Eighteen sound 2-way fully active - Aktiivinassakka

This post is the description and construction of my loudspeaker. I have built a 2-way sealed box active speaker using Eighteen sound 10NMB420 and NSD1095N with XT1086 horn and Hypex FusionAmp FA123. The drivers are 16 Ohm versions and the box is made by professional carpenter. This build is based on well known design in my country: Ruutinassakka (gunpowder pouch), but I made it active, thus I named it Aktiivinassakka (active pouch). I'm sure there are many equivalent builds around, at least with passive xo. I'm using miniDSP SHD in front for room correction with Dirac and subwoofer integration, but the speaker specific DSP is in the plates.

I'm relatively new to speaker building, so here are a few points I discovered. I used an existing design from a reputable designer, I only made it active and used higher Ohm versions of speakers. That mid-bass driver is really a mid-range driver and it took a ton of equalizing to get it go under 100 Hz. The HF driver is attenuated circa 30 dB lower than the bass! But luckily the Hypex works wonders with its DSP and with BTL config there is more than enough power available for domestic SPL. I'm crossing them to subs (at 80-100 Hz), but that's hardly necessary for most music of my taste. The bass driver is high-passed at 60 Hz but with my settings it still plays lower and with better feel of power than my previous speaker, which is similar design, but a passive, ported box, without high-pass, with 8" bass driver from the same brand. The bass quality was unexpected, but a fortunate surprise.

The reason for 16 Ohms for the tweeter is simple: it helps to even out the sensitivity mismatch to woofer and also hopefully less noise. 16 Ohms for woofer is another story, most simply because I hope it to sound better than 8 Ohm counter part. BTL load halves the load to the amps, but the Hypex would have been happy to drive 4 Ohms. Using higher Ohm version of woofer probably undoes the sensitivity compensation of the selected tweeter. But I selected this version because I'm crossing them over to subs, so not even nearly all of available power is required and through simulation I deducted that gain and SPL will be more than enough for home use.

Description:

• Eighteen sound 2-way fully active - Aktiivinassakka
• 10" midbass + 1" compression driver on a horn
• Crossover ~1600 Hz, LR2
• Frequency response ~60-15000 Hz, in-room.
• Hypex FusionAmp FA123 plate amplifier
• Sealed cabinet. Width x Height x Depth: 30 x 57 x 35 cm. ~22 kg.

Drivers

I did read some positive reviews of these particular Eighteen Sound drivers, but I was reluctant to try them. That's because the previous speakers that I owned, were not as accurate to micro detail in the midrange that I had hoped and the highs were missing some in frequency response. Also the name "Eighteen Sound" gave me negative first impressions, like do they have anything other than 18" boomboxes for 'hoods? That combined with low-fi experiences cheapish PA systems for PA purposes had made a negative prejudice against them. But after being in contact with the designer of the Ruutinassakka and reading some more reviews, I pulled the trigger on this design.

10NMB420 and NSD1095N unboxing.

10NMB420

10NMB420

The midbass driver is gorgeous to look at. It's simply stunningly beautiful. The only minus would be the outer ring, which has the screw holes, it's ugly on the front side, but everything else simply oozes style and yells high end. Flawless build quality.

NSD1095 bolted on a XT1086 horn.

The NSD1095N has a bit older looks, but is a state of the art, flagship model of Eighteen Sound. It's not visible outside of the speaker, so no minus points for the looks. The magnet on this driver is strong. It's actually somewhat difficult to operate the Allen wrench to fasten the bolts, because the magnet pulls so heavily.

The aluminum horn rings very loudly in free air. That can be cured by tightening it loudly to the cabinet, but I also damped it with butyl mat with aluminum foil on the back side. Video here.

Damped horn.

The horn has a 80º horizontal 60º vertical dispersion pattern. Manufacturer suggests a 1500-1600 Hz xo point for this combo and at that frequency a 12" or even a 15" woofer would fit this radiation pattern better, but I'm going forward with the 10". A 2000 Hz xo point would have ~80 degree dispersion for 10" driver to match that of the horn.

My listening environment for this speaker is not a typical one. I have a somewhat large (by european/finnish standards) living room and my main listening position is at the couch, but I have my computer hooked to my TV (double duty as a computer screen) and these speakers have to also handle this near-field duty when I'm sitting at my computer doing computing stuffs. When I looked for suitable horns with wide enough dispersion to cover both of these listening positions, XT1086 came up immediately. Now that the speaker is completed and I'm listening to them at the computer while writing this, the sound seems to "visually" come from the center, behind the TV screen, from the front wall. It was a strange sensation at first, but sounds good. It's definitely a high quality horn.

Cabinet

The cabinets were ordered from a professional carpenter, which makes loudspeaker cabinets for some DIY web shops, so they got good experience and exact drawings for this design. I decided to trust them and I got a really nicely made and finished cabinets, made to order with minor customization over the standard Ruutinassakka.

Cabinets.

The cabinet is 30 x 57 x 35 cm (WxHxD), made from 1" (25.6 mm) thick MDF, it has one horizontal brace between driver openings, one smaller vertical brace and also a separate, sealed compartment for the plate amplifier. There is no reflex port, like in the original, this is sealed. Net air volume is close to 34 litres without drivers. The cabinet is heavily stuffed with wool-like foam (Ewonawool), which is excellent material, but a bit pricey for this purpose. Still affordable in the total cost of one pair of speakers.

Heavily stuffed cabinet.

Amplifier and Filters

The HF+horn produces 110 dB/1 m/1 W, so it's obvious that we would have a severe sensitivity mismatch problem with analog filters. I had auditioned the reference build of Ruutinassakka with analog xo in an (DIY) audio store and I had to turn down the treble dial on the amp a lot because the levels were simply too much. At home I actually drove it to ear-splitting loudness with a headphone amp(!) after I measured its impedance, so if you use this combo, be sure to make no mistakes while not wearing hearing protection. Luckily I'm using a very capable DSP plate, so it's a non-issue!

At the time of putting it together, I did not have appropriate resistors at hand to reduce tweeter level to allow better matching and noise reduction, so I decided to leave it out for now. The speaker produces very little hiss, so little in fact that it's not bothering me at all. I also tested the plate amplifier without having protection cap to the tweeter, the FusionAmp is very clean and well-behaving amp. Turn on/off thumb is not a thumb, it's a faint click, which will not damage any speaker component as-is. Protection capacitor is still a useful component, so don't leave it out.

Hypex FA123 from both sides.

The FA123 has a 100 W (4 Ohms) HF module and two 125 W (4 Ohm) multipurpose modules. The high-frequency section will produce 25 Watts for the 16 Ohms driver. The MP modules are in bride-tied-load (BTL) configuration in this speaker, so they should total for 125 W for 16 Ohm midbass driver. I actually forgot which program file I last uploaded to the speaker, so I hope I'm showing the correct data here. I've been having updating the DSP program quite a lot during the first year after initially completing the speaker (yes this article is written over a year after the speaker was physically completed).

All high-pass (HPF) and low-pass filters (LPF) are Linkwitz-Riley 12 dB/Oct. High-frequency channel (LF) has a -20 dB gain without the analog attenuation. HF has a HPF at xo and a few other filters to shape up the response. The low frequency channel (LF) has a positive gain of 10 dB on top of the filters. This is fine, because the FusionAmp has a good headroom in it's DSP, so it won't clip. LF has a LPF at xo and a HPF at 60 Hz. The image shows many filter points, but not all were enabled or used in the final version of the DSP program. The goal with the LF was that the DSP in the plate amplifier EQ's the response flat to ~50 Hz, from which point it will be high-passed so that low frequency signal won't modulate the woofer. I also use subwoofers when needed with these speakers and they are now crossed at 100 Hz, so the summed crossover region should be quite good.

HF channel filter shape.

 

LF channel filter shape.

At this point it's worth noting that the 16 Ohm 10NMB420 is actually more suited to midrange duty and it's frequency response drops like a stone below 200 Hz. It really takes insane boosting to get it to 60 Hz. Surprisingly, it does get there with surprisingly little audible effects. I cannot hear any defects in it's bass response and it sounds much more authorative than any 5-6" woofer I've heard. These PA drivers really can take some serious signal levels without going up in smokes!

Measurements

There is no extensive measurements of the design available at this time. I have full trust in the original speaker's designer, who came up with this concept and physical design. I do have some measurements for my speaker and its DSP program though. All measurements were taken with the final cabinet and plate amplifier unless otherwise stated.

Compression driver + horn

HF on a separate test bench. Nearfield SPL and phase.

HF on a separate test bench. High SPL measurement.

The highest harmonic distortion component with the compression driver is the H2. I've read that this is inherent to the technology, how the compression driver works. Luckily it's still fairly low -39 dB at rather high SPL and the third harmonic is almost non-existent among higher orders.

Midbass driver

Near-field SPL and phase of the midbass.

Near-field SPL and distortion of the midbass driver.

Full-range

Mid-field full-range SPL and phase.

Mid-field, full-range SPL and distortion.

The slight differences in the graphs are caused by different DSP programs. Some of the measurements were taken at different times. The HF response of the NSD1095N+XT1086 does not go easily up to 20 kHz, but I find that an acceptable trade-off, because my hearing is no longer good past 15 kHz. There is a general tilt in the FR, the start-point of bass is roughly 6-8 dB higher than end-point of treble. That's by design.

To those who understand these measurements the graphs should be self-explanatory. For those who aren't so measurement savvy, let me say this: the distortion of these speakers is very low, less than 1 % at power levels where small desktop speaker would show nicely flaming voice coils.

Some More Pictures

Considerably larger than my previous speakers.

Completed set in use.

The Sound

My fears for undetailed, unarticulated sound were on no grounds. This speaker is absolutely stunning in all aspects, especially when used with a proper, sealed subwoofer. It's got all the detail, but not adding any extra glitter. Now having listened to this speaker for some months and then attending to a high end audio event, the only speaker that had comparable sound quality to my ears was the Kii Audio THREE with its BXT bass system. All others were either simply lacking in bass, muffled, not clean or played at so low levels that I had to hold in my farts breath to hear anything.

Funny thing that this speaker does: all SPL levels sound the same. If you crank it up, you can hear it going louder and vice versa for lowering volume, but after a minute the subjective SPL is the same as before changing the volume. I've read about similar experiences with speaker systems that have really low distortion even at high sound pressure levels. This speaker will simply not produce much of a distortion until the sound pressure reaches unhealthy levels (for your ears).

Now after a year, there is simply no going back to a lesser degree of drivers and cone surface area. I might build a small set for my work desktop, but for a living room, I just cannot see myself downshifting any decade soon.

Conclusion

All in all I'm very happy with how they sound and when they are crossed to subs the mid-range improves even more and then they really sound fantastic! I will continue to keep tweaking DSP settings and probably add analog attenuation. I may even add a discrete woofer to help with bass reproduction, although that would mostly give me more SPL, which I don't need right now. I've eyeing out some 15 inchers that would play well with a sealed box and give some bass response withouth EQing the living crap out of them. That would seriously ease up the DSP program and phase response, which could yield good results and further research subjects. My only issue is space and the need to use these speakers in both near-field at desk and at longer distance at the couch.

I highly recommend Eighteen sound drivers. If you are looking for a 2-way without using a subwoofer, I would recommend using some other model for midbass driver. You could also try out one of those new 1.4" compression drivers and a larger woofer with a sub-1000 Hz xo point.