How to correct the signal (time) of the speaker driver?

Since the early 1980s, the word "time alignment" has been used over and over for comment, but people have never been very accurate about it.

Correct signal (time) alignment for speaker drive.

In fact, people have long noticed this concept. Engineers who supplied speakers for the first audiovisual film The Jazz Singer found that in tap dancing scenes, there were different arrival times for footsteps from the high-frequency horn and the low-frequency unit from the folding horn.

From then on, planners tried their best to tune the speakers at all times. Time alignment is the trademark of E.M. Long. To facilitate comments, we will use the generic term "signal alignment" to avoid the usual use of? And TM.

Many people think that signal alignment of the drive inside the speaker box is actually measuring the difference between the voice coils of each drive and the front of the box, and then adding the corresponding delay for the drive nearest to the front of the box, so that all the drive signals can be correctly aligned.

But this is not accurate! To align the signals of high frequency and low frequency drivers, we need to know the filters and phases first.

All filters are "rotated" so that the frequency of the filter is "forward". Because 360-degree phase transfer is equivalent to a wavelength, the wavelength can be described at intervals or times, so any phase transfer at a specific frequency can be indicated by a certain signal delay.

For example, 1,000 Hz indicates a cycle of 1,000 cycles per second, so a wavelength (or period) is one thousandth of a second, or one millisecond (ms). Thus, 360-degree phase transfer at 1 kHz is equivalent to 1 millisecond delay, 180-degree phase transfer (1/2 wavelength) is equivalent to 0.5 millisecond delay, and 90-degree (1/4 wavelength) is equivalent to 0.25 millisecond delay. As for 2 kHz, the delay time of phase transfer is half of 1 kHz because the wavelength of a period is only half of 1 kHz. By this calculation, the 180 degree phase transfer at 20 Hz (1/2 wavelength) is equivalent to a 25 millisecond delay or 28.25 feet (at the speed of sound).

Practice and Application

What does this have to do with my topic? All frequency dividers and equalization filters are electronic filters that cause the signal passing through to be phase-shifted/delayed. Similarly, all speakers are acoustic filters and signal delays. Therefore, in order to align the signals of the bass unit and the drum driver (or the high-pitch unit), we should not only compensate for the difference of the physical interval between each driver and the front of the box, but also compensate for the frequency divider, the unique post-frequency equalization filter of each driver and the phase transfer delay of the filter formed by the speaker box as the acoustic filter. At the time. The pre equalization filters are not taken into account, because they cause the same delay for the two drivers.

Now let's apply the new knowledge to practice and align a two-way speaker system. The system consists of a 12-inch low-frequency unit (low-frequency part) and a 90 x 40 microphone / compact driver (high-frequency part).

Before you start, make sure that the polarities of both drives must be common, or at least the associated polarities need to be common. Polarity can be viewed in a variety of ways: (1) viewing the wiring; (2) using a polarity viewer for each driver without an equalizer or a frequency divider filter enabled; (3) using a measurement system to view the pulse response to the initial positive sensation.

Figure 1 shows the echoes of low frequency and high frequency. The measuring Mic is placed in the center of the two drive, and the interval is five times the diameter of the low voice unit. Note that before adding the 24 dB/octave (4th order) Linkwitz-Riley frequency divider, I have now adjusted each part of the frequency response to be smoother over EQ, as well as at the frequency divider.

Correct signal (time) alignment for speaker drive.

Figure 1: Independent low-frequency-high-frequency response using a 24 dB/octave Linkwitz-Riley frequency divider with a frequency dividing point of 1 kHz

I find that the driver can be equalized first by passing through the special frequency division filter of each driver, which can provide the smoothest frequency response for the frequency division area when the high and low frequency response merges. This also allows the frequency divider to be merged in a closer approach.

Note that the frequency response curve intersects the acoustic frequency. For signal alignment, if a fourth-order filter is used, the intersection point should be at - 6 dB. To achieve this goal, it is necessary to ensure that the levels of each driver are common, and then adjust the frequency of the electronic divider until the desired acoustic results are achieved.

In this example, I expect to get the 1 kHz frequency divider. To achieve this goal, after all, the frequency of the two drivers is 950 Hz. Keep in mind that electronic frequency division is closely related to and affected by equalization filters and acoustic filters (i.e. speakers), and that equalization filters and acoustic filters are capable of producing truly useful acoustic results.

Fig. 2 shows the combined frequency response curves of the two actuators, which are superimposed on the independent frequency response curves of low-frequency and high-frequency. Please note the offset at the frequency division and the small increase in 600Hz proximity. The 11 dB trough indicates the need for signal alignment for the drivers, which cancel out each other's outputs because they occur at the same frequency with different phases. This can not be balanced by patching, because it will affect two drives together, and the same will cancel out.

Correct signal (time) alignment for speaker drive.

Fig. 2: the combined frequency of the two drivers has a 11 dB trough at the frequency division.

Fig. 3 adds the phase curve of the combined frequency response. Notice the change in the slope of the frequency division of the phase curve. This also indicates that the misalignment of the driver signal forms the trough of the frequency response curve.

Correct signal (time) alignment for speaker drive.

Figure 3: From the drive combination frequency response diagram with phase curve, we can see the sudden change of slope at the frequency division, showing that the driver signal is aligned.

At this stage, most engineers performing signal alignment operations add delays at the beginning to the drive nearest the front of the box and investigate the phase curve until the slope becomes as straight as possible ("straight" rather than "flat"). If your real time analyzer (RTA) can't measure the phase, it's too bad. This can also be a very mundane task, since the last few delayed processes look almost identical on each side of the phase alignment optimization.

This may not matter for frequency response, but the signal alignment here also determines the direction of the lobes on the axis at the frequency division. To make the lobe straight on the front side of the box, it is better to obtain the best alignment settings at the location of the Mic location.

To find the most accurate alignment settings, the easiest way is to use real time parser.

The polarity of the rotary high frequency actuator (polarity), rather than "phase". Then start adding delays to the drive closest to the box -- in this case, the drive closest to the box is the bass unit.

Find the largest offset in the frequency division. Unlike the previous method of straightening the phase slope, this method can easily determine the delay step at the maximum offset. The trough may be 30 to 40 dB deep, and this trough will be several dB smaller even if it is further or lower than the best delay.

Fig. 4 contrasts the combined response curve before and after high frequency rotation polarity. Fortunately, the frequency response curve of the polarity of rotation looks quite smooth.

Many people stop here and start using the system. People often do this before DSP (digital signal processing) is presented. The passive frequency division system built in the loudspeaker system is usually the 10dB/ octave (2 order) divider.

Correct signal (time) alignment for speaker drive.

Figure 4: Composite frequency response curve (trough) with the same polarity of the driver and combined frequency response curve (flat) with polarity rotation of the high frequency driver.

The 2 order divider has 3 dB attenuation at the frequency division point, and the phase difference between the drivers is 180 degrees. The phase of the revolving high frequency part allows them to phase together and achieve an increase of 3 dB at the frequency division. Many speakers with passive frequency dividers are designed in this way. At this point, an important question is: Can you hear the difference between positive polarity and rotational polarity signals? The short answer is: if the signal is a very asymmetrical waveform, you can hear the difference; if the signal is a very symmetrical waveform, you can't hear it.

Fig. 5 So unless you're listening to flute solo, you might need to use modern DSP functions to provide optimized frequency dividers for both drives to keep their polarities in common. As can be seen from Fig. 5, in the combined echo of high frequency polar rotation, the phase slope at the frequency division frequency changes slightly, indicating a certain degree of asymmetry.

Correct signal (time) alignment for speaker drive.

Fig. 5: the combined response curve of high frequency polar rotation. Please note the small turn of the phase curve at the frequency division.

Figure 6 shows the process of finding the offset at the frequency division when the high frequency driver turns polarity. The trough depth at the frequency division is 37 dB, and the best low frequency delay is 0.417 milliseconds. Please note that it is 10 dB deeper than the nearest 0.396 millisecond delay step.

Correct signal (time) alignment for speaker drive.

Fig. 6: the most profound offset is found when the high frequency driver turns polarity.

Figure 7 depicts the deepest offset phase curve. This is a straight straight line indicating that it is just 180 degrees out of phase.

Correct signal (time) alignment for speaker drive.

Fig. 7: the deepest offset phase slope is a straight line indicating that it is just 180 degrees out of phase.

Once the polarity of the high-frequency drive is found to rotate, the delay step with the deepest cancellation occurs, and only the polarity of the high-frequency drive can be rotated again. Your system is now aligned.

Figure 8 is the final result. Compared with the high frequency response curve of rotation in Fig. 5, the slope of the phase curve in the frequency division area is more straight and gentle, and the response curve of the bass unit near 600 Hz does not have the high frequency offset trough.

Correct signal (time) alignment for speaker drive.
Fig. 8: the final signal alignment after the high frequency phase is reversed again.

If you have the measurement system of phase measurement, please admit that the final phase slope is a straight line. This is to avoid adding delays to the wrong driver, or delaying the correct driver 360 degrees too much or too little at the short wavelength frequency division, resulting in a deviation from a frequency cycle. After all, the frequency echoing may look the same. If the real time analyzer is used without phase measurement, special attention should be paid to this.