Cabinet      Crossover      Measurements      Crossover Layout           

I always liked the SP38/13 and seeing the large ScanSpeak D3806/8200 dome getting out of stock here and there, John/US came in handy with a pair of Accuton C244-8 inverted domes and a pair of HIQUPHON OWI tweeters as well, thus the SP44 project was on track. We could also call it Ekta-Accu, but SP44 it will be. Comparing SP44 to the Ekta is obvious and despite not having them side by side, I'm afraid the SP44 will beat the Ekta from a less coloured upper mid/lower treble. These ceramic domes are something - and I find the 12M/4631G00 at 229 € and the C44 at 235 €, so price is not an option here.
The total kit will be something like this: 2 x 18W8531 = 370 €, 2 x C44 = 470 €, 2 x OWI = 210 EUR, crossover super cap version = 320 EUR, in total: 1370 € plus cabinet and damping materials. Not cheap!

Now, what can the C44 do the D3806 can't? Well, it goes deeper and it goes higher and it has a remarkable flat response all up to 20 kHz where a serious cone break-up occurs. 20 kHz is really high and nothing to worry about in a 3-way system. The C44 is not particularly sensitive, around 85-86 dB/2.8 volts - but all the same a good match to the 18W/8531 driver that usually can be tuned to a system sensitivity of 86 dB/2.8 volts. Looking at C44 horizontal dispersion, 6 kHz seems like a good starting point for crossing over to the tweeter, thus the C44 is really able to handle all of the important treble range, because we can go even lower compared to the D3806. From simulation 900 Hz looks ideal, but no guarantee this will also sonically make the best transition to the 18W midbass. Has to be tried.
Last but not least, the C44 has the most clean cumulative spectral decay I can remember, i.e. distortion is really low.

The 18W/8531 and OWI drivers are well known from numerous constructions, so not much to add here. 8531 still the best 6-7" midbass I can think of and the OWI is the 3/4" dome where everything seems to have been done right!

We need some 28-32 liter volume to please the 18W driver the best. From 32 liter we get an F3 = 39 Hz. Not many 6-7" drivers can do this. The price to pay is sensitivity being on the low side. Vent tuning is a never ending topic and anything from 28-38 Hz may do depending on taste. The higher we go, the more response we get in the 50-100 Hz range, where 28 Hz renders an almost flat response. 28 Hz is not all that easy to achieve as we need a long port in a 32 liter net volume cabinet. 70 x 290 mm - or 60 x 210 mm, the latter actually OK if we flare the vent to reduce port noise.

So, why should you possibly try a design like this?

1. Well, first of all enjoy the world of diy loudspeakers and enjoy something you'll have to pay at least 6,000-8,000 US $ for over the counter. Finished speakers with the current set-up come with serious price tags.
Mostly thanks to the C44, this speaker will merciless tell you if you amps and musical sources are less than - well, really good. I do not have a state of the art solid state power amp, but my tweaked Rotel RB981 at least has the power to make the 8531 bass drivers going. Do not run this speaker on low-wattage anything. The 8531 bass driver can go deep for the size like no other bass driver, but it likes lots of amperes to perform its best. I ran the SP44 from my 20 wpc AM amps and the sound is great - much better than the Rotel, but bass grip is lacking as soon as I crank up volume.

2. Next you may build this speaker because you want deep bass from a small footprint and an overall modest cabinet volume. The price to pay for this is low efficiency. Removing my DTQWTs, having been on the stand for long, was a bit shocking. Our livingroom seemed almost empty and going from a total bass radiating area of something like 1850 cm^2 to a merely 300 cm^2 for the two 8531 drivers calls for some mental preparation as well. I won't even try to describe the difference in sound, but it is huge.

3. Last but not least: To try out the ceramic drivers and hear what lack of colouration does to your music. The C44s are truly something as stated elsewhere on this page.

Cabinet

Test cabs

Facetting the front panels. Use 1) chisel, 2) planer and 3) sandpaper. Takes a few minutes from MDF.

Right: Test set-up for initial measurements to generate data files for LspCAD simulation.

Initial measurements

Left: SPL from drivers in cabinet, no crossover. Red = 8531, green = OWI, blue = C44.
Right: Impedance of drivers in cabinet. Red = OWI, blue = C44, green = 8531.

C244-8/C50-8-044 driver

Left: Horizontal dispersion, 0-10-20-30 deg. Right: Impulse response.
Horizontal dispersion suggests point of crossover should be placed around 6 kHz between mid and tweeter.

Left: Step response. Right: ETC Plot.

Cumulative Spectral Decay, C44 driver at 20 dB and 40 dB.
(I don't think I've seen anything this excellent before)

The Crossover - Simulation
version #2 (#1 not shown)

Crossover version #1 for SP44.

Measurements and crossover fine-tuning

Left: After crossover assembly, the bass and mid turned out pretty well and inverting mid polarity produced the wanted deep suck-out, indicating point of crossover around 900 Hz. All as predicted from LspCAD simulation. The RC circuit across L2041 in the mid crossover was omitted.
Right: The OWI is a quite capable driver, allowing fairly low points of crossover, so it takes serious measures to get it down at 6 kHz, thus a 4th order filter to make proper summation and phase-traking. Inverting tweeter polarity produced the wanted deep suck-out between mid and tweeter (yellow). Now, this almost looks too easy, but the trick here is to predict (measure) the acoustic distance from drivers to the listening point before doing the simulation. Based on measurements and experience, a few millimeters have to added to the mid and bass distance before starting simulation.

Left: Final response from crossover V1. Blue = minimum phase. Rather flat I dare say. Not often I tune a speaker this flat.
Right: The litmus test: Reverse mid polarity.
The suck-out at 6 kHz can really look like anything depending on microphone placement, so this is only show-off.

Left: Horizontal dispersion at 0, 10, 20, 30 mand 40 deg. An even power response up to 10 kHz is seen over a very wide listening area.
Right: Vertical dispersion measured a 1 meter distance. Red = 10 cm above tweeter. Blue = tweeter height. Green = mid height. Yellow = between bass and mid. Purple = at bass height.
Now, making a point of crossover at 6 kHz is calling for trouble with regard to crossover lobing and this is no exception. The good thing is that 6 kHz is really high and not representative of what we actually hear, as our two ears (fortunately) do not work the same way as a 5 mm microphone capsule.

Crossover - Simulation
version #3

The sound from crossover version #2 was indeed very promising. Lots of detail and transparency, but also a distinct character from the C44 working up to 6 kHz. A sound very hard to describe. Not really fullrange kind of sound, but slightly "cuppy", i.e. like when you hold your hands around your mouth and speak. This doesn't appear to be derived from a too high upper mid level, rather limited dispersion, which is strange as the horizontal dispersion is excellent. But measurements are one thing, not always reflecting what we aurally perceive. At 6 kHz we're close to where the C44 will start beaming due to the diameter of the cone and maybe we're just pushing our luck from taking a 2" dome this high.
Back to LspCAD, targetting 5 kHz point of crossover. What appeared from this exercise was an almost perfect 4th order LR crossover as can be seen from automatic optimisation below. Not often this is possible.

Crossover version #3. Only L2041 was changed during crossover optimisation from 0.68 mH to 0.56 mH to smooth 4-8 kHz range.

Mid and tweeter attenuation

Depending on your overall front end and room acoustics, you may want to
balance upper mid and treble. I do not forsee you want to decrease mid and
tweeter attenuation, but if you find the sound too forward, try
2R7 or 3R3 to the mid and 5R6 or 6R8 to the tweeter.

The current tuning is rather flat and you'll merciless be told how bad your - bad - CDs are.

4th order LR crossover simulation. Points of crossover at 900 Hz and 5000 Hz.

Upper left: Predicted response of drivers driven from crossover and summed response (black).
Upper right: Phase tracking.
Lower left: Final system impedance. Minimum impedance = 4 ohms at 400 Hz.

Measurements on crossover #3, 4th order LR

Left: Bass low-pass section and mid low- and high-pass sections seems to be in place.
Right: Tweeter and mid appear to join hands at 5 kHz as predicted.

Left: Actual point of crossover between mid and tweeter is 5200 Hz.
Right: Final response from V3 crossover. Blue = minimum phase.

Now, this all seems pretty straight forward, but how about the sound? Well, just what I had hoped for. The slight cuppy sound from the inverted dome is gone and after listening to a few vinyls and CDs, I can't help feeling this is it! Very enjoyable indeed. The 8531 - as always - delivers firm and deep bass - that is, if driven from not less than 100 watts solid state amplifier. The upper mid/lower treble? Never heard it better. These ceramic cones are something. Clean, clean, clean. Treble? The HIQUPHON is totally cool in this environment. Doing treble from 5 kHz is surely a piece of cake for this magnificient tweeter.

Some comments on crossover components.
The drivers alone for this kit is a little above 1,000 EUR and if you plan to do the complete crossover from standard polyprop capacitors, I suggest finding another construction because you'll never render the full quality of the C44 and HIQUPHON from less than the very best components. The 22 uF and 68 uF caps in parallel with the bass driver can be standard PP, but for the mid and tweeter I suggest high-resolution super caps, i.e. caps made from two series connected capacitors, e.g. Jantzen Audio Superior Z-caps (~180 EUR), Mundorf MCAP Supreme (285 EUR) or AUDYN-CAP Plus (160 EUR).
Use proper wound coils with non-ferrit core for the bass driver. 1.2 mm wire gets DCR really low. Make sure the coils are baked. For mid and tweeter, use air cored baked coils, 1 mm wire for mid coils and 0.7-0.8 mm wire for tweeter coils.

Complete crossover kit available from Jantzen Audio: contact@jantzen-audio.com

Crossover layout:

Bass low-pass section crossover layout.

Mid and tweeter crossover section layout.
Layout is shown for Superior Z-cap version. This should do for most super caps.
Standard PP caps may require a smaller board.