Post Your Triple Reflex Bandpass Designs

[Patrick Bateman]

Over on my forum I've been having some fun exploring why tapped horns sound so good, and in the process I stumbled across a subwoofer alignment that isn't used much by the DIY crowd.

In the interest of exploring it's benefits, I thought it would be fun to post our triple reflex bandpass designs.

Based on my experiments with it, it has some interesting features:

Pros:

• It's possible to get absurdly low cutoffs with a triple reflex. In fact the biggest challenge I've had with these boxes is getting the F3 *high* enough, not low enough!
• Like all bandpass boxes, the vents roll off distortion. Triple reflex boxes do it to a greater extent than single or dual reflex. I do not like the nomenclature of calling these "eight order bandpass" because the rolloff isn't always eight order; it could be as shallow as fourth order.
• Because triple reflex boxes have three excursion dips, you can get absurd power handling. I'll post a design for an eight with an F3 of 15hz and 1000 watts of power handling. (out of an eight!)
• If your goal is a little bit more down to earth, you could probably get some deep bass out of small cheap woofers (like Bose did )

Cons:

• This is a hideously complex design. I am still trying to wrap my brain around it, and it seems that changing one variable affects all the others. For instance, if you change the tuning frequency of one port, it affects the other two ports as well.
• Don't even THINK about building one of these without a woofer tester. Bose stopped making these, and I'll be it's because even a bit of a drift in the QTS of the woofer will change the frequency response in a big way.

[Patrick Bateman]

Here's a triple reflex with a single eight that can do one thousand watts without running out of displacement*, with an F3 of 18hz.


Frequency response of the sub. We get about an octave and a half of bandwidth.


Here's the excursion. Due to the three dips in excursion due to the three tunings, we can handle over a thousand watts with a single eight. (* two ohm load - I designed this for the car where 2ohm is where amps deliver their maximum output.)


The group delay of the sub.


For comparison's sake, here's the Danley TH-Spud. F3 is about the same, but the TH-Spud has higher efficiency because it's bigger and it has twice as many woofers. The TH-Spud also has wider bandwidth. I think the triple reflex bandpass is a good option when you don't have the space for a tapped horn. Folding a tapped horn can be time consuming also.


Frequency response of the triple reflex using the same scale as the TH-Spud.

[Patrick Bateman]

Here's the Akabak script I used to make the sub. Note that the port on the third chamber is flared, just like the Bose subwoofer.

System 'S1'

|================================================= ================================================== =====
|REQUIRED AKABAK SETTINGS:
|File > Preferences > Physical system constants:
|Sound velocity c = 344m/s
|Medium density rho = 1.205kg/m3
|Sum > Acoustic power:
|Frequency range = 10Hz to 20kHz
|Points = 533
|Input voltage = 2.83V rms
|Integration = 2Pi-sr
|Integration steps = 1 degree ... 1 degree
|Integration method = Cross

|================================================= ================================================== =====

Def_Const |Hornresp Input Parameter Values
{
|Length, area and volume values converted to metres, square metres and cubic metres:

S1 = 248.22e-4; |Main Port Throat Area (sq cm)
S2 = 182.37e-4; |Area of main port tube (area betweent the flares)
S3 = 248.22e-4; |Main Port Mouth Area (sq cm)

L12 = 2.54e-2; |Main port segment 1 axial length (cm)
L23 = 50.0e-2; |Main port segment 2 axial length (cm)
L34 = 2.54e-2; |Main port segment 3 axial length (cm)

Vrc = 31.0e-3; |Rear chamber volume (litres)
Lrc = 22.5e-2; |Rear chamber average length (cm)
Ap = 45.6e-4; |Rear port cross-sectional area (sq cm)
Lpt = 82.6e-2; |Rear port tube length (cm)

Vtc = 14.16e-3; |Front chamber volume (cc)
Atc = 550.00e-4; |Front chamber cross-sectional area (sq cm)
Ata = 45.6e-4; |Front port cross-sectional area (sq cm)
Tpt = 46.07e-02; |Front port tube length (cm)

V3c = 85.44e-3; |Third chamber volume (litres)
L3c = 30.0e-2; |Third chamber average length (cm)

|Parameter Conversions:

Sd = 200.00e-4; |Diaphragm area (sq cm)
Arc = Vrc / Lrc;
Ltc = Vtc / Atc;
A3c = V3c / L3c;

}
|================================================= ================================================== =====
|Network node numbers for this horn-loaded vented-box system:
|0-Voltage-1-Resistance-2
| |
| 3-Chamber-4-Driver-5-Chamber-7-Port-12-Chamber-8-Segment-9-Segment-10-Segment-11-Radiator(1)
| |
| --------Port--------------------------
|================================================= ================================================== =====

Def_Driver 'Driver'

Sd=200.00cm2
Bl=18.30Tm
Cms=3.85E-04m/N
Rms=5.68Ns/m
fs=28.90Hz |Mmd = 74.30g not recognised by AkAbak, fs calculated and used instead
Le=2.90mH
Re=7.40ohm
ExpoLe=1

Driver Def='Driver''Driver'
Node=4=0=5=6

Coil 'L1'
Node=1=4
L=4mH
Rs=0.35ohm

Duct 'Rear Chamber'
Node=5=7
SD={Arc}
Len={Lrc}
Visc=0

Duct 'Front chamber'
Node=6=8
SD={Atc}
Len={Ltc}
Visc=0

Duct 'Third chamber'
Node=11=12
SD={A3c}
Len={L3c}
Visc=0

Duct 'Front Port'
Node=8=11
SD={Ata}
Len={Tpt}
Visc=0

Duct 'Rear Port'
Node=7=11
SD={Ap}
Len={Lpt}
Visc=0

Waveguide 'Main Port Segment 1'
Node=13=12
STh={S2}
SMo={S1}
Len={L12}
Conical

Duct 'Main Port Segment 2'
Node=13=14
SD={S2}
Len={L23}
Visc=0

Waveguide 'Main Port Segment 3'
Node=14=15
STh={S2}
SMo={S3}
Len={L34}
Conical

Radiator 'Horn mouth'
Node=15
SD={S3}

[jbell]

is it possible to model in hornresp?

[FrankWW]

Do you think you could post your forum address, please? Maybe add it to your signature.

[revboden]

I've got another con for you. Transient response is terrible in triple reflex bandpass designs.

[Brian Steele]

IMO 8th order bandpass = looks like a good way to destroy drivers :-)

While the sim suggests that the driver can absorb 1000W before excursion limit is reached, think about what's going to happen when an 8" driver is hit with 1000W. Even 500 W. Hell, even 250 W. T/S params will shift, and the alignment no longer matches the nice Akabak predictions. Then of course there's the nonlinear effects of three vents tuned to different frequencies that are going to mess things up too.

[454Casull]

Originally Posted by Brian Steele

IMO 8th order bandpass = looks like a good way to destroy drivers :-)

While the sim suggests that the driver can absorb 1000W before excursion limit is reached, think about what's going to happen when an 8" driver is hit with 1000W. Even 500 W. Hell, even 250 W. T/S params will shift, and the alignment no longer matches the nice Akabak predictions. Then of course there's the nonlinear effects of three vents tuned to different frequencies that are going to mess things up too.

Most woofers can handle very high peak music power inputs, especially for transients only.

As far as "non-linear effects" go, well, I would assume that all the graphs show the response AFTER taking into account the tuning... that's what the math is for, right?

[Brian Steele]

Originally Posted by 454Casull

Most woofers can handle very high peak music power inputs, especially for transients only.

As far as "non-linear effects" go, well, I would assume that all the graphs show the response AFTER taking into account the tuning... that's what the math is for, right?

Wrong.

Most t/s param-based modelling tools model the response to small signals.

E.g. 1W or less.

1kW does not count as a "small signal".

[blakkshepeaudio]

Originally Posted by Patrick Bateman

<snip>
Cons:

• This is a hideously complex design. I am still trying to wrap my brain around it, and it seems that changing one variable affects all the others. For instance, if you change the tuning frequency of one port, it affects the other two ports as well.
• Don't even THINK about building one of these without a woofer tester. Bose stopped making these, and I'll be it's because even a bit of a drift in the QTS of the woofer will change the frequency response in a big way.

Yeah, I am looking into oddball enclosure arrangements ATM and a weakness that many have is a serious dependency on EXACT driver, enclosure and port specification.

Modelling these kinds of enclosures is tricky enough, but prototyping them can be a nightmare! Effective lengths of ports vs physical length, end corrections, tiny air leaks, even changes in enclosure volume due to driver displacement(!), plus many other factors, conspire against the tireless investigator.

Most manufacturers, rightly, can't justify the required substantial R+D for a system that probably won't work for the next batch of drivers they buy-in.

This leaves the field for those companies (eg. Blose) who can finance the R+D, and demand and get the requisite component parameter stability.

And, of course, the enthusiastic DIYer.

The results achievable from some of the more obscure enclosure design techniques are truly amazing and I would not discourage anyone from trying anything. Put saw to wood. Make stuff. That's what we do.

The reasons why commercial outfits don't use some of this tech has little to do with the quality of results, and more to do with the realities of operating a business that has to make LOTS of speakers that all sound the same, usually using components with a fair spread of parameters.

Some oddball ideas are just crap, hampered by fundamental flaws that mean they will never work, or by the originators poor grasp of physics or math. Others are ingenious and very amenable to the stroking of a delicate DIY hand. Yet others are rediscoveries of principles that just got forgotten in the whirlwind surrounding the publication of Thiele and Small's papers.

Anyway, did you build one? How did it sound?