|A pair SFB/10 units acquired from a fella
for whom I repaired a Super 8 and gave him a matched unit. They
can be seen HERE. One of those tens,
the one on the right had an open voice coil which was easily repaired as
the break was on the front of the cone by the dust cap. Both were
in need of new annuli.
Actually, both could use new cones as the paper is dry and brittle which can cause all sorts of resonances, the end result of which is a response curve loaded with peaks and dips.
Note the damaged circumference of the cone on the right, caused by removal of the remaining foam stuck in the glue, despite the liberal use of pure acetone.
|Note the machined recess on the left unit which, I assume, was used to center the spider. The left and right orientation of the speakers is the same in all three photos.|
|The damaged cone on the right has no effect on performance but looks ugly.|
|This drawing of the SFB/10 was made
because of an anomaly I discovered. With the spider laying flat,
the coil is raised too high; the bottom of the coil is in line with the
bottom of the top plate, TP. The drawing shows the coil in the
correct position with equal number of turns above and below the top
plate. The coil in these speakers being too high may account for
the leftward lean in the impedance curve at resonance.
The blue curve is that of an original 10"BRONZE/CSB and that coil is symmetrical with respect to the top plate, TP. It's weird as all three baskets are identical. (They were carefully measured) The only way for this aberration in the SFB's to be explained is by assuming the coil form has been fixed too high on the cone or the spider too low, or both.
Since the magnetic field in the gap will be present above and below the gap by a slight amount, and with the bottom of the coil flush with the bottom of the top plate, the force will decrease as the cone moves out and increase when it moves down, due to the decrease and increase of wire in the field, respectively.
It is my theoretical conclusion that this accounts for the leftward leaning in the impedance curves at resonance.
|A = Annulus; B = Basket base; C = Cone (diaphragm); CF = Coil Form; DC = Dust cap; VC = Voice coil; S = Spider; TP = Top Plate of magnet; M = Magnet; BP = Bottom Plate of magnet|
|The response curves below are all measured
under the same conditions, black, red & blue corresponding to the
SFB10J1, SFB10J2 and the 10"CSB, respectively. All are driven
by 1/4w at1/2m which is equivalent to 1w1m. Both responses are
gated to reduce room reflections.
The large peaks between just under 2 khz and 5.5khz seem to be due to traverse waves propagating along the cone. Since the length of the cone from apex to annulus is 3.875", this corresponds to a wavelength of 3512 hz. A half wavelength at half this frequency, 1756 hz will also traverse well, its length being the same. This can be seen in the rise around 1500 hz in the curve. As the frequency decreases, we approach 878 hz and a quarter wavelength at this frequency will traverse to a much lesser degree but can be seen on the curves.
Going up in frequency, above 3512, we come to twice that at 7024 hz. This can allow two full wavelengths to traverse the cone. At a frequency in between the two, 5268 hz, with a wavelength of 2.58", we have 1.5 cycles traversing the cone. If there is a reflection at the annulus, the the second half of the second cycle will be reflected back, partially canceling the first half of the second cycle. What's left of the energy in the reflected wave will support the second half of the first cycle and decrease the first half of the first cycle. The reflected waves are out of phase with the incident wave causing the various peaks and tips. These variations are also referred to as resonances within the diaphragm.
Despite that these frequencies are spot tones, they last for less than 7.6 milliseconds, the time for the first reflection to reach the microphone. The mic is turned on just about 2.9 milliseconds after the tone signal is sent to the speaker. So, the mic is on for about 4.7 milliseconds. However, since the speed of sound is faster in the cone material than in air, it takes less than 19 microseconds for a traverse wave to travel from the apex to the annulus. This is 1/247th (0.004) of the time the mic is on, ergo, they will be sensed by the mic and if strong enough, more than once.
|The following are spectrum analyses of the
pair, the J1 on the left and the J2 on the right.
Going down each column are the spectra at 1, 2, 3, 4, and 5 khz., the fundamental being the tallest trace on the left of each photo.
The horizontal scale is marked at 5, 10, 15, 20 and 25 khz and is linear.
The vertical scale, also linear is in dBu, referenced to 0.775v. At around -44 dBu, the peak at the fundamental, this would indicate a microphone output of about 4.5 millivolts.
|SFB/10 J1||SFB/10 J2|
Back to the Wharfedale Index
Back to the loudspeaker main page