EMPYRE

 

I've always been fascinated by the EMPIRE Royal Grenadieer 9000M, mostly because of its appearance.  Originally designed to be obscure in a fashionable living room of the early sixties, it stands only 29" high, placing the midrange and tweeter rather low unless one is seated.  They could easily blend in with the other furnishings, sometimes being used as end tables, rendering them very compliant  to the WAF. (Wife Acceptance Factor)  In 1960, they sold for $299 each. (about $2600 in 2020)  Consider that a new Chevy or Ford cost about $1800 back then.  The average cost of a new car then was about $2600.  This average included Chevy, Ford, Dodge, Cadillac, Lincoln and Chrysler, placing the first three below that average.

The drivers used are the Dayton DC300-8 woofer, a 6" wood cone speaker from Midwest Speaker, the MW-7065 and a Peerless (Tymphany) BC25SC55-04  The cabinets are 39" high, 21" wide across the flats of the sides and weigh about 100 lbs.

Below is a photo of the 9000M.  It can easily be seen that my pair are in dire need of cosmetic surgery.  Should I permanently assemble my pair, they will get that much needed cosmetic surgery, including the white marble top.

 

 

 

 

Photo 1

The test setup.  Originally, the bottom box was only 3 ft^3 but has been enlarged to 4.8 ft^3, the closest I could get to the intended 5 ft^3 using left-over MDF. The octagonal cabinet is 5.1 ft^3 without the mid-range chamber, vent crossover and crossmembers and, of course, the woofer.  This was of no concern as the reason for the presence of the woofer is solely to determine an optimum crossover frequency and slope as well as its effect on the crossover vs a resistive load. 

The crossover components were determined by running a gated frequency response at 1w1m and recalculating component values to obtain a response flat within 5 dB from about 300hz to at least 15khz.  See Fig. 7.

 

 

 

 

Photo 2

Just the 16 sides with a foam strip attached to one edge.  This was done to seal the enclosure after assembly due to space requirements.  This leaves the option to disassemble them for storage under a bed.

 

 

 

 

Photo 3

The internal bracing.  Each crossmember is screwed to two blocks which are glued to the inside of each panel.  The crossmembers are bolted together at their point of intersection.

The two holes in the vent were an afterthought due to poor planning.  They allow access to the two screws in the glued block.  The vent should have been placed higher.

 

 

 

 

Photo 4

The crossover, obviously.  Being labeled as A and B is to correspond to the drivers used in each enclosure.  There are slight differences in component values to do slight differences in speaker impedances. Coils were adjusted and capacitors were chosen from dozens using a capacitance meter with a 1% accuracy.  This method proved better than using 1% capacitors as exact values weren't available.  Using a set of smaller value capacitors of 1% tolerance wired in parallel proved too expensive.

 

 

 

 

 

 

Photo 5

View in my living room from behind the blue chair.  There is no wall behind the blue chair.  It opens to the dining room.

The dimensions of the combined rooms are 37'x15.5'.  The center speaker system is that which was built in 1957 by a half-brother.  The bass reflex is 9.8 ft^3

The three speakers are Wharfedales,  W15FS woofer, W10FS/B mid-range, Super 3 tweeter

 

 

 

 

Theoretical results from Bas Box Pro ver 6.

Fig. 1

 

Fig. 2

This shows the bass to be -5dB @ 30hz although that's under anechoic conditions.  The room takes over below about 250hz as can be seen from comparison of figs 8A or 8B with figs 9 and 10.  The room influences all frequencies but its influence on then higher frequencies is controllable, whereas control of low frequencies is extremely difficult, if possible.

 

Fig. 3

This shows that in this alignment, the power handling capacity of the woofer falls below 55hz, reaching a low point of 25w at 30hz.  This power handling is determined by the diaphragm's reaching its Xmax.  However, looking at the next graph explains this.

 

Fig. 4

Despite the 25w handling at 30hz, the system will produce an SPL of 105dB at that frequency.  Again, this can vary immensely by the room.  Fig 8A shows 87dB at 1 watt in that room.  At 25 watts, by calculation, that's 100.93dB (101dB) at 30hz and that enough to make the room shake, rattle and possibly roll.

 

 

 

 

Fig. 5

Woofer impedance with impedance eq. circuit.  This not only removed the high impedance peak at 47hz as intended but also brought the woofer impedance to 6.6 W from 8.5 W at 300hz.  This has to be considered when designing the crossover filters although the difference is small.  

 

 

 

 

 

Fig. 6

Crossover measured with resistive loads, 8, 8 and 4 W for W, M, T, resp.  The woofer uses a first order low pass; the mid-range uses a second order bandpass and the tweeter uses a second order high pass.  The first order low pass of the woofer gave a flatter response around 300hz than a second order.  This resulted in reversing the woofer polarity with respect to the mid-range and tweeter, which are wired in phase.  Various polarity configurations were tried.

 

 

Fig. 7

1w1m gated (quasi-anechoic)

The dip after 15khz could have been remedied by tweaking the crossover point between the mid-range and tweeter but that would have taken more time than it was worth. See Fig.11

It was later noticed that by turning the treble control to about 2 o'clock, this drop was eliminated.

Compare this to the curve in fig.8B

 

 

 

 

Fig. 8A

Blue chair about 9 feet (3m) from each speaker with a power transfer of about 1 watt.  The result shown was obtained with a 31 band 1/3 octave graphic equalizer.

Black is 1/6th octave smoothed. The cause of the dip around 15khz may be due to interference between the pair of speakers.  That it could be caused by the rolloff after 15khz (fig.7) is unlikely because of the rise thereafter.

Fig. 8B shows each speaker measured under the same conditions as in Fig.8A (green & blue traces)

 

 

Fig. 8B

Green = left; Blue = right; RED = both.

The green and blue traces reflect the result above 10khz shown in Fig.7.  The red trace shows the dip at 15khz only when both speakers are active.

  All three traces are 1/6th octave smoothed and are NOT equalized..  By comparing this curve to that of fig.7, one can see the effect of the room

 

 

 

Fig. 9

Woofer near field measured in the rectangular box.  Here, the speaker and the vent are facing the floor.

The response was intentionally stopped at 2khz.

Fig. 10

Woofer near field measured in the octagonal tower. Here, the speaker faces the floor but the vent is in the rear about 2 feet above the floor.

The CLIO seems to do a full 20hz-20khz.  If there's a way to limit this, I have yet to find it.

Green = left speaker; RED = right speaker.  The difference most likely is due to their placement.  See photo 5.

 

 

Fig. 11

The original schematic.  Note that the second order low pass section has been changed to first order in revision H (green).  Each filter design was checked using a gated sweep at 1w/1m.

The attenuation resistors were determined the easy way by using an L-pad and then measuring the two resistances.  Further tweaking the low pass mid section and the high pass tweeter section may have resulted in a flatter response but this was considered to be treading into the realm of diminishing returns considering the time involved recalculating 6 components. See Fig. 7.

 

 

Fig. 12

A cleaner view.  The chicken scratch at the left is the actual measured values of the three capacitors making up the 654uF capacitor versus the printed values.  The actual values sum to 648uF.  The printed values of 300, 220 and 68 sum to 608uF which places the required value within the 10% tolerance of the capacitors but that didn't come close to the purple curve of fig.5.

These are non-polarized electrolytics with a 10% tolerance. Adding another 6uF capacitor was pointless as the 648uF value obtained is less than 1% (0.9%) of the required 654uF and besides, I didn't have a pair of 6uF caps available and making up 6uF with a pair of 3's or a 4 plus 2 wasn't even considered..  Using 1% capacitors in this eq circuit wasn't cost effective.

 

 

 

Back to the loudspeaker main page