The Second Thooper Dooper Whizzer

 

This unit uses a stiffer and/or thicker paper, 65# cardstock with a stated thickness of 0.010".  The stock used here measures 0.009" in thickness despite its being called 65#.  Details, details.

The paper used for the previous model is 0.006" which means the speed of sound through this thicker paper will have to be measured.  There will be other effects.

 

PHOTO 1

The magnet for this unit had to be borrowed from a Wharfedale Super 12.  There are two such speakers here that were salvaged from a speaker repair shop sometime in the eighties.  One has the original aluminum coil and phenolic spider; the other has a replacement copper coil and spider.  Both have replacement Waldom cones.  The magnet used is from the second unit described above. The wood cup is to protect the coil which protrudes beyond the basket base.

 

 

 

The wood ring is to elevate the coil to be centered about the top plate and is bolted with the original bolts.  The winding is the same height as the plate thickness, ergo it's underhung.  Since the excursions will be very small above 500hz, this will not be a problem.

This coil and spider are from my salvage inventory. The coil is aluminum and fortunately, I have aluminum solder and flux.

 

PHOTO 2 PHOTO 3

 

 

 

The dowel is glued to the pole piece thanks, J for the idea.  The dense foam home made stabilizing device will be screwed to it and glued to the cone.  The cone, being lightweight and of uniform density should maintain stability by itself but that's an assumption.  The acoustical effect, if any of the centering device hasn't been studied.

 

PHOTO 4

PHOTO 5

 

 

 

Two views in my living room.  Both cones are 6" tall but the grey one is 5.75 inches in diameter at the top and the black one is 4 inches.  There is a reason for this which will be detailed later in the link provided at photo 8.

The woofers are 12" Eminence BASSLITE S2012 with neodymium magnets.  The cabinets are test units with internal volume of 2.8 cubic feet.  There are a few different fronts to accommodate various 12 inch and 10 inch units and all can be converted to vented and sealed.

The corner enclosures are bass reflex of 6.5 cubic feet and each contain a W15FS.  The upward firing mid and tweet are the Super 8 FS and the Super 3. All three loudspeakers are vintage Wharfedale.

PHOTO 6

PHOTO 7

 

 

 

 

 

FIGURE 1

The 5khz half cycle pulse.  A half cycle square wave may have been better but CLIO doesn't generate square waves and connecting another generator to the system wasn't worth the effort.  Besides, music waveforms are anything but square unless one clips the amplifier. 

 

 

FIGURE 2

The red trace is that of the upper mic.  It is advanced by 13uS.  The human ear doesn't come close to detecting this difference.  The reason for that advancement is due to the speed of sound through the cone material being faster than the speed of sound through the air.  Again, see fig 1 in the link at photo 8

This trace was acquired by the setup shown in photo 8 below.      

The initial dip in the blue trace from the lower mic is attributed to the sound coming from the voice coil.  The rest is attributed to vibrations in the cone caused by the initial pulse that set it into motion. This wave will also generate a horizontal wave into the air which will be detected by each mic.

The time between the center of the image (yellow diamond) and the right end is 1mS.  Each vertical division is 200uS.  What should have been done was move the yellow diamond to the left side.  This would give an idea of how long the cone continues to vibrate after the initial 0.0001sec (1/10th of a millisecond) pulse of the 5khz half cycle.  By the looks of the graph, it could have taken as long as 2 milliseconds.

At 1806m/S, it should take 74uS for sound to travel from 1 to 3 along the cone and 70uS to travel from 1 to P6 which means the lower mic should get a signal 4uS before the upper mic.  This figure is in disagreement with that.  Given the crudeness of the apparatus used and the modus operandi employed, the 17uS difference is damn close enough.

 

 

 

 

PHOTO 8

Mics spaced 13cm; 0.013m apart.  The idea is based on the fact that the sound coming from the cone to the bottom mic at 345m/S will reach the upper mic at the same time due to the wave traveling several times faster through the cone material.  This is what produces the cylindrical wavefront.

The whole thing is explained in detail in figure 1 HERE  It's about half way down the page.

 

 

 

 

The three photos are taken from an earlier page to save you the trouble of linking back and forth.  They show the process for aligning the mics, after which they are carefully moved to the cone with the upper mic as close as possible to the cone, about 2mm.  There, the mics are powered by a 9v battery which was later replaced by a DC power supply as the batteries became awkward. Photo 8a

 

 

PHOTO 9

PHOTO 10

PHOTO 11

 

 

 

FIGURE 3

The initial impedance curves with (black) and without (red) the 500hz high pass filter

 

 

 

FIGURE 4

The orange curve is the same as that of the previous figure.  The black curve is the impedance with the high pass filter, a notch filter centered around 3100hz and the cone with two coats of black latex paint on both sides.

 The slight bump in the black trace between 3khz and 4khz is due to the notch filter.  A zoomed part of that trace is shown in fig 5 below.  The vertical scale is so large due to the unusually high impedance at resonance.

 

 

FIGURE 5

The impedance at the peak of the notch filter is about 34W

The system nominal impedance is about 9W around 1khz

 

 

 

 

Initially, the DEQ2496 was used to smooth the curve but was abandoned due to having to reset the equalizer to allow separate left and right eq settings. This would seem to be the better way to go but when two speaker systems are played simultaneously, the subsequent room response may be different than either of the two systems.  The system now used sets each channel separately and by itself.  The two response curves are summed in CLIO and the resultant curve is used as a guide to equalize both channels simultaneously.  This method has been used for decades using analogue parametric equalizers and LMS1 

All subsequent responses are done at 1 watt measured at 8 feet and at a height of 64 inches, ear level when standing as there is no seating position in the room with the speakers at this location and with the L-pad at full CW which essentially is out of circuit.  It would have eliminated some wires thus cleaning the rat's nest of a crossover.  see photo 5

NOTE 1:  The reference of 65p is to more easily differentiate these curves from similar curves made with the other cone.

NOTE 2:  All curves run with normal polarity between woofer and cone.  Also, it was later tried with one and both cones out of phase with respect to their woofers.  With one out of phase, in mono, the sound smeared from between the two speakers.  This aberration was less heard in stereo.  With both out of phase, the sound stayed between the speakers in mono but there was a loss in the lower midrange, very noticeable with a male singer.

FIGURE 6

This is the first attempt at smoothing the response without resorting to the parametric. Several notch filters were tried with unsatisfactory results.  Two eventually did work and they were centered at 3100hz and 5000hz but the hump between 2500hz and 6000hz didn't want to settle down to a satisfactory level.  Wool was stuffed into the cone and the notches kept in circuit.  The response settled down from the red trace (no filters) to the black trace with the filters and the wool.

The 5khz notch was removed due to the dip at 5khz.  However, the wool wasn't intended to be permanent but damping the cone became obvious.  Two coats of black latex paint were applied to each side and left overnight to dry.  On to fig 7

 

 

1w 8ft 65p living room standing norm pol L-pad full CW  RED-as is; GREY two notches cone wool stuffed

 

FIGURE 7

The black trace is without the wool and also without the notch at 5khz.  Note the reduced and shifted dip at that point.  Attempts at narrowing the 3100hz notch filter to reduce the dip failed as that raised the hump around 3khz.  Lowering the notch center frequency wasn't attempted as this would require inductors and capacitors of non standard values.  This would have been attempted had a digital parametric equalizer not been available.

An unintended effect of the latex can be seen beyond 7khz in the black trace but this was of little concern as use of the parametric had already been accepted.

Now, to raise the valley between about 700hz and 2khz and drop the hump centered at 550hz. Since an equalizer is required to raise the valley, it was decided to use the same for the 550hz bump.  On to fig 8

 

1w8ft 65p standing L-pad full CW RED-No EQ; GREY with notch at 3100hz; latex

 

FIGURE 8

Further equalization.  The hump at 550hz is gone as is the valley between 700hz and 2khz.  The response is flat within 5dB from 1khz to 15khz.  The few dB rise below about 800hz wasn't noticeably objectionable and that's subjective.  At this point, the 65p unit can be played at levels as high as 95dB to 100dB without being over bright in the treble.  However, it's still ear shattering but with some music, it sounds awesome. Again, subjective.

1w8ft 65p standing L-pad full CW RED-No EQ GRN-Eq GREY-THD

 

FIGURE 9

A comparison of  left-GREEN, right-RED and both-BLACK.   The left has the notch.

A larger image is available by clicking on this one.  This will show the colours better.

 

FIGURE 10

Both units operating.  GREY-NO EQ;  ORN-PEQ and GEQ A little graphic eq. was added to smooth the low end.  Graphic was used as this is a separate equalizer in the DEQ2496 unit and either can be switched in or out.

Very old jazz and blues popular music dating to the 20's and 30's don't need low end attenuation.

 

 

 

 

PHOTO 12

 

PHOTO 13

 

 

1  LMS.  A system made by LinearX Systems but is no longer manufactured since the passing of its founder.  This system is used by several loudspeaker manufacturers. I still use it as it has several features not included in the CLIO CP-01 unit I have such as gated response which nulls the room reflections.  Another feature is the ability to step the cursor across a curve to get the exact value at that point.

The CLIO CP-01 doesn't have gated response but it will sweep a full response in about a second whereas the LMS can take up to 2 minutes, depending on how many data points are selected.  The LMS I have was bought in 1994 and uses a ISA slot on the motherboard, no longer seen.  As a result, the system runs on XP or earlier; I use Windows 98/ME and run long XLR cables to and from another room.  The CLIO is USB operated, hence portable.

 

Third Thooper Dooper Whizzer

Back to the loudspeaker main page

Table of Contents