# How good is your port

```How good is your port ?
By Bohdan Raczynski
Ported enclosures are known to extend the low frequency output of the
loudspeaker system by exploiting Helmholtz resonator created by the compliance of the
air inside the enclosure and inertance of the air in the port. Acoustic transformer created
this way has its own resonant frequency, Fb, at which most (or all, if there was no losses)
of the system acoustic output comes from the port. A fairly obvious implication of this is
significant velocity associated with the air flow through the port. This in turn, causes all
sorts of non-musical noises to be generated by the port and also distortion and acoustic
compression. Depending on port geometry, and required low frequency SPL of the
system, the issue can be quite significant.
The problem described above belongs to rather complex field of fluid flow.
Assuming incompressible flow, some sophisticated FEM programs would be able to
model air turbulence and associated vortex shedding in more detail3, but this is well
beyond the scope of my article. Fortunately, existing research results, design material and
tests results enable us to formulate an approximate macro view of the problem and look
at the acoustic impedance of the port under high air velocity. I would like to stress, that
the approach I am taking here is a significant simplification of the physics of the problem.
However, the resulting model is quite useful and finds confirmation in practical tests1.
Intuitive approach to port non-linearity
We are all familiar with the need for good carpentry skills when building speaker
boxes. Accuracy of joints and sealing the box is essential for proper operation of all types
of boxes, be it sealed, vented, passive radiators and so on. Sealed box means exactly this
– the air inside the box is trapped and sealed from the external world. There is no
parasitic or accidental leakage from the box, so that mathematical model developed for
the enclosure continues to be accurate. Consequently, the box Qb factor is controlled by
the designer and NOT by sloppy workmanship.
Vented enclosure also needs to be “air tight”, of course with the exception of the
purposefully introduced opening – port. Just as for the sealed box, the cabinet needs to be
sealed, so that no accidental air leakage from the box can occur. Properly executed design
would include all sorts of seals and gaskets to ensure, that connector boards or drivers
themselves do not cause air leakage. Assuming you have your perfect vented box build,
you may expect that the SPL curve and input impedance curve will look as on Figure 1.
The impedance curve is very familiar and has two characteristic peaks with the dip
between them. The dip is located exactly on the box tuning frequency, Fb = 25Hz. My
design is a QB3 type with system parameters as shown on Figure 1, and I have also
assumed, that Qb = Qp = 1000, so I have a very low-loss design. Port diameter in this
example is 15cm or 6 inches and the input power to the system is 1watt.
1
Figure 1 SPL and input impedance @ 1watt, port diameter = 15cm.
Now, assume, that I start reducing port diameter. Initially, the picture will not
change by much. However, when the port eventually becomes very small, intuitively,
there should be very little difference between the vented box with a “very small port” and
sealed box with a “large leakage” problem4. In this case, you would expect than the SPL
curve will resemble that of a leaky sealed box and the input impedance curve will loose
the lower peak and will become a “single peak” curve just like the sealed boxes have.
Between those two extremes, you may expect problems commonly labeled as “port nonlinearity”.
Historical research on fluids in tubes
Bies and Wilson8 actually experimented with an orifice (more like a decent port)
9cm (3.5”) in diameter and 2.9cm (1.13”) long. They found, that the acoustic resistance,
R, of the tube varies with the particle velocity in a similar way as shown on Figure 2
Thurston11, working with fluid flow through circular tubes found that analogous
acoustic resistance, R, and inductance, M, again vary in similar way as shown on Figure
2.
A landmark paper by Ingard10 offers an empirical formula for the non-linear
component of the acoustic resistance of a tube. Later on, this work has been expanded by
Ingard and Ising9 who offered more complete mathematical treatment of orifice
behaviour.
2
Figure 2. Acoustic impedance R+jωM of a short port.
Backman1, experimenting with port non-linearity, plotted driver’s input
impedance for various input voltages and determined, that the most sensitive to the
amplitude variations is the magnitude of lower impedance peak. He pointed out, that the
behaviour of the vented enclosure approaches a sealed enclosure at high levels. The inner
diameter of the port was 102mm (≈4”) and the length of the port was 168mm (≈6.6”).
Vanderkooy2 measured and analyzed ports performance at high levels and
proposed a simplified nonlinear model of a port. The comprehensive analysis included jet
formation, acoustic compression, turbulence noise and distortion.
Simplified view of the problem
Typically, the area of the cone is many times larger than the area of the port.
Since the volume of the air pushed via the port needs to be approximately equal to the
volume displacement of the cone, the only way to accomplish this, is to push the air in
the port much faster. At low SPL levels, the volume displacement of the cone is small
and consequently, the cone velocity is low as well. This, in turn, stipulates low air
velocity in the port or what is known – a laminar air flow. The process is characterized by
a low Reynolds number, Re<2000, and there is no turbulence in the air flow.
Re = (r*Vmax*ρ)/µ
Where Vmax is the peak flow rate, ρ is the density of, µ is the viscosity of air, and r is the
radius of the port. At this point of time, the acoustic resistance of the port, Rp, (port loss)
is linear and is relatively low (50-200ohm).
3
Also, the mass of air in the port, Mp, is constant. Remember, that mass of the air
in the port resonates at Fb (box tuning frequency) with the compliance of the air inside
the box, halting the movement of the cone. If there were no losses, at resonance, Fb, all
acoustic output of the system would come from the port.
Now, lets start reducing the port diameter. As we can guess, the volume
displacement of the cone remains the same, so the air in the port must move faster and
faster. Soon, the turbulence occurs in the air flow, jets of air are forming and the process
is no longer linear. Reynolds number corresponding to the onset of turbulence is Re ≈
20000. When the Reynolds number hits 50000, your vent is compressing. It is now more
and more difficult to push the air in the vent. The acoustic output of the port is below the
expected level (if the process continued to be linear) and you begin to experience “port
compression” effect. Port impedance, Zp, has now additional component relating to
turbulent air flow and depending on the air velocity, Rp(v). Also, the Mp has changed as
well, drifting towards 60-70% of its original value and becoming velocity dependant.
The Rp(v) is an interesting element. If plotted in frequency domain, it would
resemble the curve depicting port air velocity, shown on Figure 3. This should be of no
surprise, as Rp(v) is so heavily dependant on the port air velocity. In the frequency
domain, the Rp(v) will quickly reach its peak not far away from the lower peak of the
input impedance curve. Depending on the geometry of the port and the volume
displacement of the driver, the air velocity in the port may reach 50-100m/s. Such a high
values of the air flow result in Rp(v) reaching 5-10kohm levels.
Figure 3. Port air velocity (curve 1) for port diameter = 5cm.
4
The “double-peak” impedance curve is a clear result of the port action. The port
air velocity curve plotted on Figure 3 clearly indicates, that the lower impedance peak
will be much more affected by the port non-linearity than the upper impedance peak.
Therefore, the Rp(v) needs to be incorporated in the driver’s model.
Changes in driver’s model
Vented enclosure shown of Figure 4, provides different loading for the back of the
diaphragm, as compared to the sealed box. The vibrating system and front loading of the
diaphragm are represented on the mechanical mobility circuit the same way as for the
sealed enclosure.
Introduction of the vent adds several more components such as: (1) mass of the air
in the port Mmp and its losses Rmp and (2) radiation impedance of the port represented
by Rmrp and Mmrp. The air in the port is treated as a mass because of its small volume
and more importantly, because it is incompressible. Particles of air will move on both
sides of the vent with the same velocity. The air compressed in the box by the back side
of the diaphragm has only one path to escape - pushing the air mass trough the vent.
Therefore, the pressure path consists of series connection of Cmb, representing compliant
air in the box and the four elements of the port.
Since the air in the port is incompressible, the immediate layer of air in front of
the box (radiation impedance) will be connected to the same velocity line as the entry to
the port inside the box. The other end of the masses is connected to the U=0, or reference
velocity as required in mechanical mobility circuits.
The mechanics of the above process can be easily demonstrated on a physical
model of a vented box. Connecting a small (1.5V) battery across vented box terminals,
we can displace the cone in or out of the box. Small air-flow detecting device (candle)
positioned in front of the port will show significant air movements, in the direction
opposite to the diaphragm. The volume of air displaced by the cone should be similar to
the volume of air leaving the port. If the difference is significant, than leakage losses are
significant.
The above experiment clearly shows the pressure (current) path in the mechanical
mobility model, so it should now be easy to explain why the compliance of the box is
connected in-series with port elements. It is observable, that Cmb and Mmp+Mmrp form
series resonant circuit in the mechanical mobility representation. The circuit will act as a
“selective short circuit” for the volume velocity Uc, shorting it to U=0 (ground) at the
circuit resonant frequency. Because of the circuit losses, the short is not perfect, but
velocity Uc will be much reduced. In the practical system this situation translates into
much reduced cone excursion at the box resonant frequency.
Acoustical impedance representation shows Cas and Map+Marp forming parallel
resonant circuit. Electrical circuit theory advocates that very little energy (current) needs
to be fed into the circuit for it to resonate and for the current (volume velocity) in the
resonant circuit to be still very high.
5
Therefore, volume velocity in the “feeding” branch, which contains diaphragm
output will be very small and volume velocity in the resonant circuit containing port will
be high. This effect, although the strongest on the resonant frequency Fb, will extend
over some narrow frequency range and on the low-end side creates extended system
output. It is the enclosure/port resonance effect, that is being exploited here to augment
system output at low frequency. Figure 4 shows mechanical mobility (top diagram) and
acoustical impedance (bottom diagram) representation adopted for the vented enclosure
model. The components are:
Cas, equivalent compliance volume Vas transformed to acoustical side.
Mad, mass of the vibrating system Mms transformed to acoustical side.
Ras, vibrating assembly loss Rms transformed to acoustical side.
Mar+Mab, air radiation of the front side of the diaphragm + air load of the back
side of the diaphragm.
Rar, air radiation of the front side of the diaphragm.
Cab, enclosure compliance Vab transformed to acoustical side.
Rab, absorption losses of the enclosure transformed to acoustical side.
Map, mass of the air in the port.
Rap, frictional losses in the port.
Figure 4. Modified model of the vented box including Zp(v) = Rp(v) +jωMp(v)
The non-linear port impedance was implemented as Zp(v) = Rp(v) +jωMp(v) in the port
branch. Please note, that Map+Mp(v) will exhibit slight reduction in value as the air
speed increases and Rap+Rp(v) will exhibit significant increase in value as the air
velocity in the port increases.
6
Resulting performance
In order to gain some insight into system performance affected by port nonlinearity problems, I plotted the SPL for a port of 5cm in diameter for 1W (curve 0), 10W
(curve 1) and 100W (curve 2) input power – see Figure 6. As we can see, with the
increased input power, there is a sort of “saddle” developing on the SPL curve around the
box tuning frequency of 25-30Hz. This is exactly the frequency range, where we would
expect the port to contribute most to the system SPL. Our small port is clearly not
performing as anticipated.
Figure 6. SPL for port diameter of 5cm, @1W, 10W and 100W input power
Next, I modeled SPL for the same power levels, but this time, I used larger port,
15cm in diameter. The resulting plots on Figure 7 do not exhibit the “saddle” any more
for 100W power level. Clearly, as the input power is increased, the SPL curves go up,
maintaining the approximate shape acquired at 1W power level.
It is also easy to observe, that the SPL curves now have resonant peaks above
200Hz, not seen on Figure 6. This is the result of enlarging the diameter of the port. You
may remember, that larger port must also be longer if tuned to the same frequency. Now
the length of the port is such, that the self-resonances of the port tube fall into much
lower frequency range – just to be displayed on the screen.
7
Figure 7. SPL for port diameter of 15cm, @1W, 10W and 100W input power
In order to estimate “port compression” effect, I plotted the SPL of the two ports
on the same screen – see Figure 8. Visual inspection of the graphs shows about 5dB of
port compression at Fb = 25Hz. Well, I have just about lost my vented box performance
gains if I was to use the smaller port.
In the next step, it would be interesting to compare the input impedance plots to
see if the lower impedance peak, characteristic for the vented enclosure, indeed
disappeared from the plots. The answer is clearly evident on Figure 9, where the input
impedance for three ports is plotted at 100W power level. The 15cm vent exhibits the
expected “double peaks” curve, however the compression is still registered on the
impedance plot. Ideally, the port should still be larger. The Reynolds number calculated
for these conditions is Re = 68000. Therefore, the port is indeed compressing slightly.
However, the 10cm port has the lower impedance peak significantly reduced. This
is a sure sign, that this port is too small for the job.
The severely undersized 5cm port produces “single peak” impedance curve for
100W input power. This port would also prove to be inadequate for 1 watt of input
power. The Reynolds number calculated for 1W conditions is Re = 20000, which is a
clear indication, that the port becomes turbulent. Indeed, the corresponding input
impedance plot shown on Figure 10, is a clear indication of the non-linearity problem @
1 watt for this port.
8
Figure 8. SPL for port diameter of 5cm (curve 0) and 15cm (curve 1) @100W
Figure 9. Impedance @100W for port diameter of 15cm,10cm and 5cm.
9
Figure 10. Input impedance @1W for port diameter of 15cm, 10cm, 7.5cm and 5cm.
Remedies
First and foremost, the problem is related to air velocity in the port. As we know12, air
velocity in the port, Vp, depends on volume velocity Up, via the port branch divided by
the port's area, Sp.
Vp = Up/Sp
Since the area Sp = πr2, it is easy to observe, that air velocity in the port depends on the
inverse squared of the port's radius. It seems rather obvious, that port radius should be
kept as large as possible.
Dickason13 recommends 15cm (6") ports for 15" woofer as a minimum and 10cm (4”)
ports as a minimum for 12” woofers.
Salvatti4 offers a number of recommendations relating to port geometry. This includes
balancing inlet and exit flows by using different tapers for inlet and exit.
Roozen3 advocates port geometry based on 6 degrees diverging contour towards the ends
of the port. The inlet and outlet are rounded with relatively small curvature.
10
Conclusions
A simple modification of acoustic impedance of the port in the vented enclosure has been
described. The modification consists of adding extra components such as Mp(v) and
Rp(v) which are dependant on air velocity inside the port. The modification reflects
acoustic compression of the port and shows the changes to the input impedance of the
driver under high SPL levels. The changes in input impedance reflect experimantal
findings1.
Looking at the performance of the 15cm (6”) vent, I was surprised to see the compression
evident at 100Watt input power. The vent seems quite large and to tune it to 25Hz in
120liter box, I would have to make it nearly 60cm (23.6”) long. This could be a
construction challenge, and remember, this vent is still not completely linear at high SPL.
This brief analysis of the vent performance clearly indicates, that nearly all practical size
vents will compress at higher SPL levels. Loudspeakers intended for high power stage
applications should be give very careful consideration regarding port non-linearity. For
this type of application, port compression problem will tend to further degrade system
SPL, already affected by the thermal compression during prolonged stage performance.
References
1. J. BACKMAN, The Nonlinear Behaviour of Reflex Ports. AES preprint 3999.
2. J. VANDERKOOY, Nonlinearities in Loudspeaker Ports. AES preprint 4748.
3. N.B ROOZEN, J.E.M. VEAL, J.A.M. NIEUWENDIJK, Reduction of Bass-Reflex Port Nonlinearities
by Optimizing the Port Geometry, AES preprint 4661.
4. A. SALVATTI, D. BUTTON, A. DEVANTIER, Maximizing Performance from Loudspeaker Ports,
AES preprint 4855.
5. G. TRURSTON and others, Nonlinear Properties of Circular Orifices, JASA, Volume 29, Number 9,
1957.
6. A. NOLLE, Small-Signal Impedance of Short Tubes, JASA, Volume 25, Number 1, 1953.
7. G. THURSTON, C.MARTIN, Periodic Flow through Circular Orifices, JASA, Volume 25, Number 1,
1953.
8. D. BIES, O. WILSON, Acoustic Impedance of a Helmholtz Resonator at Very High Amplitude, JASA,
Volume 29, Number 6, 1957
9. U. INGARD, H.ISING, Acoustic Nonlinearity of an Orifice, JASA, Volume 42, Number 1, 1967.
10. U. INGARD, On the Theory and Design of Acoustic Resonators, , JASA, Volume 25, Number 6, 1953.
11. G. THURSTON, Periodic Fluid Flow through Circular Tubes, JASA, Volume 24, Number 6, 1952.
12. R. SMALL, Direct Radiator Loudspeaker Analysis, An Anthology of Articles on Loudspeakers from
JAES Vol. 1 -Vol. 25 (1953-1977)
13. V. DICKASON, Loudspeaker Design Cookbook, 5th Edition, Old Colony Sound Lab. 1997.
All graphics included with this article were generated using SoundEasy V3.32 computer program.
11
```