The Wharfedale SUPER 3, a cone tweeter initially made by Wharfedale of England in the 50's.  It had a paper cone, foam annulus and a foam spider inserted inside the speech coil and screwed to the top of the pole piece.  The speech coil is about an inch in diameter and is wound with an aluminum wire with a silk insulation.  It's underhung, meaning the coil winding length is the same as the top plate of the magnet assembly, in this case, about 1/4 inch.  The field strength is 14,500 gauss (14.5 T/M); the magnet is 1 lb. AlNiCo.  The original aluminum coils measured about 7.6 ohms with a nominal impedance of 10-15 ohms.  Later coil wire was coated with a dark enamel. 

According to my calculations derived from a gated response at 1m from the cone, the unit is capable of producing a sound pressure level of 97 dB with 1 watt applied and at a distance of 1 meter.  This corresponds to an average impedance in the passband above 5 kHz of around 12.1 ohms.  

This was subsequently verified directly using the ADCOM amplifier to sufficiently drive the speaker. (second pic below)



THD+N  (Total Harmonic Distortion plus Noise was measured at a distance of 1m and a sound pressure level of 100 dB.  This corresponds to 3.66v into 13.4 ohms, the impedance at 6 kHz and is 1 watt of power.    As the frequency rises, the power drops because the impedance rises.  Nonetheless, 100 dB from a tweeter at 6 kHz and above is deemed unrealistic for sensible music listening.  Also, the analyser was set to a low passband below 30 kHz.  This was decided upon due to the fact that the ear can't hear above 20 kHz under ideal conditions, typically. Due to age alone, the ear will not hear above about 16 kHz for a 45 year old male (probably higher for females) and it usually worsens with age.

Freq (kHz)       % THD+N

5                       0.67

6                       1.16

7                       1.05

8                       0.96

9                       0.69

10                     0.69

The drop in distortion above 7kHz is thought to be due to the drop in  power delivered as a result of the rising impedance;

the input voltage to the amplifier is held constant at 3.66 v.


Tektronix AA501 distortion analyser and AC voltmeter. (center)

DM501A multimeter (left)

SG505 oscillator.

This unit was used to measure the distortion.

The setup, however crude, works.

The ADCOM  GFA-535L is fed by the LMS in the computer.  The microphone is a SHURE cardioid condenser powered by a 1.5v or 12v battery.  Most moderate condenser mics are far less guilty of distortion than any speaker by 2 or more orders of magnitude.  The plate to which the speaker is mounted is an original Wharfedale shipping plate.

The mic is fed directly into the AA501 which has a sensitivity of 200 uV, (microvolts),  full scale.

Bandwidth is 300 kHz.



This one was reconed many years ago and on a few occasions, the connection of the aluminum coil wire to the braided lead-in became open.  That it lasted as long as it did was probably luck, or a good mechanical connection because conventional soldering techniques don't work on aluminum.  Notice the left connection, it's been resoldered a few times.  This unit was shelved for years until recently.  It came in handy to make this page.
Here, the bad connection is more obvious.  The gap between the inside of the coil foirm and the pole piece is about 0.008".
Here's the original aluminum voice coil shimmed in  place.  The form length is 0.25", the same as the thickness of the top plate so the windings should be flush with the top.  For larger excursion speakers, the form length is usually twice the top plate thickness; the difference being split above and below.  This allows the amount of windings in the field to remain constant.  It's called Xmax.  If the coil is driven to decrease the amount of windings in the field, nonlinear and harmonic distortion is one result while destruction of the speaker is usually next.
A closer vies of the coil and shims.  The blue one and the orange ones are 0.005"; the green is 0.003".  Another thicker one wouldn't fit easily.  These shim strips are primarily needed to lock the coil in place during the rest of the assembly process.
The basket is now installed.  Usually, for larger speakers, removal of the basket isn't necessary and in many cases, not possible due to its being riveted to the magnet assembly.  Some speakers with cast baskets or frames are held in place with bolts.  Some are even precision machined and keyed so the basket can be replaced with a new cone and coil assembly installed therein.  This is common among many professional speakers.

Be sure to align the terminal strip with the coil wires.  This should be considered when the coil is first installed.

This is the same speaker as above but now has a new cone and coil assembly.  Removing the blue shim wasn't easy due to the flange around the apex.  The foam spider was cut from 0.125" foam using f\a flat washer of diameter 1.012" for a stencil.  It's about 0.010 larger in diameter than the coil form.  The phenolic washers original Wharfedale as is the screw.  Mention of the screw uis important as although it looks like an 8-32 SAE, it's not.  It's a BSW, British Standard Whitworth and trying to force an SAE into the pole piece will damage the threads.  A worst case scenario would be to tap the hole with an SAE 8-32; this will leave enough thread to hold the spider and washers in place.


Another view of the above.
The glue, Titebond 2, was applied with a very tapered toothpick.  It was later worked into the crease.  This adhesive takes about 20 minutes to set after which it has become too cured to hold the dust cap.  Plastic cements set too fast.  They can be applied after the cap is in place but the result doesn't look as nice. 
See?   No visible adhesive.

Purdy, ain't it.

Oh, the annulus is felt. 

I used purple because it looks cool.



The following are three Super 3's repaired in 2011.  Two of them, P1 and P2 are currently being used in a set of Briggs' three way corner enclosures.  The mids are Super 8, FS.  They were acquired decades ago and without cones and coils so I had to hand wind the coils.

There will be a page coming detailing that along with a restored pair of W10FS/B's.


These are most likely P1 and P2 with identical and later version original voice soils.  The coil in the middle is from a Super 8 FS/AL and is the same as that used in the Super 3.

Note the phenolic spider.  Larger versions of this type of spider as used in the Super 12 FS/AL, when viewed from above, resemble a spider's legs, hence the name.

These are P1, P2 and P3.  The P is for the purple annulus; it has nothing to do with the apparent shell game.
A screen capture from the LMS showing the averaged response curves of the P1 and P2.  The data was acquired with a gated response by which means the instrument is capable of ignoring room reflections.  It does this by the operator's inserting of mic to source distance and the distance of the closest reflection.  The instrument pulses the speaker and turns off the mic before the first reflection get's to the mic.  In my room of 17.5' by 20', it's good to about 400 hz.  As the wavelength increases, the difference between the incident wave and the nearest reflection becomes moot and the unit can't differentiate and get a good sample.  It works better outside but is not really needed as the nearest reflection is about 50 feet away.  The ground reflection can be considered minimal as it's present over the entire sweep and is not present in the higher register due to absorption of the ground.  The lower frequencies amount to less than 3 dB reflection unless the driver and mic are both placed on  the ground, then  the sweep is elevated by 6 dB.
A spectrum analysis of a Super 3, unit unknown.  This was done with a 6 kHz fundamental (large peak).  Note the second and third harmonics at 12 kHz and 18 kHz.  they are about 65 dB down.

This analyser is software.  the dip between 40 and 70 hz is the hum filter which wasn't needed for this test but is usually turned ON.

The software works well above about 100 hz but I question it's accuracy below that.  Hey, it's not a Tektronix or HP; it cost me $75 and works better than expected.

Back to the Wharfedale page

Back to the loudspeaker main page