With loudspeakers, virtually every part of the structure vibrates and adds to the overall sound. Even the parts you want to vibrate – the driver diaphragms – don’t behave as you might like and you need to understand how the whole of the structure behaves if you are to stand any chance at all of making a really good sounding transducer.
Senior Product Manager Mike Gough continues his Tools of the Trade blog series with this post on laser measurements:
At low frequencies, driver diaphragm motion is relatively large and you can view it under stroboscopic light, effectively slowing it down or even freezing it. This is typically used to check that the moving parts don’t hit any end stops or tilt over at extremes of motion. But at these low frequencies, the diaphragm motion itself is relatively well behaved. It’s only at higher frequencies that diaphragm break-up sets in and the motion is too small to be seen by the naked eye. Lasers can detect these small vibrations and the resulting output can be amplified so that the motion can be studied.
It is outside the scope of this article to elaborate on the measuring apparatus itself. What is important is seeing the results and interpreting them. Let’s start at the beginning……
In its most basic form, holography is used to create a three-dimensional image of an object that is illuminated by a laser (coherent light) beam spread out using a concave lens to cover the whole object of interest. The reflected light is mixed with light coming directly from the laser and the interference pattern captured on a photographic plate. When a laser beam of the same frequency (colour) is shone through the plate, a 3-dimensional image of the object is formed in the space beyond the plate.
You have to know pretty well which frequencies to use in order to see specific break-up modes of a cone, for example, so it was common to look for peaks and dips in the frequency response first, as these were usually caused by specific resonance modes.
The resulting images show a series of fringes corresponding to nodes and antinodes of resonance and, although the stroboscopic method allows you to see the situation at different points in the cycle, there is no calibration of relative amplitude.
This image, taken in the 1970s, is typical and shows the vibration of a plastic tweeter dome. The outer roll surround is easily discernible and the dome breakup is very complex. It tells you that there is something seriously wrong, but it’s difficult to interpret in detail
This method is similar to holography except that, instead of looking at a hologram, you look at a video sensor that is illuminated by laser light reflected by the object itself, superimposed on light directly from the laser. The picture is speckled, but shows similar fringe patterns and lack of amplitude information.
Both Holographic and Speckle Interferometry are not only difficult to interpret, they are also susceptible to extraneous noise in the system (such as amplifier hum) and the effects of creep (where the object may move its rest position by an order of magnitude similar to the very low amplitude vibrations you want to examine). In order to overcome these limitations, a system of scanning the surface of the object was developed.
Scanning Doppler Velocimeter
This technique is the only one that has been used at Bowers & Wilkins and overcomes the limitations of both earlier approaches. It still measures velocity, so the shape of the object is not directly captured. Modern vibrometers are able to superimpose an image of the object in order to restore absolute shape, but this is not necessary in order to examine the motion and, in the sample images that follow, the target appears flat. The technique captures both forward and backward motion with correct relative amplitude and allows movies as well as still pictures to be produced.
The laser beam is kept focused and narrow, and shines on a single point on the object of interest while the device is excited by the electrical signal. Once the data from that point are captured, the beam is focused on a different point and the measurement repeated. The apparatus maps the whole area of interest as a series of points and constructs a complete picture of the motion. Although the technique can be applied to all parts of a speaker, for simplicity we will concentrate on driver diaphragm motion.
The most common signals used are steady state sine waves (pure tones) and an impulse. Like the earlier techniques, it helps to know which frequencies to choose when using sine waves and the standard frequency response graph is the usual starting point.
Here, we show a driver excited by a sine wave at a single frequency and the outer ring of the diaphragm can be seen to move in the opposite direction from the central portion. It is, in fact, the roll surround moving in the opposite direction from the cone and this cancellation results in a significant dip in the frequency response.
But this is only one frequency and it would be rather nice to see what was happening throughout the effective bandwidth of the driver. To display the whole cone movement over the full audio band is a daunting task, but this surround dip and many other effects are close to being axisymmetric – that means the motion has axial symmetry and you can get most of the information you need simply by looking at a single diameter of the diaphragm instead of scanning the whole surface. If the excitation signal is a swept sine wave, it is then possible to display a 3-dimensional graph in perspective.
In this graph, the lines represent the motion across a diameter of the diaphragm. Amplitude (maximum values are plotted) is the vertical axis and frequency forms the third axis, with the lowest frequency at the back. You can thus examine the axisymmetric non-piston behaviour at all frequencies of interest in one go. In this particular example, the lack of movement right at the centre is the result of the driver not having a dust cap and the laser is scanning the stationary centre pole of the magnet system. There is also an apparent reduction in effective radiating diameter as the frequency increases. This is actually a useful attribute, as it gives a wider dispersion of sound than you would get from a theoretically ideal piston.
In this next example, we look again at the whole cone, but now the excitation signal is an impulse. The cone is formed from a uniform sheet of polypropylene, so shows a great deal of axial symmetry. The initial impulse applied to the voice coil causes first the central section to move, after which a double circular wave moves to the outer edge of the diaphragm. You can then detect a wave being reflected back to the centre, after which the outer portion continues to flap. This sort of ringing motion obviously results in sound radiation well after the original excitation signal has stopped and causes a reduction in clarity. Reflection actually occurs wherever there is a boundary between dissimilar materials. So it happens not only at the coil to cone joint and where the surround sticks to the chassis (or basket), but also where the cone and surround meet.
Again, this type of movement is amenable to being viewed as a 3-dimensional graph, looking across just a single diameter, but in this case the third axis is time (going from left to right) as opposed to frequency. In this example, amplitude is further indicated by changes in colour, so it’s rather like a contour map. You can see wave motion starting in the centre, travelling to the outer edge and being reflected back.
So far, we have looked at drivers with homogeneous cones, but our woven Kevlar® cones are anything but. The behaviour of the cone changes with radial angle, depending on whether you measure in line with or at some angle to the fibres. So in this example, we are going to look at the impulse response of a typical Kevlar driver. You should compare this with the one shown above that has a polypropylene cone, but is exactly the same in all other respects. The motion once again starts at the voice coil, but this time there is just a single circular wave that moves out. This dies away much more rapidly and a 4-fold pattern appears towards the centre, which shows peaks moving both forward and backward. The forward and backward peaks have an almost equal area and so, although air is shuffled across the surface from areas of high to low pressure, very little sound energy is radiated to the far field. This is part of the reason why woven Kevlar® cones sound as good as they do – they radiate very little delayed sound and clarity is better preserved
Looking at diaphragm behaviour is all well and good and gives a pretty useful insight into how the driver will sound, but it’s also possible to see how sound is radiated away from the driver towards the listener. Of course, you can’t get an image from air, but you can from sheets of cling film that are suspended at regular intervals in the line of fire of the driver. Cling film is so light that it barely has any effect on the wave motion and it can be coated with a powder that will reflect the laser light. The driver, in this case a tweeter, is mounted and receives a signal in the normal way, but instead of scanning the diaphragm surface, the surface of each sheet of cling film is scanned. It is thus possible to visualise how sound progresses from driver to listener. The driver is actually positioned close to the sheet second from left and the left-most sheet indicates radiation behind the driver.
This article gives just an insight into how powerful laser measurements can be. In a further article, we shall see how the Klippel Distortion Apparatus not only shows you what is wrong, but points you in the right direction to make corrections.