Let’s start with some bad news: A flawed speaker in an acoustically difficult room won’t become a test winner with using and applying all these techniques.
The good news: As long as you have got a well-designed, good sounding speaker in a well thought through acoustical set-up, you will dramatically benefit from this concept.
The basic mechanical properties of you speaker are the effective limits of this approach. If for example your loudspeakers directivity plot looks like a Christmas tree this cannot be cured by the suggested approach. In most cases, you will have a well-designed speaker, in which the designer has optimized the transition frequencies radiation patterns to make them as smooth as possible.
In addition, any membrane resonances, diffraction or other artefacts, mainly those originating from mechanical parameters or constraints, aren’t really correctable.
However, as said above, a well-designed loudspeaker will be able to be dramatically enhanced. The only thing you have to do is removing the internal crossover and route all chassis connections to the outside, so that you can directly connect them to the amplifiers.
Acourate offers a whole range of possible crossovers. One of the key properties is the possibility to use linear phase crossovers. Uli Brüggemann, author of the Acourate toolbox, is describing its advantages in a white paper. This cannot be achieved with conventional “classic” analogue crossover design, without major performance limits.
Experience and many listening tests have led to the preferred selection of the so called UB-class crossovers. Choosing a higher polynomial (such as jpol11) to set behaviour in the transfer region is favourable. Any order higher than two may create adverse sonic effects.
Once selected, you get an almost perfect crossover. It is linear phase and the impulses add-up to a perfect Dirac pulse.
There are two steps necessary with driver linearization: phase correction and amplitude linearization. Please make sure, that when measuring your loudspeakers, that they stand free from any obstacles and that measurement distance is somewhere between 20 to 40 cm on axis. Feel free to also experiment with other set-ups, mostly generating a sonically different result. Each chassis must be measured individually. The phase of each chassis needs to be aligned so that at the crossover frequency it is in-phase with the chassis it crosses over. If not – in worst case at 180 degrees – it cancels out completely at the crossover frequency. And that means we will have a non-correctible dip! Technically, the crossover for each channel is convoluted with an IIR filter which corrects the phase. Then you can linearize the amplitude.
If you also want to linearize the bass chassis you may either chose a ground plane measurement or a close-up measurement technique so that you do not accidentally measure the room instead of the chassis. Important: Do not attempt to linearize any sort of dipole speaker. In case you use dipole speakers, it is much better to apply the Acourate overall room correction only. Having a close look at the result or this one illustrates the potential for linearization, even for very high-performance chassis.
Bear in mind, that while linearizing your individual chassis you also alter their way of reverberant sound pattern in your room.
Finally, the linearization is convoluted into the originally generated crossover and is replacing it.
The total outcome of this process will create a re-worked crossover.
With the loudspeakers’ chassis acoustical centres never being totally aligned your system does require time alignment. This is an easy task for a digital system, meanwhile in the analogue domain it is very complicated or even impossible.
To do so, one introduces an artificial delay. Knowing how Acourate works, the timing difference between the tweeter’s impulse response and the artificially delayed chassis can be identified. This is expressed as a delta in samples and compensated for. At 48 KHz one sample delay equals a physical distance of approximately 7 mm.
Having now created a linearized, time aligned crossover, one can use the Acourate toolbox to optimize your loudspeakers in your room!
Alternative Time Alignment
A more sophisticated method of time alignment is the sinc convolution. The procedure as described above is an approximation for a system when it is in steady state. It does not consider the fact, that the system swings in and out.
Taking an ideal digital crossover and convolute the filters with a sinus chirp at the transition frequency produces an ideal waveform when added. Convoluting the same sinc with the measured response of the filters doesn’t. Optimization is done with modifying the signals rotation (e.g., time shift) and amplitude so that its swing-in and the periodic behaviour approximates the ideal one.
Let’s have a look at a practical example: The integration of a subwoofer. The transition frequency is 70Hz. The standard method shows quite some phase shift at the crossover frequency. When optimized with the sinc convolution, this improves dramatically.
The alternative procedure is a bit more time consuming, but worth the effort. The integration of the low frequency drivers into the sonic image is a great improvement.
Reduction of low frequency distortion
Adding a subwoofer to an existing speaker system is always a challenge. The phase relationship may be arbitrary and the physical distance of the sub from the main speaker is hard to overcome. Even a digital system can get to its limits. Using the alternative time alignment as described above is a huge step to overcome this challenge. The real benefit (and yes, this is audible) is the dramatic reduction of distortion when professionally done. The following chart shows a high-end sophisticated speaker without subwoofer.
It does have sufficient low frequency power, but its distortion goes up towards lower frequencies. It is mainly k3. The perfectly integrated subwoofer completely changes the picture. Now, mainly k2 is dominant, but on a very low level. If you aim at maximum bass clarity, this is the way to go!