Tír Na HiFi

Hi all,


I thought I'd start a thread about a set of speakers that I'm undertaking as a COVID project......

So I've had a hankering to make a new set of speakers that would fit somewhere in the line up of all sorts of oddball stuff I have here. The design goals (some of which I'm making up as I go along):

1. I'm only interested in making a pair that sound bloody good. They need to do at least one thing special.
2. I don't need crazy low subwoofer level bass but I do want good bass response thats tuneful.
3. No harshness or excess sibiliance
4. Excellent soundstage
5. Non-fatiguing.
6. General design philosophy of KISS

Yeah, not asking much am I?

Anyway, after a lot of thinking and looking, and remembering various speakers that I've heard over the years, I took a generous drop of inspiration from the Audioplan Kontrast. A member here, Briano, has a set that I heard a fair bit of material through, and although they are 2 way rather than 3 way, they do a great job on the design goals.I remember hearing them ages ago in Dereks house as well. In short, if I got that sound I wouldn't be doing too bad.

At this point, you should go off and google them. There is precious little info out there on them, and less than that in terms of technical information that actually might be useful. I did get just a few nuggets that helped hugely. The mid/woofer is an Audax aerogel driver - these are not plentiful, but have a great reputation. The tweeter is a Seas model I think. The early models of the Kontrast line might have been transmission line, but the most recent ones are all bass reflex. The other bit was the crossover is at pitched at 2500hz. The front facing port also helps with room positioning.

So with those bits of info I went off to see how I could use them as a starting point to develop my own take on it. I knew from experiments over the years that designing a speaker is a holistic process, or at least doing it that way gives you best chance of success. Having said that, you have to start somewhere.... so I started with the Audax 170mm aerogel driver. I looked at so many tweeters!!! You need a tweeter that's a little more sensitive than the woofer, so I wanted something about 92/93db or so. I knew I wanted great soundstage and no bleeding ears. I reckoned fabric domed would be the way to go, and I also reckoned that picking one with a larger diameter would likely help that. The downside is that the bigger the diameter, the more they roll off at higher frequencies. Anyway to save you having to read more ponderings, I ended up with another Audax driver, a 34mm fabric dome in a huge 5" mount.

Box design next....

This should get to be an interesting read...

More! More!

Those of you familiar with the Kontrasts will know that like a lot of other high end speakers, they put the tweeters in a separate box. This complicates matters quite a bit but also offers some advantages. So the conventional wisdom is that having the drivers as close together as possible helps coherence in sound (part of the attraction in single drier horns), although as usual it isn't as simple as that. The portion of the band where each driver does its business is also important. Building a second box is also more money in a commercial setting, and you have to figure out connecting everything together in a robust way. In the mass market world, if its possible for someone to connect something wrong, they will. Luckily for the home builder this isn't a big deal.

So with a crossover pitched at about 2500hz, all the human voice range, and many many instruments are handled by the mid/woofer alone. I elected that I wanted all that to be at ear height when planted on my arse in the listening chair. That left me with the necessity to put the tweeter in a separate box that I could position/angle to blend and get the soundstage right. Its a really interesting experiment to only connect the mid/woofer and listen to it n its own, then reverse and listen to the tweeter on its own. You find that on its own, the tweeter appears to produce very little but it absolutely transforms the sound with air and life.

So, looking at the specs on the mid/woofer, I knew if I went for a box around 38-40L I could get meaningful output down at about 40hz, but that this would be the limit. So given ear height, some guesstimating, I went for a bottom box 950mm tall which when put on feet would put the mid/woofer at ear height.

The next decision was what to make it out of. Nearly all recommendations are for either baltic birch ply (high quality ply) or MDF. The ply is quite hard to get, and I knew I had a supplier of good quality MDF not all MDF is equal). So I went for 25mm thick MDF to give a balance between using MDF and achieving stiffness in the box. Tweeters are essentially sealed units, so the box doesn't matter as much for them, although you do still want a stiff rigid support.In any case, given I was already cutting 25mm MDF, the tweeter boxes are cut from that too. Now, I don't know if any of you have ever lifted a sheet of 25mm MDF, so let me tell you right now its heavy as f%&k. Very luckily, the company supplying the MDF also cut it for me to the final panel sizes on their CNC saw and I tell you now, its an absolute Godsend. The pic below is the pile of cut pieces. I got a couple of extra front panels and bits cut to allow for the inevitable mistakes.
More shortly!

Crazy man Fran! Good luck 😉

This is going to be a great thread. I had a pair of Audioplan Kontrast 5's a while back, a latter model to the ones Briano had, The 5's too were transmission line. They has a gorgeous sound, not forward or bright or shouty just a lovely effortless voicing to them. I believe Audioplan tune these speakers with Jadis amps.

They has a gorgeous sound, not forward or bright or shouty just a lovely effortless voicing to them. I believe Audioplan tune these speakers with Jadis amps.

That's exactly what I'm after. Hopefully I get there in the end!!

So the next step was to define the crossover. This is made up of two filters; one to cut frequencies above a certain level for the mid/woofer, and the other to make the reciprocal cut for the low frequencies that would otherwise go to the tweeter. The thing is that it is not possible to have a hard limit on this, and it is in itself a bag of constantly moving snakes. Such filters are defined in "orders" - this means how quickly the frequencies tail off. So a first order has quite a gentle slope, a second order is quicker and so on. The higher the order, the more and bigger parts you need. You loose something in every extra part you add, but you also loose something if you don't get the slope right as the driver will be either trying to do something its not designed for, or the overlap between both will be less than perfect.

I don't currently have the facilities to do measurements on the drivers, so I was dependent on the spec sheet that comes with the drivers. The simple crossover calculators that you find online are pretty much completely useless as they don't take into account impedance or the box. However, you can use more complicated software that will give you a really good starting point. The one I used is called vituixCAD and is pretty user friendly.

The first step was to get the graphs for SPL and impedance into the software. On the modern drivers this is usually available as a file you import, but for the ones I was using it means copying the graph and tracing it in using a tool in the software.
After this the next step is the first stab at the crossover components.......

This is great Fran! I remember hearing Audioplan Kontrasts with Jadis JA-80s at a Cloney Show a good while ago: it was perhaps the most enjoyable sound of the show for me.

I don't envy you trying to work with 25mm MDF, I made some curved shelves out of it once, which looked great, but it was very hard to hold and move about. And I was 20 years younger then!

Very interested to see how this works out! Enjoy the project, Fran!

Thanks Ciaran!!


So, the great thing about the better speaker design software is that you can endlessly mess about and see (in theory) what each one does to the response. My goal was to have a proper crossover, but with as few components as possible. The tweeter here is a little more sensitive than the mid/woofer, so the first step I took was to drop in a series resistor to bring that down a bit. A resistor like this affects all frequencies equally, so it drops the level right across the board. With an approximation in place, the next task was to add the 2 filters to cut the highs to the woofer and the lows to the tweeter. If you scoot back up and look at the graph for the mid/woofer, you'll see a peak at around 3000 hz. That is what is called cone breakup, and basically we need to stay away from that. Likewise if you look at the tweeter graph you'll see it slopes upwards to about 1000hz, so we want some comfort and stay well above that. That leaves a crossover point somewhere between 2000 and 2500hz as the optimum spot. After much modelling and trying different configurations, the V1 crossover is shown here:
The graph here shows the resultant modelled SPL response curve, which is not as ruler flat as you'd like....
That little bump at the low end should help give a bit of heft down low. The crossover point ends up nicely around 2300hz. The heavy black line in that graph is the cumulative response, and the other two lines show the individual responses for each driver. Lastly, a resistor across the woofer helps with the impedance curve which should make them a bit easier to drive.

Bear in mind that the crossover here gets you in the ballpark only - there's plenty of tweaking as you go and actually hear them in the box.

Next up is assembly.....

Fascinating. I even understood some of that.