The purpose of this study is to compare the difference between a conventional vented enclosure using a cylindrical vent and the single port RJ design on bass response. Vertical rectangular and elliptical slots/ports were tried but neither had any difference from the RJ slanted parallelogram seen in Photo 11 and are not further discussed here.
The original design used a vent opening of odd shape, kind of like a truncated parallelogram rotated 45o clockwise. The area of this shape is the same as the area of the internal vent, about 27 sq.in. (26.625)
The vent of 27 sq.in. was tested using Bass Box Pro as well as good old fashioned equations used years ago. Both were co-incident which was expected as Bas Box, in all probability, uses the same equations.
The 27 sq. in. vent is derived from the perimeter (16.75" on a side) of the internal square baffle and its distance from the exterior baffle, 0.75 inch. Since this inner baffle is square, two opposite sides were sealed. Square was easier to cut and also since it's bolted to the outer panel, made changing the front assembly easier. It would also ensure a better seal than if it were cut to fit inside.
RJ8 pdf file aka The Original Paper (10 pages). http://downloads.bbc.co.uk/rd/pubs/reports/1955-08.pdf
Some Construction Piccies
The cabinet is only screwed together as this is a one time study.
|In the original report,
mention was made of the dip around 1khz being caused by the internal
front to rear distance of the box. It was also mentioned that
additional padding had little effect. So, this idea was put to
test. More on this later.
This pyramid is mounted on the inside of the rear of the box. Response sweeps were run with and without the pyramid and padding. First, neither had much effect on the response and there never was a dip at 1khz observed with or without the pyramid and padding. There is, however, a wide rise in response between 2400hz and 4200hz but that's blamed on the speaker, not the cabinet.
|PHOTO 2||PHOTO 3||
The regular vent design. The speaker is mounted to the front panel, unlike the RJ design. This was done for comparison between the two designs. The vent is normalized for a tunnel length of 1/2 inch, the panel thickness.
The vent area is 27 sq.in. Had the front panel been 3/4", the vent area would be slightly larger, about 30 sq.in.
The pyramid and padding are inside.
The layers were rotated to introduce some irregularity between the front and rear. However, it can be said that this irregularity is also regular.
It seems oxy-moronic.
If nothing else, it might add a slight twist to the music.
This shows the padding. The pyramid is behind it.
This is just to avoid trying to describe how much padding was used and how it was located.
All three of these speakers are original.
That includes the foam on the FS unit which surprised me as it hasn't rotted. It was actually driven to about +/- 3mm without tearing.
The two upper units were used for comparison of Fs with the RJ8/AL (Figure 12)
Top Left: Super 8 CS/AL
Top Right: Super 8 FS/AL
Bottom: The RJ8/AL
Obviously the same three suspects.
The exterior panel is bolted to the smaller baffle, separated by 3/4 inch.
On the right, the bolted pair.
Two opposite sides were later blocked so as to have the required vent inlet area of 26.625 sq.in. (~27 in^2)
The truncated parallelogram opening.
Responses were also made near field to the cone as well as the lower left right angle corner. This was done between 20hz and 350hz since the LMS gating becomes erratic due to reflection distances. The microphone here is at a distance of 1m from the baffle.
This vent has an area of 25.5 sq. in., slightly smaller than the internal vent area of 26.625 sq. in., the difference being attributed to measurement and/or cutting variables. The difference is about 4%, not worthy of correction.
Regular vented Enclosure
1a(left) & 1b(right)
Regular vented Enclosure
Two alignments as seen by Bass Box Pro for the Wharfedale RJ8/AL.
On the left (RED) is the optimum vented alignment and on the right (ORANGE) is the alignment for a 2.9 ft^3 enclosure. It should be noted here that the vent of the smaller enclosure is a regular vent normalized for a tunnel length of half an inch, the panel thickness.
Neither of these alignments use the vent design of the RJ enclosure. Bass Box Pro doesn't make allowances for such a vent design as used in the RJ enclosure.
Expected response normalized to 0dB. used primarily for response comparisons among different cabinets. These two curves are those of the above two alignments.
Impedance curves of eight Super 8 units. The units measured in #19 and #37 are original assemblies.
Initially, the 78hz resonance of the RJ8 (not in this set) was thought to be high but this seems to be around the same as all the cloth surround 8" units I have. Only the foam surround unit(s) are lower. That may be another reason for changing to cloth (felt) surrounds, the other being foam's tendency to rot.
These Wharfedale 8 inch units are intended for mid-range or low level full range; they were not classified as woofers.
A 1957 data sheet specifies two 8" units having an Fs of 70-85hz and three foam surround units as having an Fs of 45-65hz. The other two were not described as having cloth or corrugated surrounds.
The units here with an Fs above 100hz were modified as such to improve high frequency transient response. This was done by stiffening the spider, not the annulus (surround). The stiffness of the spider should be equal to or greater than that of the annulus. If not, the cone can buckle when driven against the stiffer end.
(REF: PHOTOS 8 and 9).
This pair, of which there are 4 units in all, are old Jensen commercial units intended for background and public address in small auditoriums. The transformers are not connected here. All 4 speakers are new old stock. The reason they are shown here is that they have very similar cones to that of the RJ8 and their resonance is also close. These two have an Fs between 85hz and 95hz. The RJ8 has an Fs of 79hz. It is lower than these with very similar cones probably because these two ate later vintage and never used. The RJ8 in my possession shows signs of extended use as well as being about 15 years older.
These are response curves of three Super 8's taken at different drive levels for reasons not important here. What is important is the shape of the curves. All units are either FS or CS, foam or cloth (felt) surround.
Note that there is no severe bump between 2400hz & 4200hz as seen in FIGURE 14. (next)
It is this observation that led to the conclusion that the aforementioned peak is characteristic of the RJ8/AL and not its enclosure.
|In the following figures, all curves labeled as [Gnn + Gnn] are the summation of a near field responses at the cone and the vent. The dB adjustment is to bring the summed pair to the level of the others for easier comparison. Attempts to adjust the power levels to make this adjustment proved too tedious so the power level was adjusted to bring the SPL within the limit of the microphone and the final adjustment was done with the summation algorithm in the LMS. All such algorithms in LMS are done on the board, not with software loaded onto the hard drive. This makes the algorithms run faster, independent of microprocessor speed.|
All these curves were run with the RJ slot vent (parallelogram). Photo 11
Curve 60 (red) is the left part of curve 55 below 350hz and is derived by summing the near field responses of the cone and vent and adjusting the magnitude. The reason for magnitude adjustment is due to lowering the power to the speaker during near field measurements due to the proximity of the mic to the source; no attention was given to actual sensitivity. The main purpose was the shape of the curve to see the possible differences among the vents.
The green curves are those of the Jensen speakers, SEE PHOTO 14. Both these green curves were run with the smaller truncated parallelogram vent. Low frequency response below 350hz was not run on the Jensens. The purpose of their presence is to show the high frequency response differences.
Compare the red curve with those seen in Figure 15a
This red curve (fig 14) is also the same as the red curve in fig 15b, although run several days earlier.
All curves above 350hz were run at 1w1m and gated.
Notes on Figure 14 too lengthy to fit the left side of the figure.
Regardless of the vents used, the peak between 2400hz and 4200hz persists. This peak seems due to the nature of the speaker, whatever that may be. An aged cone was a likely suspect but all the 8 inch units here do not have that peak.
Another possibility is the annulus being stiffer than the spider. If so, the cone will buckle when driven out. It is noteworthy that the length of the cone from apex to annulus is about 3 inches, the length of half a wavelength at 2400hz and a full wavelength at 4800hz Note also that at half a wavelength of 4400hz, 8800hz, there is a small peak. This seems more than coincidence and is worth investigating. Fortunately, there's a pair of Wharfedale tens here that have easily removable cone and coil assemblies, like some Tannoys. Wharfedale W10FS There are also a few or more cones available as well as hand wound voice coils. The band in which the peak occurs will be lower but should still be present if the annulus is stiffened more than the spider.
The results obtained in the green curves, (Jensens) was surprising. A similar peak was expected considering the stark similarity of the cone material and the corrugated annulus. By the way, those Jensens have dual winding voice coils, hence the dual 25 volt transformers. responses were run with the coils in series, parallel and one coil. While amplitude differences of about 3 dB were observed, (the amplifier output wasn't altered), the curves were the same if power was set to 1 watt.
The light green curve is that of the unit without the whizzer. It doesn't have that wide peak. The bump at about 3400hz is not considered a similarity in any way. The result seen in the dark green curve, the unit with the whizzer was somewhat expected but not to the degree seen. While it is fraught with resonances, it is linear within about 5dB from about 900hz all the way to 15khz. Perhaps Mr. Briggs should have considered a whizzer on the RJ8.
The conclusion of this is that the RJ vent was an idea worth investigating but unfortunately, it didn't make the Top 40. That may be due to the complexity of the double parallel front panels which requires more work than just drilling a hole and inserting a tube of appropriate length. That is not to say it wouldn't sound nice as studies such as this will show how a speaker will perform but give little to no idea on how it sounds. That last statement is reminiscent of what we, as teenagers, did to cars. A larger engine and the right choice of transmission and differential gear ratios would guarantee a faster acceleration, higher top speed but no one ever considered how quickly the car could stop or how it would handle on curves.
Comparison between the cylindrical (conventional) vent and the RJ slot
(slanted truncated parallelogram) vent
FIGURE 15a RJ8/AL in three variations of the cylindrical vent enclosure.
All are measured at 1w1m gated. Near field and vent response below 350hz.
Black: No internal padding and no pyramid. Pink: Moderate padding (see photo 7) Blue: Moderate padding and pyramid.
All three show the rise between 2400hz and 4200hz albeit slightly altered.
Using the cylindrical vent (regular/conventional) was to show the difference between that and the RJ vent design, next.
FIGURE 15b RJ8/AL in four variations of the slot/slanted truncated parallelogram vent enclosure.
The dip between 1000hz and 2400hz is due to the effect of the dual front panels and the cavity between them.
The peak between 2400hz and 4200hz has changed a little. This peak is, in all probability, caused by the speaker. Note the similarity of this peak with that in Fig. 15a
Another speaker was tried in this enclosure, a Dayton DC200-8. The results follow.
Dayton DC200-8 in the RJ enclosure
Before it is neglected, this speaker is still available at Parts Express. All the data on that unit can be found at the following link.
|The urge to try this design with a
modern woofer was irresistible. The idea of firing a full range
through two baffles spaced 3/4 inch apart wasn't appealing due to the
potential cavity resonances at higher frequencies.
Since the original idea of this design was to enhance the bottom end, using a modern woofer seemed like a good idea. see figure 18 (below).
Regular vented enclosure, volume 2.9 ft^3 with 3"D by 8"L vent
16(left) FIGURE 17(right)
This speaker is a woofer, not a full range. In fact, it's not recommended for use about about 800hz. As one fella who commented on P/E, it gets nasty in the mid-range. The reason should be obvious in the response curve, figure 18. The response falls after 1khz and has a nasty peak at 2khz with further roll-off beyond.due mainly to voice coil inductance and cone mass. An inductance of 1.5mh with an 8 ohms load will be 3dB down around 900hz, first order filter.
The Normalized Response Curve here is purely theoretical in that it probably represents that which is obtained under anechoic conditions.
The data on the left, figure 15, shows an optimum high fidelity cabinet design, assuming Xmax is not exceeded.
As implied earlier, this speaker performs better with a conventional vent mainly due to its low Fs of 32hz, a full octave below the Fs of the RJ8.
Dayton DC200-8 in regular vent and the RJ vent
Here we have three pairs of curves. The lighter shades are those of the summation of cone and vent near field responses below 350hz. They are magnitude adjusted to coincide with their respective curves. They are paired as follows.
Grey, 79, with 75, black.
Pink, 85, with 80, red.
The grey & black pair is the response of the Dayton in the RJ enclosure using a cylindrical vent 3" in diameter and 8" in length.
The pink & red pair is the same speaker in the same RJ enclosure but with the double panel RJ vent.
Admittedly, this slot vent isn't right for this speaker. If the two open sides of the internal vent were lengthened towards the rear, a vent length of 17" can be achieved but this would only need an opening of 4.11 sq.in. and that would heavily load the diaphragm. A longer such vent wrapping slightly over the rear can be made as long as 25" if the cabinet is enlarged to compensate for the volume occupied by the vent. A 25" vent would require an opening of about 5" in diameter but would still load the diaphragm not to mention it's area being only 19 sq.in., still 7 sq.in. shy of the required 27 sq.in. area at the inside of the cabinet. Further juggling of cabinet dimensions to accommodate such a vent is an exercise in futility since the speaker performs very well in 3 cubic feet with a conventional cylindrical vent.
When the difference between the responses of the cylindrical and RJ vents was noticed, it was concluded that the dip between 1000hz & 2400hz was caused by the cavity between the pair of parallel front baffles. This is mentioned on page 3 of the original paper. (ref fig.2 in that paper)
It wouldn't hurt to have that paper handy here. It's only 10 pages, 5 sheets if printed in duplex. Here's the link again. http://downloads.bbc.co.uk/rd/pubs/reports/1955-08.pdf
The aforementioned conclusion is corroborated here. Four responses were run on two IEC type baffles, actually the front baffles of the enclosure Two responses of each baffle were run, with and without fiberglass on the front to negate the effect, if noticeable, of frontal reflections. The resultant response curves are shown below in figure 19.
The photos below are of the front baffles. Showing the single baffle with fiberglass was deemed redundant.
RJ double baffle
RJ double baffle with fiberglass
The black & red curves are run on the RJ double baffle, without & with padding, respectively. Similarly for the blue & pink curves on the single baffle.
As can be seen between the pairs of pink & blue and the black & reed, the padding has little effect.
What is quite noticeable is the effect of the cavity between the two baffles of the RJ system. The peak between 2400hz and 4200hz, in my opinion, is probably due to the speaker itself, as it is some 60 years old. The cavity even tames that but has a more severe effect between 1000hz and 2400hz. Could this be the dip between 1000hz & 1500hz seen in fig.2 of the original paper? That the Misters Robbins and Joseph missed the cause is surprising. Admittedly, the relationship between the wavelength of the frequency at the dip being the same as the internal front to back distance can be misleading and result in an erroneous conclusion. The fact that the effect of the cavity would elude two acoustical engineers is even more surprising, especially after the dip was unaffected by the addition of internal padding, something that should have led them to look elsewhere.
Comparing figures 15a and 15b, one may wonder at the low frequency improvement of the RJ design. While the conventional vent has more low frequency output, there are two peaks separated by a null, which is typical of a vented enclosure.. With any luck, an equalizer may fix that but a bass control won't. The latter will boost or attenuate the low end but will have no effect on the peaks and null with respect to each other. The RJ design, however, being void of those irregularities will respond well to a bass boost without the use of an equalizer. It seems the Misters Robbins and Joseph were on the right track to achieve a better low frequency response.
Back to the Wharfedale page
Back to the loudspeaker main page