W90 modification

Initially, the intent was to do a complete analysis on the W90 but while searching my files for info, a file titled The Wharfedale W90 Project was found. After reading that paper, the author didn't include data on that which was done to smooth the response. That may have been due to the numerous variations of the W90. His units were of the original six speaker model. Since I have in my inventory, one 12" woofer, the one without the "bung" and the AlNiCo 5" midrange units as well as the Super 3 tweeters.

The initial tests were performed on the two high frequency sections only, with crossover points of 1500hz and 6000hz. Later, a single woofer was added to get data on the transition from the upper response of the woofer to the lower response of the midranges. No attempt was made to evaluate the bass for two reasons. One, the woofer with the bung wasn't available and two, the system would have to be taken outdoors to get away from the room effect. The LMS analyzer will perform quasi-anechoic responses (gated) but fails around 300hz due to the effect of the room.

Here's a link to the Wharfedale W90 Project It's quite an extensive piece of work.

Despite the results of that project and this one, those systems did sound damn good. I knew a fella in 1961 who had a pair. Of course, back then none of us knew a helluva lot about speaker design which may have been for the better. After all, if they sound good, who cares about specifications?

These two photos show the two upper section of the system. The baffle is within 1/2 inch of spec. The response of this arrangement will be different that in the W90 Project paper due to the lack of reflections within the chamber housing the four units.
PHOTO 1	PHOTO 2

The addition of a woofer section. This cabinet is 3 cubic feet whereas the cabinets housing each `12" unit in the W90 is about 2 cubic feet each. However, this will not affect the upper register of the 12". Also, that this cabinet is sealed, the response in the upper register is unaffected. Both chambers in the original W90 are ported. the one with the cone type woofer may have a vent tube of unknown dimensions and the one with the bung filled woofer has 12 slots also of unknown dimensions. (See photos 6 and 7)

The woofer with the bung won't have much, if any, appreciable output at 1500hz so its absence is moot.

These photos show what looks like the original W90. There doesn't appear to be any evidence of the bungless woofer (probably the one on the left, looking at the front) having a vent, yet the upper chamber and the woofers are plainly seen through the grille cloth.
PHOTO 6	PHOTO 7

FIGURE 1

FIGURE 2

Data and amplitude responses for two 5" units and one ;'bungless' 12" unit in a sealed cabinet. The addition of the 12" unit with the 'bung' will elevate the yellow trace by about 3dB. Doubling the diaphragm area increases the SPL by 6dB but in this case, the power is split between the two parallel units resulting in a loss of 3dB. The net gain is therefore 3dB.

FIGURE 3

There is an error here in the schematic in the lower left corner. It's encircled in red. That line places the 24uf capacitor and the series 50 ohms pot in parallel with the two parallel midrange units. The tweeter section is OK.

For some weird reason, this is the second time I found an error in a vintage Wharfedale product. The other was in a crossover that just didn't seem to work right, so it was opened.

That whole story can be found HERE see fig 2

FIGURE 4

1w1m MID and TWT with 1st order network 1500-6000 BLK-THD

Response at 1w1m with THD, about 2.24% which is high above 1khz

The crossover is the original first order at 1500hz and 6000hz

The level pots are full CW and the 4 speakers are open back but without the rectangular chamber. This response is smoother that that shown in the W90 project because there are no internal chamber reflections.

There is a good possibility that had these 4 units been individually isolated, a smoother response could have been achieved. Alternatively, lining the rectangular chamber with sound absorbing material with the four speakers open back may also have achieved a smoother response.

FIGURE 5

1w1m RED and 1w2m GREY MID and TWT with 1st order network 1500-6000

A comparison of the responses at 1m, red and 2m, black. The latter is more linear which is more likely what one will hear as most will be at least 6 feet or more from their speakers.

The crossover is the original first order at 1500hz and 6000hz

Admittedly, what the response looks like at greater distances is anyone's guess and highly dependent upon the room.

Therefore, take this with the proverbial grain of salt. It was placed here just to show the effect of distance.

FIGURE 6

Closed Back with Woofer/1w1m GRN woof norm pol ORN woof rev pol xover3 800-3000 pots full CW

Woofer added to fill the transition between woofer and midrange. Both at 1w1m with a first order crossover at 800hz and 3000hz

Green is with woofer normal polarity; red with woofer polarity reversed

The chart shows at the top and bottom that the crossover as 1500/3000. This was due to my copying and pasting filenames when making changes and forgetting to make the necessary file name correction(s).

The crossover is xover3, see figure 7

FIGURE 7

The crossover worksheet.

The impedances just below the schematic pertain to the xover 1A and 1B

The two impedances for the mid-range pertain to the impedances of the driver pair at 1500hz and 6000hz.

Those were changed in xovers 2 and 3

FIGURE 8

Here we have the original with the 1500/6000 crossover in black and the modded version with the 800/3000 crossover in red. The black trace here is the same as the red trace in figure 4 The bass extension of the red trace is due to the presence of the woofer. Had the woofer been used for the original design, the bass extension would very likely resemble the red.

Both traces are run with the level pots full CW. The modded version has the woofer polarity reversed.

How much this would affect the aural perception is a matter of subjectivity. Suffice it to say that the 6dB lower response between 1000hz and 2500hz will remove some brightness but this too is a matter of subjectivity.

The lack of extension as mentioned in the W90 project may be due to the perception of the 6dB rise between 900hz-2500hz and 4khz-10khz. See figure 4. The several dB dip between 2500hz and 4000hz will certainly make the system sound a little muted or, as some say, lack of presence.

The following eight spectrograms are taken in sequence at 300hz, 500hz, 700hz, 1khz, 2khz, 3khz, 5khz and 7khz. They all show a lot of harmonics which explains the 2.24% THD. These were obtained with wool packed cylinders housing the four upper frequency speakers which eliminates the effect of the rectangular chamber of the original design. Constructing such a chamber housing the four open back speakers to compare the possible difference inflicted by the internal reflections seemed more trouble that it was worth. Besides, the Wharfedale W90 Project already stated that effect.

These spectrograms were all taken at 1w1m.

These two photos show the individually wool stuffed chambered speakers of the upper section. This would have eliminate internal reflections in the rectangular chamber of the original.
PHOTO 4	PHOTO 5