Fine Tuning


1st Attempt

Bessel filter with subwoofer and woofer overlapping at a certain frequency


This is the result of my first attempt


You can see that the subwoofer in the left side can boost bass response all the way down to 20Hz. But the right side only 35Hz. This asymmetry can also create other problem.

Here is the step response.


Not bad. But I am thinking of ways to do it better.

IACC for 10ms and 20ms is quite good. But 80ms and late pulse are quite poor giving the overall result of 71.3%


Listening to my current setup, I can say that the high and mid rivals the best system out there. The details, resolution, the dynamics and the musicality is one of the best. But there are a lot going on at the bass region that I need to work on. I am still missing the layering of bass that I heard in some systems. One possibility is the low IACC at later stage which represents the asymmetrical reflections clouding the sound. At this moment, I wonder if I can improve on this if I give up overlapping of the woofer and subwoofer……. More testing to do.

2nd Attempt

In order to identify the problem in the bass, I think I should isolate the sub and the woofer, so i created this crossover for testing. W210 is switched off


Here is the frequency response curve, you can see that with woofer alone, I can get down to about 35Hz.


The impulse response is also a closer match than before


IACC improved to 78.3%, so my subwoofer is doing more harm than good.


Using this new crossover, with the W210 switched off, I am actually getting very strong bass, too much in fact for some track. Now, I think the problem is that W210 uses Vifa drivers and is quite different from AT drivers, when they run at different speed, they tend to cancel each other out at certain frequencies, making the bass less convincing.

Actually, there is a lot you can learn from the impulse response curve. You can see the first impulse is located at 0.1360s, there is another peak at 0.1367s. The distance between these 2 values, if travelled by sound is 0.0007 x 300 = 0.21m. Now, where can this first reflection coming from? You can see that the second pulse is symmetrical for right and left signal so it is not coming from the side wall. The distance between the speaker and the measuring mic is 2.94m, the woofer is 0.58m from the floor. The distance for the sound to travel from the woofer, hit the floor and bounce back to the measurement mic is 3.15m. The difference is 0.21m! So confirmed that it is the first reflection from the floor!!! I order to get rid of the peak, I need to do something to the floor! Of course, this can also be something placed close to the measuring microphone.

From the amplitude curve, I can also get useful information about the bass and where the cancellation occurs.


On the top of this amplitude curve, you can easily read out the information


For example, in the picture above, The point that the cursor is located showed the point at -4.859dB and 52.335 Hz. The corresponding wavelegth (lambda) = 340 / 52.335 = 649.67cm. Allison effect will occur at 162.42cm. Now you can use your ruler to measure the distance from the speaker to the measurement point Vs distance from the speaker to the first reflection point and then from this point to the measurement point. The difference between these measurements that give you the value 162.42 x 2 = 324.84cm confirms that this is the first reflection point that need to attend to.

When doing the physical measurement, an infrared distance meter will come into handy


3rd Attempt

My latest attempt in setting up the digital crossover involves the use of multiple slopes crossover. The setup is now like this

Sub/Woofer : NT 1st order 100Hz
Woofer/Mid: LBessels 1st order 550Hz (set at 465Hz)
Mid/Tweeter: LBessels 8th order 4500Hz (set at 1600Hz)

This gives more coherent and powerful bass with midrange and high clarity. Most recently, Uli has also updated his Acourate program which include pre-ringing removal. Look at the impressive impulse response here. This is as good as you can get.


However, you can still see the pre-ringing of the impulse response. Now, after pre-ringing removal:


This translates to better dynamics and more punch. The soundstaging also gets better too!