HS/CR3/2 crossover

In January 2015, a pair of corner cabinets were built for an all Wharfedale three way system. It's components are W15FS, Super 8 FS and a Super 3. It was later observed that the left channel sounded a little brighter than the right one. The data collected for the mid-ranges and tweeters was checked and were found to match. It should be noted that the voice coils for all four units were hand wound months prior to building the cabinets. The original voice coils were damaged beyond repair or even salvaging the aluminum wire, so they were rewound with copper wire. However, they were removed and checked again and found to match, It was thought that the drapes aside the right channel on one wall had an effect but that was later found not to be the case. The crossovers were considered but since they are sealed in wax, further investigation was abandoned. In retrospect, what should have been done but wasn't, if memory serves me well, is reversing the crossovers. That this wasn't done is indicative of the fact that if it had, the source of the problem would have been discovered. However, being sealed in wax didn't allow an easy fix; the effect of the problem was easily remedied by a few tweeks on an equalizer. The left cabinet is shown in photo 1.

Fast forward six years and that problem was again addressed as the the result of a conversation in an audio forum on AudioKarma.

Having three of these crossovers, all three were taken to the test bench. The results are discussed below.

Photo 1

If you're curious, the other audio paraphernalia is described below from lower left and clockwise. Restored hallicrafters S38-B AM receiver covering from 540khz to 33mhz. Bought and restored in 2006 for nostalgic reasons; my father had one and I just happened to stumble on this one on eBay. Above that is a 1954 H.H.Scott 99B integrated amplifier. A full description of it and its restoration can be found by clicking here HHScott99b The tape recorder is an Akai GX4000-D, one of 7 tape decks connected to these systems via a patch bay.

On the wall is a short line array made comprising 4 DynaVox full range speakers of diameter = 3.5" in each channel. They cover the range above 300hz; the eight woofers are under the sofa and are HiVi B4N 4" units. The original idea was to use the W15 bass units for the bass but the switching became too complicated. the corner systems are already relay controlled to be driven by the Scott or the Adcom via the little dark blue box on the end table.

The octagonal cabinet on the right houses a Dayton Audio PS180 point source speaker. these guys are flat from about 100hz to 17khz with exception of a nasty slightly greater than one octave peak centered around 2300hz. An adjustable notch filter made short order of that. the associated woofers are MCM (now Newark) 55-2981 ten inch units. Those cabinets are 3 ft^3; the table lamp sits on the rear half of that cabinet. This system is controlled by two other Adcom amplifiers through an electronic crossover. The Adcom preamp can be switched between the 200 watt arc welder to a pair of Adcom amps. the switch is mounted in a well grounded aluminum box behind the arc welder. Reference to an amp as an arc welder goes back to my teen years when anything over 30 watts was facetiously considered an arc welder.

Figure 1 The schematic as presented in the audio forum. What surprised me were three things. First and foremost, the low pass section is second order and only the high pass section of the mid-range band is second order. The low pass section of the mid-band and also the high pass section of the tweeter are first order. The second was the manner in which L3, the low pass filter of the midrange section is shown attached to pin 2 of R1. It's usually shown between C2 and pin 3 of R1 as seen in figure 1a here. The third is the high pass filter of the tweeter being wired in series with the high pass filter of the mid-section, giving an effective capacitance of 2.7uF. This was first noticed in a Klipsch design.
Figure 1a The dashed line shows the familiar wiring with C3 taken directly from the input. (fig 1a) The solid line from the junction of C2 and L3 to C3 would be eliminated. The input of L2 could have been tapped at the junction of C2 and L3 (fig 1a), which is electrically equivalent to that which is shown. The low pass section is omitted for clarity
Figure 2 This shows the schematic with each component identified. Note that R1 and R2 are not L-pads. With the pad fully counter clockwise, pin 2 to pin 1, maximum attenuation, the 50W resistor will dissipate the power delivered by the amplifier to the mid section; the mid-range would then be connected to AC return. These are wire wound variable resistors and will handle 15 watts with ease, perhaps more. A 10w power resistor is usually encased in a ceramic casing and is wound with a much thinner wire that that found in these pads. As for crossover point shift, this will shift by a few hundred hz depending on the position of the pad. An L-pad maintains a much smaller load variation, somewhere around +/- 1.5W. Keep in mind that these systems weren't designed with the fanaticism seen today. The Briggs' three way was actually compared against a live orchestra many times in Royal Albert hall in London with astonishing results. The schematics shown in figs 1 & 2a are electrically equivalent. Nuffa dat. On to the problem found. In figure 2, note the yellow highlighted line and the ground symbol just above the mid-range. The ground symbol should not be present. This is what was found in the wiring under the wax. The line highlighted by yellow wasn't there and that end of L3 was connected to ground, actually AC common or return, point B as shown in photos 4, 5, 6 and 7, further down.
Figure 2a The familiar manner if a parallel network. This is equivalent to that shown in fig 1 with the exception of the input to C3.

Figure 3

This resulted in the mid-range operating in two way mode regardless of the switch position. Initially, the switch was suspected and after isolation of the switch, measurement with an ohmmeter showed that when closed, the contact resistance was, at minimum, 10W at best and tapping on the switch with a pencil caused that resistance to change radically. The switch was then left open (disconnected) and the response of the mid section remained in two way mode. The response at the mid section was always as shown by the yellow trace. The red trace is the woofer, green is mid-range in 3-way, yellow is mid-range in 2-way and blue is tweeter.

The left scale is relative, not actual SPL. It is dependant on the voltage applied to the speaker.

The orange trace is from a previous sweep that was not removed.

After rewiring the innards (L3), the green trace was noted in three way mode with the switch open. A temporary slide switch was inserted in place of the defective toggle switch pending delivery of parts from Mouser Electronics. See photo 2 below.

Photo 4

Photos 4, 5, 6 and 7 show different views of the innards. The letters A and B designate the inner and outer leads of the coil, L3. The wire surrounded (it surely isn't circled) by the black line is the wire going to the return line, aka ground, signal return, at point B. After clipping it was when it was noticed to be aluminum. So, out came the aluminum solder and flux. This solder has a much higher melting point that regular solder. The aluminum solder is 2mm in diameter (12 AWG) and difficult to melt with a 25 watt soldering iron. So, the aluminum solder was rolled flat and cut into strips which facilitated the flow. That aluminum solder was purchased at a plumbing and AC supplier and is intended to be used with a propane torch which I wasn't about to try here.

Note the short wood match at the upper left corner. It was found in the core of L1 also in that corner. It wasn't broken as the end looked just like that of a normal wood match. Odd that it's so short and odder yet is why it was there.

Yes, I left it there, just fer schitts 'n giggles.

Photo 8

If it looks like the ends of two wires, it is. The larger right one is that of the L3 coil after being removed from AC return. To its left is taped a piece of 18AWG copper wire. Both are solid conductors. The right one is aluminum, as are all the coils in the crossover. It measures as 16AWG.

Aluminum is 30% less dense than copper but its resistivity is 1.5 that of copper. By juggling the numbers, it seems that weight may be the prime consideration as these units were shipped all over the world.

Perhaps an aluminum wire manufacturer made them an offer they couldn't refuse.

FYI. A 4mh air core inductor wound with 18AWG copper wire will have a dcr of 0.9W to 1.2W depending on the coil dimensions. This is similar to the 1.17W dcr of the L1 and L2 aluminum coils here which takes me back to the aforementioned unrefusable offer.

Photo 9

A brief note on oil filled capacitors. This is a motor start/run capacitor purchased as a replacement for use in a tape recorder motor. The look like the oil filled type but I think those were discontinued decades ago other than those manufactured for boutique audio stores to separate people from their money, much like the purveyors of exotic wire to connect speakers to amplifiers and also exotic interconnecting cables and the beat goes on, and on, and on ... ad nauseam.

These modern units have a polypropylene dielectric, are resin filled and are claimed by the manufacturer to be self healing. the 50/60hz stated is for maximum current handling and is of no concern if used in loudspeaker crossovers. These modern capacitors cost but a few dollars and in a speaker system, will last forever. They are used in AC units which are outdoors, exposed to ambient temperatures especially in Arizona well over 100F for months, yet they work. My AC units that have been replaced a few times in the 47 years I've had this house have never needed a capacitor replacement.

The accuracy of this tester is +/- 4% for capacitance and +/-1.5% for ESR. This fella tested as 1.98uf and 0.32W ESR on the PEAK. On an AGILENT it tested as 2.005uF +/- 1.2%. On a TENMA, it tested as 2.00uF +/- 0.5%. The Agilent doesn't check ESR; it's a multimeter, not a capacitance meter. The Tenma is a combination capacitance and resistance meter.

I'm sure there are those who will insist that they hear a sonic improvement with certain types of boutique priced capacitors and inductors, yet none would dare to participate in a blind test.

A fool and his money are soon parted

Thomas Tusser 1524-1580