the distortion numbers to back it up.

The two tweeter samples have almost a perfect consistency between each other. It simply doesn’t get better than this.

The rising top-end above 15kHz is nothing that needs to be corrected for. If it at all can be heard it gives the high frequency a bit more air.

For further details see:
ScanSpeak D3004/660000

I think these two mid-woofer and tweeter drivers complement each other perfectly. 🙂

When using a symmetrical Linkwitz-Riley 2nd order cross-over filter topology, the mid-woofer and the tweeter “acoustic center off-sets” is a very important factor to consider for a successful design.

The acoustic center off-set between the drivers in the Sequence Two – Monitor design is about 26mm, when measured at tweeter-axis and at a 2.5m distance.

There are several solutions to adjust the acoustic center off-sets between the mid-woofer and the tweeter and here’s the most commonly used techniques:

1.    Use an asymmetrical cross-over slope.
2.    Use a stepped baffle.
3.    Use a slanted baffle.
4.    Tilt the enclosure back-wards.
5.    Use a ladder delay network circuit.

Each technique has its pros and cons. The most commonly used “asymmetrical cross-over slope” technique can often cause poor phase tracking around the cross-over point.

Using a “stepped baffle” can cause additional baffle diffraction and using a “slanted baffle” cause added enclosure building complexity.

In my prototype design I will use the option 4 and 5. The “ladder delay network” adds complexity to the filter design, but doesn’t have the same issues in the vertical off-axis response roll-offs as the option to “tilt” the enclosure back-wards.

Listening tests will decide which of the solution 4 and 5 that performance the best. If they are equally good, I will publish both options as a part of the design.

The first (left) prototype mid-woofer cross-over section is very simple and consist of an electrically first-order section (L1) and a response shaping circuit (C1+L2+R1) to achieve a full “baffle step correction” and an acoustical LR2 cross-over slope with a cross-over point at 3000Hz.

The tweeter cross-over section consists of an simple electrically second-order section (C2+L3) and an adjustable tweeter level attenuation (R2) together with a “notch filter” (C3+L4+R3) to flatten out the sharp impedance peak at the resonance frequency of the tweeter.

The tweeter “notch filter” is a must in order to achieve an acoustical LR2 cross-over slope with a cross-over point at 3000Hz and to avoid audible distortion and “ringing” at the tweeters resonance frequency.

The second (right) prototype cross-over filter has the “ladder delay network” (L4+C3 and C4+L5) circuit connected between the cross-over filter and the “notch filter” section.

Observe! In this filter the tweeter isn’t directly connected to ground.