My Swansong speaker build

[Toppsy]

Scott and I have never built and designed a 3-way speaker so whilst I still have the mental, physical and hearing ability, together with the means to fund a new speaker project, I have been pondering a new speaker build for a while now that would (hopefully) surpass the performance of my Edingdale GT speakers.

After discussing with Scott my thoughts and aims for this swansong speaker, Scott sent me some links to potential speakers that could fit the bill. Of these, one speaker in particular piqued my interest. This was the Tekton Design's MOAB speaker. This appealed as it had polar arrays for the midrange drivers and twin bass drivers - both design features I wanted to incorporate in my own speaker build. Surprisingly, I showed Lynn a photo of these speakers that Scott sent me and she actually liked them but asked if the cabinets could be made a little smaller.

To achieve this I agreed with Scott to use 10" woofers instead of the original 12" Eminence units. So armed with information Scott researched suitable 10" bass drivers for me. This Scott did most thoroughly and sent me details and specs for a dozen suitable drivers by Faital Pro, Seas, ScanSpeak, Dayton Audio and Visaton. As this is intended to be my swansong I went for the Scan Revelator 26W/8867T00.

The original MOAB design uses the same basic SEAS tweeter for the midrange polar arrays and the HF driver. I didn't like this arrangement of using 14 tweeters in 2x polar arrays for midrange duties. Having spoken to Scott I suggested and agreed with him we instead use the SB Acoustic beryllium dome tweeter for HF duties and MarkAudio 2" full range drivers for the midrange arrays. I believe this will be sonically better.

So having now agreed the drivers for the build, calls went out to Jerry at Falcon for the Scan Revelators and Beryllium dome tweeters and to Stefan at KJF Audio regards the MarkAudio CHN-50 drivers and orders placed. These are all now safely delivered.

One other design consideration I wanted to incorporate was to go active (with DSP) for the bass but passive for the HF and midrange. Scott suggested I speak to Chris (chris661) in Sheffield regards going active with the bass and using DSP as he was more knowledgeable then he. This I duly did and we had a very useful chat. Scott did mention this arrangement was something that Troels Gravesen had been doing on a few of his designs and sent me a link to his web site. Troels uses Hypex FusionAmps (FA251 and FA501) for controlling the bass on his designs.

Having checked these out the Hypex units get very good write-ups and seem to be regarded very highly. So another call to Stefan at KJF Audio as he is the UK agent and distributer for Hypex. This resulted in me ordering a couple the FA251 FusionAmps. Well if it's good enough for Troels they must have something going for them. These arrived yesterday. Much reading and learning me thinks to get to grips with this, new to me, technology. Me thinks another call to young Chris for potentially some help.

So armed with all the above I have drafted up the first pass speakers design. Copy of CAD file below.



Next up is to get quotes for having the front baffles CNC routed in 25mm thick sheet and the other panels cut from 18mm sheet using a CNC controlled flat bed Beamsaw. Having checked out the current price of Finish Birch ply I might have to bite the bullet and use Furniture Grade chipboard for this build. I think I've ranted on and taken up your time enough for one day. I'll post as I make further progress with this build.

[IslandPink]

Bravo Colin, this is quite an undertaking. Also a big departure from anything you've done before.
Like the Scanspeak driver, had a quick look at that.
What's the name of the MarkAudio 2" one ?
Are there any downsides to having the tweeter below ear level ( if it is ) ?

[pre65]

An ambitious project Colin.

I look forward to progress updates and I hope the finished articles meet your expectations.

[Toppsy]

What's the name of the MarkAudio 2" one ?

The MarkAudio driver is the CHN-50. This is a little full range driver that has a frame the exact same size as the Beryllium dome tweeter. It's a nice little driver. I thought the cones would match the cone colour of the beryllium driver but on receipt of the drivers the MA cones are a bright silver!

As an experiment I have lightly sprayed the cones on 1-pr the drivers matt black using acrylic spray paint. Car paint rattle cans. These look 100% better. Before I do any the other 26 drivers I'm going to knock together a couple pairs small sealed boxes and mount the black ones in one pair and the silvers in the other. I can then A-B listen to see if this has had a degrading effect on the sound or not. I'll report further when I have finished the small boxes.

Are there any downsides to having the tweeter below ear level ( if it is ) ?

I had thought of this. The speakers will have tilt levelling feet front and rear under the base. This will allow for a few degrees of tilt to the cabinets to aim the tweeter towards ear level at the listening distance some 2.5m from the speakers in my room.

[Nick]

Just a thought, is it worth considering 16 instead of 14 mid range drivers? You could then wire them as four sets of four and so would electrically look like a single driver, but still be able to expect equal division of the signal to every driver. I also wonder if you could then arrange the 16 drivers around the HF tweeter or below it. And so maybe as Mark suggested shift the tweeter up a little.

But other than that, looking forward to seeing how it works out.

[Toppsy]

Mark and Nick,
Thanks for the feedback observations.
The tweeter is a little low and checking ear height at my listening position means a 5 deg tilt of the cabs. Too much me thinks.
I can reconfigure the driver array to something like this:



This will bring the tweeter up to ear level. However I am mindful of the visual aesthetics as I have to get Lynn's approval of how they look when built. Might struggle to get 16 the midrange drivers to work? More CAD work needed.

[ed]

good luck, looks awesome...
I've probably done my swansong after the discover of birch prices with the last one. Some outlets are looking for upwards of £200 for a sheet of BB grade.
I've got 2 sheets left of the Chinese stuff but it's absolutely pants for finish.

When I was making clocks a cabinet maker moved into the area and we got chatting. He took over my place in the band I was with....another story...but he suggested I use skinned chipboard for my clocks....outrageous for a period clock but the stuff he was using for furniture seemed really good for finish.

I was impressed with his furniture though, he had a 3' wide belt sander....I'd never seen one before....phwooaarrr.

I digress, back to your build

[simon]

Some outlets are looking for upwards of £200 for a sheet of BB grade.

Clucking bell!

[Scottmoose]

Colin asked if I'd shove my head around the door re the Tekton configuration.

Despite appearances, Moab is a 3-way WMTMW. The bass frequencies are handled by the two woofers. Where it gets unusual is that the two clusters of seven tweeters are actually used as a pair of midrange units. Those finally hand over to the single central tweeter. Actually quite a clever idea -not strictly speaking 'necessary' (the patent justification is worth a read if you enjoy a good laugh) but most people who've heard Moab report it does work well, so why not? It's not perfect though: even though 14 x 1in dome tweeters have a decent amount of radiating area, they have very little excursion, & they're run very low indeed, so distortion at the bottom end is going to be higher than ideal. Colin, being brighter than the average bear, hit on the idea of using 14 x Markaudio CHN50 for the midrange elements instead of tweeters as their baskets are more or less the same size as a regular dome tweeter, they have better performance at the lower end of their range (since they have 4x the cone area and 10x the excursion) and they're also quite inexpensive.

Fourteen CHN50 in two paralleled packs of 7 series-wired units (a la the attached) gives the following:
Nominal array impedance 11.9ohms
Efficiency gain (over single unit) = 11.46dB
Sensitivity gain (over single unit) = minus 5.44dB (as impedance is 3.5x that of a single unit)
Net array efficiency = 100.56dB 1m/w
Net array sensitivity = 95.12dB 1m/2.83v

Shifting to 16 units in 4 parallel packs of 4 series-wired units would level efficiency and sensitivity gains at a whisker over +12dB with a net array efficiency / sensitivity of 101.14dB [nearly]. Bit more challenging on the physical size & configuration though, since for best results in an MTM you want to minimise the distance between the tweeter and the outermost midrange element. Not impossible though; Colin & I were discussing it earlier. With the 14 unit configuration the main issue is that 12ohm impedance, but we can damp that with a resistive shunt to a slightly lower level; since the lower crossover frequency will be near the box peak we'd likely need to do that anyway so assuming we take that avenue rather than upping the number of drivers it can pull double-duty.

[Nick]

With the 14 unit configuration the main issue is that 12ohm impedance, but we can damp that with a resistive shunt to a slightly lower level

I cant think of a better demonstration of the differing world views of loudspeaker and amp designers

[Scottmoose]

In which way[s] did you have in mind? (genuine question)

[Nick]

The idea of burning power with a resistive shunt. If there was no other way of avoiding it I would have considered an autoformer to impedance match.

I wasn't suggesting your way was wrong it's just it seems to be wrapped around the idea that driving power is free. I inherently migrate towards efficiency I guess.

[Scottmoose]

Curses, this is what happens when you try to write something over a couple of hours while doing other things -damn thing logged off in the meantime so lost it all. A brevity version of what was there, as it is now beer o'clock:

Yes, I suppose it's a matter of perspective. A passive speaker system efficiency / sensitivity is ultimately governed by the least efficient leg. To use this one as an example, and sticking with 2.83v sensitivity for consistency (since each leg is of a different basic impedance) the bass end is about 93.5dB, the midrange about 95dB and the HF about 95.5dB. That being the case, the midrange and HF have to be padded anyway to the level of the bass units, so since we're going to be burning up a little power in those legs whatever we do, I generally think it makes sense to do so in a way where we can achieve several things at once. While the speaker isn't going to equal a full-sized Altec setup on the efficiency / sensitivity front, it's fairly good already in that sense, & damping down a midrange impedance that is a little higher than ideal (which should also help make it a little less reactive) should if anything make the amplifier a bit happier than otherwise with the overall load its presented. I could do the same with LCRs and autoformers, but the extra complexity & costs are a consideration, and we would still need the level padding.

A couple of the questions I usually ask when I'm designing or helping design a new speaker are: what is the maximum headroom requred at the listening position, what is the projected speaker efficiency, how reactive is the impedance load, and what is the driving amplifier like (specifically power & output impedance, since we're dealing with a combined system rather than individual components). All other things being equal, I try to aim for relatively unreactive loads as for a given impedance & sensitivity it should give the amplifier an easier time.

[Nick]

That being the case, the midrange and HF have to be padded anyway to the level of the bass units

I thought Colin was talking about using a Hypex amp for the LF?

Yes, I suppose it's a matter of perspective

Yes, that's all I was saying, it seemed to me, to shine a light on the perspective of the speaker designer versus that of someone coming from the amp side.

[Nick]

12 drivers would also work surrounding a single tweeter, giving a 4.8R driver. Just thinking out loud. Sorry.

and we would still need the level padding.

Well, not with the autoformer, that was my point. I use one on the tweeter of the OB's precisely because of the monkey on my back about wasting power in an L-Pad