Investigating port resonance absorbers and port geometries

To be perfectly honest, the only thing I would care about, is the first order mode.

The rest will fall well above any practical crossover frequency and therefore won't be really an issue.

It's the first order mode that is hard to push over 2kHz without any tricks.
Those are the spectrum of an 80 hz tone to look at the harmonic distortion and chuffing noise.

Those higher frequency noises seem related to minimum croasection more than the flare, so might be this strouhal effect.

I did multiple levels too, I'll overlay and post tomorrow.
 
  • Like
Reactions: 1 user

stv

Member
Joined 2005
Paid Member
Those higher frequency noises seem related to minimum croasection more than the flare, so might be this strouhal effect.
Minimum cross section does not have any significant (adverse) effect, as far as I have experienced.
A straight not or just slightly flared port with 4 cm diameter (12,6 cm2) behaves much worse in all aspects than a generously flared port with 2 cm minimum diameter (3,14 cm2).

next videos will show this!

(And I still need to listen to your sound recordings!)
 
I'm probably making deductions too soon. To be fair about it, I need to tune all ports to the same frequency which means pushing the minimum flared port walls wider apart or the medium flare walls closer together. This should open up the performance gap.

Changing the flare also changes the resonance modes both of the port and how it interacts with the cabinet which could be having a knock on effect on the turbulence noise also. Very difficult to control all those variables.

Certainly the 'max' flare version of my port is poor in all aspects compared to the medium and minimum flares.
 

stv

Member
Joined 2005
Paid Member
A little bit of info I gathered today on 3 ports - Minimal flare, medium flare, maximum flare. Spectrum is an 80 Hz tone at 90dB 1m.

Here are the audio files:
thanks for the interesting results and mostly the sound examples!
I think all your port variants behave very well, considering the reasonable high SPL and when playing music all the noises will probably be completely inaudible.
would you mind sharing some information about the driver used and the port dimensions?

some remarks, considering my investigations so far:

even if the longitudinal resonance chuffing peaks are "buried" in the harmonic distorsion peaks, they are the annoying and audible noise.
they are NOT related harmonically/musically to the fundamental and mostly, they are being modulated by the 80 Hz fundamental tone, which makes them most audible.
a very rough overlay of your spectrum graphs with very roughly marked chuffing peaks:

chuffing_pk.png
these chuffing noise spectrums make the differences in port sound, as far as I can tell.
also note that some chuffing peaks match harmonic peaks and may be masked and less audible.

as far as I can tell from my investigations the chuffing noise is heavily dependent from the port terminations.
the port terminations are responsible for (small) turbulent airflow at the port edges that again induces longitudinal resonance ("blowing noise", just like a pan flute; this is not the vortex shedding, happening at very high displacements). here the exterior termination has the advantage of being flush mounted to the enclosure, thus less turbulent airflow "around the port edge".

therefore I think the differences of your ports do not so much depend on the flare rate or the central spacing, but much more on the different internal port endings:
tenson_flares.png
the "minimum" port has the best internal ending, having more than 90 degrees of flange bending on both sides. I suspect that is the reason for it's best behaviour.

I also think that you could try minimizing your port, for example by including a curved width. by doing so you would reduce the port lenght and improve it's resonant behaviour (better "diameter"/length ratio).

As I heard in your sound examples and can be seen in the spectrum graphs, there is no sign of vortex shedding creating wide band noise.
the exit cross section surface and thus the "virtual port radius" is probably high and for 80 Hz the particle displacement also should already be quite low.
thus the "strouhal number" will be quite high (good).
 
Last edited:
  • Thank You
Reactions: 1 user
Those are the spectrum of an 80 hz tone to look at the harmonic distortion and chuffing noise.

Those higher frequency noises seem related to minimum croasection more than the flare, so might be this strouhal effect.

I did multiple levels too, I'll overlay and post tomorrow.
Oh right sorry, for a second I thought we were still looking at modes/standing waves.
 

stv

Member
Joined 2005
Paid Member

slow motion port air movement video 2: big flared port​

here is video number 2:
again a 4 cm cardboard tube, but with a generous flare and roundover at the exterior end.
you can find a drawing in post #103.


and the correlating chuffing spectrum measurement graph:

240203-P2-flared_big.png

a clear correlation between visual turbulence and the wide band noise in the spectrum can be seen!
up to 4 V the air movement in the port opening seems to stay visibly laminar and does not produce broadband noise.
the longitudinal resonance chuffing noise peaks at 650 and 1300 Hz are most likely generated at the small rondover at the interior end of the port.
once the air movement gets turbulent (big circular movements at port exit) there is noise shown in the graph from 20 Hz up to around 3 kHz (for 8 V) and further up (for 16 V).
of course and again, the higer SPL do not seem to be usable for any decent music reproduction!

compare the turbulences and the laminar air flow to the "hard edge tube port"!
I did not use water mist for this port because the air motion seemed visible enough with the flour particles.

for the hard edge port I tried to see small turbulences at the sharp port edge using water mist - I'll have to watch that video again...
 
Last edited:
  • Like
Reactions: 1 users
thanks for the interesting results and mostly the sound examples!
I think all your port variants behave very well, considering the reasonable high SPL and when playing music all the noises will probably be completely inaudible.
would you mind sharing some information about the driver used and the port dimensions?

some remarks, considering my investigations so far:

even if the longitudinal resonance chuffing peaks are "buried" in the harmonic distorsion peaks, they are the annoying and audible noise.
they are NOT related harmonically/musically to the fundamental and mostly, they are being modulated by the 80 Hz fundamental tone, which makes them most audible.
a very rough overlay of your spectrum graphs with very roughly marked chuffing peaks:

View attachment 1269379
these chuffing noise spectrums make the differences in port sound, as far as I can tell.
also note that some chuffing peaks match harmonic peaks and may be masked and less audible.

as far as I can tell from my investigations the chuffing noise is heavily dependent from the port terminations.
the port terminations are responsible for (small) turbulent airflow at the port edges that again induces longitudinal resonance ("blowing noise", just like a pan flute; this is not the vortex shedding, happening at very high displacements). here the exterior termination has the advantage of being flush mounted to the enclosure, thus less turbulent airflow "around the port edge".

therefore I think the differences of your ports do not so much depend on the flare rate or the central spacing, but much more on the different internal port endings:
View attachment 1269382
the "minimum" port has the best internal ending, having more than 90 degrees of flange bending on both sides. I suspect that is the reason for it's best behaviour.

I also think that you could try minimizing your port, for example by including a curved width. by doing so you would reduce the port lenght and improve it's resonant behaviour (better "diameter"/length ratio).

As I heard in your sound examples and can be seen in the spectrum graphs, there is no sign of vortex shedding creating wide band noise.
the exit cross section surface and thus the "virtual port radius" is probably high and for 80 Hz the particle displacement also should already be quite low.
thus the "strouhal number" will be quite high (good).
Thank you STV, that is appreciated input! That color coding on the port edge is a nice touch :giggle: You might well be correct about the cause, the minimum flare port is more symmetrical in all aspects. I will try extending the shorter radius on the internal end of the medium flare port and see what happens. I can also push a few dB harder to see if that opens up the performance gap more.

The shorter internal radius is that way due to interference from some electronics that will be placed in the back of the enclosure at a later time. I therefore flared the left internal side more so the total flare is the same as the top external end of the port. I'm not sure if that is an optimal approach, it might be the reason the maximum flare performs badly if the rate of growth on one side only is too high.

I feel this discussion really now belongs in my design study thread not to mess up the flow of this one. Shall I ask a mod if they can move these last couple of posts?

P.S. the driver is the Dayton ND91-4 and the port has dimensions of approximately 100 mm width and 160 mm length.
 
Last edited:
  • Like
Reactions: 1 user

stv

Member
Joined 2005
Paid Member
A quick note regarding distorsion of ports:
Taking the sensible and still non-turbulent 35 Hz port output at 4 V, using the flanged port shown in the video above, the port (edit: of course the entire speaker) produces:
H2: 1,4 %
H3: 0,26 %

The measurements of hificompass of a similar size but much better driver (dayton RS-125) show the following distorsion numbers for frequencies at 100 Hz and 4 V (315 mm distance):
H2: -30 dB (= 3,2 %)
H3: -35 dB (= 1,8 %)

Distorsion rises further below 100 Hz.

So, I guess the port is usually no relevant contributor to low frequency distorsion. On the contrary: bass distorsion will usually be minimum at the port output maximum, unless the port output becomes turbulent or the driver is exceptionally good.
 
Last edited:
a clear correlation between visual turbulence and the wide band noise in the spectrum can be seen!

Turbulence from vortex shedding has a particular spectrum that follows from the physics of turbulence. These plots do not follow that shape but some of your earlier plots or parts of them did. It is likely some noise from turbulence is present but it doesn't seem to be dominant. There will be noise from the acoustics of what is going on within the cabinet and leaking out the port. There will be noise from the driver vibrating the cabinet. I suspect the vibration of the port itself may be significant driven by not only the driver but also the fluid motion (vortical and acoustic) within the cabinet and port. This is in addition to the expected tones from the helmholtz resonance, motor distortion and port acoustic resonances. I would suggest calling it broadband noise while waiting for further information about it's source.
 
  • Thank You
Reactions: 1 user

stv

Member
Joined 2005
Paid Member
These plots do not follow that shape but some of your earlier plots or parts of them did. It is likely some noise from turbulence is present but it doesn't seem to be dominant.
Thanks for your explanation!

I think maybe now I understand - the low frequency noise in the 50 Hz plot of post #349 is sign of vortex shedding.

If that is correct then only the 16 V plot in the 2nd video post does eventually show signs of vortex shedding, added to the "broadband noise", see slight emphasis on low frequency noise that the hard edge port does not show.

I suspect the "broadband noise" may be result of chaotic (thus broadband) turbulent airflow outside the port.

There surely is a combination of several effects going on!
 
The highest flow velocities are in the port and at the lower voltages they are too small for the flow in the port to be turbulent. The flow velocities away from the port are going to be smaller and the flow will not be turbulent. Aerodynamic sound sources from turbulence tend to be quadrupoles and relatively weak compared to monopole and dipole sources. Monopole, dipole and quadrupole sound sources scale differently with velocity which can be a useful way to identify which sound generating mechanisms are present.
 
  • Thank You
Reactions: 1 user

stv

Member
Joined 2005
Paid Member
so might be this strouhal effect
just a small remark: the strouhal number must not be interpreted as a singular physical effect.
thus it is important to keep in mind that the strouhal NUMBER is a rule of thumb that includes several effects.

I also noticed in my experiments that the noises produced by the port/the helmholtz resonator will be unbearable, at least for any serious music reproduction:

1) long before the strouhal number will be near 1 or below, thus when an "unsteady separation of acoustic flow" (roozen) happens
2) also long before any vortex shedding will happen (see post #412 by andy19191 for explanation of characteristics of vortex shedding)
 

stv

Member
Joined 2005
Paid Member
To be fair about it, I need to tune all ports to the same frequency

That's the part where it becomes all tricky 😉
yes!
the advantage of a simple tube port is that it can be calculated with simple equations and correction factors.
the effective weight of the air in the port is clearly defined and ends abruptly at the port terminations.
this is also the main weakness of such a design.
how should the velocity of the air in the port turn immediately into pressure? it can't.

when designing a velocity-pressure-transformer (a flared/rounded port transition) it is not that clear how to calculate the weight of the air in the port, because the velocity decreases steadily at the terminations.
also, depending on the SPL the tuning may change. this will of course also happen with the tube-port, by the way.

anyhow, I built some port variants resulting in the same tuning for chamber 2, see post #103.
for chamber 1 the port designs in these posts lead to the same tuning: #157, #228 and #328.

as far as I have experienced the cross section (surface area) development of the port is most relevant.
the shape (round, square, rectangular, flat, two curved walls, four curved walls) does not make a big difference for the tuning.
 
Last edited:
Turbulence from vortex shedding has a particular spectrum that follows from the physics of turbulence. These plots do not follow that shape but some of your earlier plots or parts of them did. It is likely some noise from turbulence is present but it doesn't seem to be dominant. There will be noise from the acoustics of what is going on within the cabinet and leaking out the port. There will be noise from the driver vibrating the cabinet. I suspect the vibration of the port itself may be significant driven by not only the driver but also the fluid motion (vortical and acoustic) within the cabinet and port. This is in addition to the expected tones from the helmholtz resonance, motor distortion and port acoustic resonances. I would suggest calling it broadband noise while waiting for further information about it's source.
Can you maybe help me understand by highlighting the part of the spectrum in an earlier post you think is associated with turbulence?

You mentioned the lower voltages are not enough to be turbulence and that seems to be shown in the spectrum of post 407 just up the page where only the 8V (red) and 16v (green) really show an increased broadband noise floor. The spikes at 650 Hz and 1300 Hz are of course modes of some sort and not pure turbulence.

The air particle motion in the video also looks pretty chaotic at 8V and 16V, is that not turbulence?
 

stv

Member
Joined 2005
Paid Member
is associated with turbulence
Maybe andy will answer himself, but here are my 2 cents:
All noise is produced by turbulence.
Vortex shedding (as a special case of non-chaotic turbulence) has a characteristic spectrum.
But, as I mentioned earlier, several other nasty turbulent effects will happen long before vortex shedding.
My next videos (or at least one of them) will include vortex shedding, if i understood the term correctly.

Also see my explanation attempt here.
 
Maybe andy will answer himself, but here are my 2 cents:
All noise is produced by turbulence.
Vortex shedding (as a special case of non-chaotic turbulence) has a characteristic spectrum.
But, as I mentioned earlier, several other nasty turbulent effects will happen long before vortex shedding.
My next videos (or at least one of them) will include vortex shedding, if i understood the term correctly.

Also see my explanation attempt here.
Andy's third sentence seems to be talking about turbulence in general. I might have misunderstood.

Andy said:
It is likely some noise from turbulence is present but it doesn't seem to be dominant.

650 Hz seems very low for a port mode unless the port effective length is about 25 cm. I'm currently discovering much of what I thought was port modes in my small speaker are complex cabinet modes. Today is dedicated to work on that :)