Trial and user error!

Two hours wasted on REW later and I was getting nowhere. I must have run 100 sweeps and still had no improvement. Thankfully, the limited number of sweeps before one has to delete earlier sweeps has gone. Thank you for that priceless improvement to your free software.

At one point I had a perfectly straight and level response from below 30 to over 110Hz. Then I made one small adjustment on the BFD and never got that flat curve back. So I ended the session without having achieved anything useful.

The following morning and I decided to reset the BFD to factory settings for the first time. I was getting odd effects which seemed intractable to any change. The reset produced a number of flashing red lights and a complete cancellation of all preset filters.

I soon found that I had been changing one IB set while imagining I was filtering the other. Quite inexplicable but true. I resolved to write down the channels affected by each box in the equipment chain. I had reversed the speaker cables from the EP2500 power amp since changing the position of the driver sets.

Behringer mixer, CX2310 and BFD (from top to bottom) 
It is usually too dark to capture the rack components without flash. 
So most of the usual lights are invisible in my pictures.

Behringer do not make it any easier by calling a channel one or right at random on their equipment. Of course the way the boxes are wired together will change which channel is right or one or two or left. It turned out the BFD was now reversed compared to all the others. One set of four 15" drivers is on each channel. The BFD and mixer affect only the bass channels. The speaker channels do not pass through either.

With two channels to play with REW is best run with stereo (or dual mono) connections. I use RCA/phono to stereo mini-jack, solid adaptors on the twin core, cable connections. This ensures that both sets of drivers can be tested together. Or one set alone, simply by muting one set, or not, on the CX2310 active crossover.

This is the result of another two hours of REW sweeping:

The bass is around 10dB hot but can be easily reduced on the active crossover gain controls to taste. Black line is the full combined IB with main speakers. Blue line is the four newer drivers. Violet the older four.

It looks quite a mess but will be further refined later. I now prefer to have a gentle hump between 30 and 60Hz to get some power and impact. My decades of chasing the infrasonics at the expense of higher bass frequencies are over. The IB has more than enough infrasonic power to wreck the house without worrying about silly house curves.

The peculiar, signature hump at 13Hz, with the associated higher frequency trough, is a known room effect. The older IB15 drivers are still getting their customary 20Hz, +16dB boost with 120 B/W. The newer drivers about 3dB boost at 20Hz. I also gave them a gentle lift in a daisy chain of gentle boost filters from 30-60Hz. Without which they follow a steep downward slope with rising frequency.

I tried every imaginable crossover point but nothing will affect the trough at 170Hz. This is obviously another fixed aspect of my system layout relative to the room. Previous testing shows it to be a cancellation effect between the IB in the left wall and the right, main speaker.

The sound quality is back to normal as I listen to my favourite organ CDs at fairly low levels. The clarity of the bass, as it ponders up and down in the acoustic basement, is exquisite. Even at the threshold of audibility the timbrel character of the great pipes is effortlessly portrayed.

Click on any image for an enlargement.


An IB rebirth?

After further REW testing and listening tests I decided the OB should be removed. It's humped response was too extreme to be useful in filling the 150Hz trough. The pair of older drivers, borrowed temporarily from the IB, were to be returned to the 6' tall, eight driver manifold.

While I was about it decided to move the four newer drivers down to the bottom four positions. The older drivers were moved up to the top four positions. I haven't tried this layout before. Logic suggests that the lower positions will enjoy more bass reinforcement. The upper drivers less so. Though this may be pure speculation given the manifold's position: Almost reaching the narrow ceiling. Just under the 45 degree overhang of the sloping attic walls.

The changeover of eight 15" drivers is not something to be undertaken lightly. Sixty four furniture screws and 64 T-nuts have to be removed and then refitted. One hundred and twenty eight loosening and fastening actions in all. Then there is the matter of the weather. If it is cold outside there is a hell of draught though each empty driver cut-out.

This time, I finally made a bit to drive the screws with a rechargeable drill. I simply cut off an Allen key close to the bend using a thin, diamond wheel in the angle grinder. Allen keys are too hard for hacksaws which just skate off.
I can't be both sides of the manifold at the same time. So I have developed a system. I leave a middle driver out on each side of the tall box. This allows me to reach through the empty cut-outs. So I can  fit the T-nuts to the inner ends of the screws inside the box.

Even this is not without difficulty. Because the screws are loose in their holes. So they want to push straight back out as I apply the T-nuts to the bare threads inside. I usually hold the screw thread with a finger nail while gently starting the nut. Once the threading is started the nut can be spun on until the locking spikes meet the inside of the plywood box.

The first screw to fit is always the bottom one on a driver fitted from the outside of the box. The heavy magnet inside the box dictates this order. The magnet wants to pull the back of the driver downwards. So the rim pushes out at the bottom. This is the complete opposite to my fuzzy logic. The first screw is pushed through the driver rim. That driver must now be held firmly against the baffle until the first T-nut is safely fitted.

This is where the missing drivers are helpful. All it takes is the bottom screw and T-nut to hold the driver perfectly safely. The basket sits in the cut-out so cannot move sideways. Now one can relax and work on all the other drivers. The final driver to be fitted is the real problem. One can't reach through an empty cut-out to fit the bottom T-nut.

My method is to push as many screws through the last driver as possible. To hold the driver in place by friction alone. I can then go around to the inside and fit as many T-nuts as possible. This ensures the driver won't suddenly slip. At least one screw will usually hold it safely in place. Once the first T-nut has a grip the battle is almost won.

Then I have only another 56 T-nuts left to fit. This usually leaves my finger tips rather sore. Even if I avoid the anchor spikes the stamped disks of the T-nuts are narrow and sharp. I never remember to have a tray to hold all the nuts and screws. Inevitably T- nuts drop as screws are removed or pushed through the holes. I always seem to end up with a nut or two missing.

Discipline demands that they must be found before they can wreck a driver! Usually they are hiding around the back of a driver magnet out of sight. The worst possible scenario is when a T-nut drops right inside the motor and coil area. There are small vent holes in the cast baskets which are just large enough for a T-nut to fall in and be held firmly by the magnetic field. Sometime a nut just sticks to the spider face by magnetism. These can be more easily rescued with extended fingers through the basket spokes. Those which make it inside the motor area need long nosed pliers, patience and some dexterity to get them out again.

One thing I noticed this time is that the older AEIB15 drivers are much heavier than the new ones. The older magnet back plates are much larger in diameter. As are the ferrite magnets. The baskets look almost identical.

The newer cones are flared, reinforced paper. The older ones are straight tapered and vinyl or polypropylene. Or whatever plastic was used in their manufacture. The older drivers have screw terminals. The newer ones are sprung. The sprung terminals will not take such large wire as the screw type. This isn't a problem with a single wire. Only when two wires must be twisted together in one terminal does it become a big problem. Unfortunately most terminals will have two wires inserted.

Now all I have to do is remember how to wire up the drivers again. The older drivers are 4 Ohms each. These will be wired in series-parallel for a total load of 4 Ohms on one channel of the EP2500 power amp. The newer drivers are 8 Ohms each. So will be wired all in parallel for a combined load of only 2 Ohms on the other channel. As usual, I will apply a 9 Volt battery to the amplifier end of the bared cables on each set. Just to ensure all eight cones jump outwards at the same time. I usually test the DC resistance as well with a DMM to ensure the combined loads are correct.

It will be interesting to see how moving the drivers around will affect the response and thus the BFD filters required. It could be a total disaster but was still worth trying. I hate "if-only I had tried that" situations. Usually the thought of removing eight heavy drivers and all those screws and nuts leads to sudden and permanent inertia! :-)

Click on any image for an enlargement.


H-frame OB


In my ongoing search for more impact I had imagined a rather convoluted idea for a large Open baffle. One which would use the AV stage as a horizontal baffle. A large slot made at the back of the stage would allow the array sub-baffle to breathe down through the stage floor to the room below. The slot would have to be of adequate size to allow free air flow from the rear of the four 15" drivers. Or the baffle would not function as intended.

Cancellation would only occur when the opposite phase, pressure waves met 5' away from the slot. At the open stairwell in front of my seat.

I quite liked this idea because it offered a reasonable size of baffle without the usual IB heat loss to the great outdoors. Not quite infinite but there was still hope of true Open Baffle sound quality with a very well extended response.

Then ThomasW, the vastly experienced guru of the IB Cult forum, suggested an H-frame OB instead.

I had no idea OBs could reach very low until I started doing my OB homework online. It seems the H-frame design loads the driver  cone(s). Pushing the free air resonant frequency downwards. All I wanted to do was fill the deep trough between 150 and 200Hz. Infrasonic response was not necessary.

A couple of hours of to-ing and fro-ing to the shed provided a working model. I borrowed two cut-down baffles from former IB projects. Then added 4" x 1" rough sawn plank frames front and back to hold two baffles together. This made a solid, but butt ugly, 2 x 15" H-frame OB to play with.

The narrow frames should help push the baffle frequency up out of harm's way. There was absolutely no point in investing quality materials in an experimental model. I had a beech strip, kitchen worktop waiting in the wings if something prettier was found worthwhile. The test subject was placed behind and midway between the right main speaker and the LCD TV. 

For test purposes I used the SVS Bash amp from my old 16-46PCI/NSD cylinder. I have no idea whatsoever if this amp has any response shaping. After series wiring and checking for polarity with a 9 V battery I ran a few REW sweeps through two of the older AEIB15 drivers sitting in the baffle. These drivers are vinyl coned with a measured Fs of 32Hz. The rest of the parameters of these non-spec drivers are likely to remain a total mystery.

This is the result of setting the SVS 12dB/octave crossover to its 120Hz maximum. Measured at the listening position with the Galaxy 140 SPL meter. No EQ is possible using this amp unless I insert the BFD in the signal leads. There is really no point in this when I can try the EP2500/BFD/CX2310 tomorrow. Though I seriously doubt I need remotely so much power. If I can pull down that wide hump from 30-50Hz, with the BFD, then that peak at 150Hz could be a useful trough filler. 

 The waterfall graph of the above curve. While the rest of the curve is tidy, 50Hz is real mess!

Now the result of bypassing the crossover on the SVS Bash amp. Rather inexplicably the hump has narrowed and a deep trough bottomed out at 90Hz. The complete opposite of what would be expected. Had I simply swapped the graphs, one could be easily forgiven for believing the curves matched the original descriptions. Very strange indeed!

Note how I have stretched the graph limits to 300Hz to capture the 235Hz baffle resonance. Quite a remarkably high figure for a baffle large enough to house two 15" drivers. Commercial designs usually use a couple of 10" drivers in an H-frame for the bass section of full range OBs like the Orion. The best designs use notch filtration to remove the peak.

As can be seen in this waterfall graph an OB is rather well damped. This is despite the awful response curve. There is some ringing at 50Hz which may be electrical rather than mechanical. I have noticed a 50Hz hump in quite a number of my REW waterfall traces.

I tried playing a Mike Oldfield CD through the OB and brick-sized Sandberg computer speakers. I was immediately impressed with the power of the OB. It seems an OB can be quite efficient. I was measuring up around 100dB(C) with the SPL meter at 1 metre with the SVS gain at halfway. Cone excursion was still quite small. The "jangly" sound I had experienced with small, naked baffles in an earlier trial was thankfully absent.

I may try a nearfield sweep tomorrow to see how well these drivers behave. It might prove whether the SVS amp has any response shaping. The OB sounds quite unlike the IB. With a lot more upper bass present. Hardly surprising really. When a 120Hz or no crossover at all is superimposed on the response. The IB has a nominal 80Hz roll off. Could SVS have applied a notch filter to remove the natural pipe resonances from the cylinder?

As promised: A series of nearfield sweeps about 4" from the upper driver dustcap. The blue curve is with the 120Hz crossover engaged. The highest curves were all nearfield without the crossover switched in. All with good agreement between them. Only the green curve is taken at the listening position with the 120Hz crossover engaged. The unfiltered nearfield curves suggest these speakers are fine and the SVS amp as flat as matters. The 30-50Hz hump may be the drivers natural free air resonance being damped by the baffle.

The SVS amp is incredibly sensitive to gain level with nearfield measurements. Merely touching the knob would change the level by 20dB. Most of the nearfield sweeps were all but inaudible. Later I matched the resulting curves at 80dB for easy comparison.

There must be some hum at 50Hz. The Creative "Live" Sound card sockets are worn out. Even the weight of the mini-jacks and very thin cables is enough to cause a break in the stereo signal when listening to the computer speakers.  

Click on image for an enlargement.
Pressing Ctrl and + simultaneously will enlarge the images and text further if desired.