The ABC’s of ABC boxes-An upcoming adventure

Matthew J Poes

AV Addict
Thread Starter
Joined
Oct 18, 2017
Messages
1,903
i very much hope this turns into a real project to share and not just an exercise in theory and modeling. None the less, I’m going to describe and foreshadow a planned future project and explain a little of why.

For those familiar with Hornresp, you may have noted an enclosure option called an ABC box. You probably thought it was a joke, some kind of placeholder for something. It’s actually the name given by the software author for a kind of series tuned dual reflex enclosure known as an Aperiodic Bichamber enclosure.

What is it:
It’s a box with a response similar to a 4th order reflex enclosure but dividing the internal volume into two chambers. On the larger chamber is a port of a given length. Between the chambers is another port of the same given length. Finally that second chamber has a third port of again, the same given length. That means we have two chambers, a direct radiating driver, two external ports (one in each chamber) and one internal port (series loading the first chamber to the second chamber/port.

What does this complex setup do, it creates two minimum motion notches roughly one octave apart of similar Q (which can be varied by the internal damping).

I believe Don Keele was the first to discuss this kind of enclosure, and it’s primary purpose is excursion control. In fact, if the ports are linear, it would thus also provide a reduction in distortion.

Why use it?

Distortion reduction and excursion control, and the distortion reduction is a direct result of the excursion control.

Here is why that is important, if a driver has high linear xmax and modest power handling, then such drivers would not exceed their xmax before exceeding their thermal limits around and above the port tuning frequency. As such, it provides no advantage for such drivers. I see no value in using this with typical home theater subwoofer drivers of say 20mm xmax one way (or more) and 1-2000 watts power handling (or often less).

So when do you use this and why am I bringing it up?

Because I define a subwoofer by its ability to meet certain performance characteristics with a minimum of distortion. For example, to produce 20hz to 100hz up to 120db at 80hz, with little distortion. There is a complete misnomer that we are insensitive to distortion at low frequencies, which is not true. In fact some forms of distortion at certain volumes are very audible at very low levels (fractions of a percent). Because we are dealing with non-linear distortion it’s inportant to remember that smaller amounts of these distortions are audible when at higher volumes. With subwoofers the mechanical limits of the driver can contribute to these distortions so controlling high excursion with ports can be a good thing, as long as you haven’t traded other problems.

Knowing all that, I like pro audio drivers with modest xmax compared to pro drivers but often larger coils, lower inductance, and higher power handling. Not all pro drivers can replace typical HT subs so this isn’t a blanket statement, but many can.

Take an 18” driver with 2000 watts rms power handling and 15mm xmax. This would be a very high performance pro subwoofer but those specs might seem modest at best compared to an HT driver. Now imagine that it’s got a RE of 2ohms, a BL or 32, and xmech of 60mm. It’s sensitivity is 99db (but not at the low frequencies for which it will be used). If we stick this driver in a modest/large ported box and tune it to 20hz it may actually come very close to the maximum output of a typical 18” HT driver. However it may be limited below 35hz by exceeding its linear excursion. While it may need less power for a given output and have lower distortion in that range, it would probably fall short by a factor of 2-4 of those HT subwoofers at 35hz and below (3-6db). Some say, well that’s the trade off. You can have loud and low distortion low bass or loud and low distortion upper bass, but hard to get both.

here comes the ABC box. Now with a ported box of typical construction a minimum motion notch would exist centered at the tuning frequency. Tuned to 20hz it wouldn’t provide much of any support in controlling excursion at 35-40hz and yet the driver may very well be exceeding xmax by that frequency. Now imagine two minimum motion notches centered at the same 20hz and at 30hz (or maybe 35hz). Now you have the exact same response shape as before, but you can gain significant output before exceeding xmax.

The Keele rule for these boxes is that volume is taken as the normal volume for the bass reflex you intend to design. You then divide it in 3rds. You then make the driver chamber 2/3 total, and the second chamber. Port as mentioned previously. Port dimensions are all identical but obviously port tunings are not.

So you can use very efficient and very linear bass drivers in very high performance subwoofers and gain additional headroom in the low frequencies allowing you to tune often a full octave lower while maintaining good excursion control.

When I have time I’ll try to post models comparing a common driver in these two boxes to show what happens. You will have to use your imagination a little bit as Hornresp doesn’t represent reality correctly. The internal port is represented as nothing more than a minimum motion notch in the response with no port contribution. I think it’s a mistake but not being as smart as Marten King, maybe just a limitation of such a modeling approach. None the less these have been built and measured by Keele and others (they became popular in car audio too) and do work as advertised.
 

chrapladm

Moderator
Staff member
Joined
Apr 14, 2017
Messages
334
Hmm sounds interesting. I will have to do some simulations on HornRep and see how they differ. I was already looking at using pro audio drivers also for my HT subs.
 

Matthew J Poes

AV Addict
Thread Starter
Joined
Oct 18, 2017
Messages
1,903
Hmm sounds interesting. I will have to do some simulations on HornRep and see how they differ. I was already looking at using pro audio drivers also for my HT subs.

As I mentioned before Hornresp can model them but it will model them wrong. You have to imagine the suckout in the response isn’t there, because it isn’t. I’ve written to David about this but not heard back. I’ll show a set of simulations to show expected and modeled behavior but obviously he ultimate goal will be to build the box.

One thing that worries me a little is that both Winisd and Hornresp show an increase in excursion below the upper tuning when Two tunings are an octave or so apart (with Winisd I just modeled a 6th order bandpass with very close tuning of 20 and 40hz). Done Keele specifically talks about this box offering excursion control and numerous experiments and tests have been performed to show advanatage. If the models are right with regard to excursion, they won’t offer any advantage. That increase in excursion between tunings would negate any other benefit. I hope to investigate this but I won’t be able to measure excursion so that may be an unknown.
 

Matthew J Poes

AV Addict
Thread Starter
Joined
Oct 18, 2017
Messages
1,903
Hopefully I can post some details later, but here’s an update. David says he believes the models are right but has never tested it against a real box, nor is he aware of anyone who has. I have a friend with a generic eBay special ABC box so I will take measurements of that box and report back with results. It will serve as an experimental confirmation and proof of concept.

A problem may have come of this though. I can’t measure xmax and it appears that placing two ports an octave or two apart will cause an excursion peak between them that is drastically higher than in a typically ported box. I may never be able to test this.
 

Matthew J Poes

AV Addict
Thread Starter
Joined
Oct 18, 2017
Messages
1,903
Here are some models including some of what I think the response will really look like. Prototypes and proof of concept measurements are pending.

Here is what the model shows for the response of a B&C 12TBX100 in an ABC box
ABC box.PNG
I know, whats with those resonances, no way you want that. No worries, because there is no way it would look like that. Those resonances are certainly being exaggerated, and the dip in the middle is the second port tuning showing up as a resonance, which is now how it works.

But we can at least see where the displacement dips take place:
ABC box displacement.PNG
But again I can't model all the behavior in Hornresp precisely, so I believe the Q of these dips is wrong, too high. I believe with damping material it will be much much lower.

Here is a model using a 6th order bandpass in WinISD with some PEQ to create what I think the response will actually look like.

Color coding is as follows
Sealed: Red
Ported: White
ABC: Purple
ABC Box Output.PNG

And displacement
ABC box displacement_Reality.PNG

I've talked to a few experts on this topic (including someone who wrote a JAES paper on the topic), and they diverge in their opinions. They all told me that few people have ever carefully built and measured these in a reputable way, so its hard to know, but that certainly if they looked like the model, nobody would have ever bothered even publishing a paper on them.

If we remember that the areas with lower xcursion will be associated with lower distortion, then you can see how this approach helps bring down distortion. If the Q of the displacement dip can be lowered, that will increase the area that has reduced distortion. We should also gain a decent amount of efficiency within the pass band due to port contributions.

I have a friend who owns a pair of Obcon car subwoofer boxes which they claim are using proprietary technology, but which I recognize as an ABC box. I want to use one as a proof of concept and see what happens. If they measure with a resonant dip as modeled, I'm going to abandon this project. If I don't see evidence of the resonant dip, I'm going to continue with the project and mentally adjust for the problem as I plan the box. I will also be talking with David McBean to see about reconciling the differences from the model and reality. My theory is that there is either a phase problem from a pole or he isn't including the contribution of the second port.
 
Top Bottom