chuffing measurement of rectangular curved walls port and 3d printed port
here are the (slightly less crowded) measurement graphs for chuffing at 1 V, 2 V, 4 V, and 8 V (in fading shades of grey) for the curved wall rectangular port and the 3d printed port, the latter with sanded, smoothed and fillered interior port surface:
both ports have a very good non-chuffing behavior, with none of the typical chuffing resonance peaks even at 8 V input.
no turbulent energy loss leads to no measurable compression.
I once more included the port response for the rectangular port to show the correlation of (higher) harmonic peaks and the response.
Last edited:
air speeds
here are the air speeds for the narrow and wide parts of the curved wall ports, as simulated with hornresp:
C = speed of sound, 340 m/s
| Input voltage | Port air speed (narrow center 8,1 cm²) | % of C | Port air speed (wide ends 18,1 cm²) | % of C |
| 1 V | 2,6 m/s | 0,77 % | 1,0 m/s | 0,3 % |
| 2 V | 5,2 m/s | 1,53 % | 2,0 m/s | 0,6 % |
| 4 V | 10,5 m/s | 3,09 % | 4,1 m/s | 1,2 % |
| 8 V | 21 m/s | 6,18 % | 8,2 m/s | 2,4 % |
| 16 V (no chuffing measured here) | 42 m/s | 12,36 % | 16,3 m/s | 4,8 % |
the straight round port (12,6 cm²) reaches an airspeed of 24 m/s (7,1 % of c) at 8 V.
from the airspeeds and measured chuffing I once more conclude that the chuffing problem is not related to the (narrow, in the case of curved walls) center section of the port but the port ends.
I should mention that while the small 5-inch driver of course has low excurcions at the resonant frequencies of the bandpass chambers it will reach it's Xmax of 5 mm at frequencies between 60 and 80 Hz with around 8 V input.
so it probably makes sense to consider this a sensible limit, also for chuffing.
thanks a lot - it's great to read about experiences from other people doing similar stuff!
I listened to your samples, your port even makes meowing sounds!
I think your "fillet hole felt" sample shows severe chuffing and very strong harmonics (uneven ones probably, simiar to clipping/rubbing).
I experienced very strong H3 harmonics whenever there was only a tiny air leak in the box.
not sure how airtight your port is towards the box interior, with the holes and the felt, so there might perhaps be some distorsion due to this restricted airflow from enclosure to port.
with my test speaker I tried to keep the chuffing noises lower than your second perforated port sample.
I will eventually record some sounds myself, for comparison!
thanks again and have a good night!
one more:
I suspect that the added "chuffing" turbulent air flow is created by the rough surface of the plywood, a not-perfectly smooth trasition between interior flange and port and eventually by sharp, not rounded corners of the port.
the cross sections are the same as for all other curved wall ports and cross section dimensions are even very similar to the 3d printed port.
chuffing measurement of square plywood port
I suspect that the added "chuffing" turbulent air flow is created by the rough surface of the plywood, a not-perfectly smooth trasition between interior flange and port and eventually by sharp, not rounded corners of the port.
the cross sections are the same as for all other curved wall ports and cross section dimensions are even very similar to the 3d printed port.
flat port chuffing graph ...
I thorougly sanded, smoothed and rounded over all interior edges with putty and painted the port with primer.
this resulted in very low chuffing, even if the port surface is increased.
I suppose the 3d printed port with even lower chuffing noise had a slightly smoother interior surface. another reason is probably the difference in interior edge length, thus reduced turbulent "air intake". and the more compact cross section.
my conclusion regarding a slot port:
the longitudinal port resonances are significantly reduced and this can be an ideal solution, as that as long as picking up enclosure resonance pressure peaks from enclosure walls can be avoided.
... and small straight port chuffing graph
as to be expected with its cross section of only 4,9 cm² and the hard edges inside and outside this port produces a very high level of chuffing turbulences, simulated airspeed at 8 V around 34 m/s = 10% of speed of sound. also the compression between 1 V and 8 V is now -1,2 dB (energy lost in turbulent air flow).
in contrast to the other ports the noise is raised over the whole bandwidth and the chuffing resonance peak seems to be similar for different outputs. this may correspond to the observation of low Q resonance of straight port in the roozen paper. I think the high excursion turbulent airflow prevents building up of a high Q resonance.
of course the chuffing of this port was very audible, however it was not as annoying as with other ports, because the resonance "whistling" was mostly masked by wide bandwidth wind noise (edit: this refers to an extended 100 Hz sine tone).
as a side note:
the absolute SPL numbers of different chuffing tests should not be compared. they are only useful as relative values to calculate compression.
in order to compare the absolute SPL levels I would need to make ground plane measurements with 1-2 m microphone distance.
Last edited:
returning to resonance absorber ideas
with these (and several more) results and insights I went on to investigate how an "absorber" (or better called an "additional resonator") mounted to the center section of the port influences the total resonance behaviour and eventually could combine the least chuffing with a non-resonant behaviour.I took the 3d printed port and the flat port and drilled a 2cm hole each in the middle of the port (half port length, for absorbing the fundamental length resonance) and another smaller one at a quarter of the port lenght (for second harmonic resonance).
I found ideal variable length resonators/absorbers using a syringe and a bike air pump.
both are cut open at the "lower" end.
the air pump needed some silicone to tighten the air intake valve of the piston.
both are mounted with airtight epoxy on small boards that can be mounted at flat boards on the ports:
combined resonance behaviour of port lenght and absorber
at first I wanted to investigate the combined low resonance of the port with added absorber as I described in post 58.
resonators (as I learned the hard way) must not be seen as isolated elements but they have a complex combined behaviour.
I re-installed the internal helmholtz resonator for absorbing the enclosure length resonance.
i mounted the the 3d printed port extending to the outside of the box.
I attached only the central air pump absorber, closed the quarter lenth hole and measured the responses of different lenght settings:
observe the port length resonance at 900 Hz with "no absorber" (or more precisely: absorber set to 0 cm lenght) and the changes when extending the air pump.
the pump is acting as simple 1/2 wavelenth "absorber" with a constant cross section.
as can be seen the port length resonance combines with the added "stub" and creates a strong lower resonance. once the "combined" resonance is at about 600 Hz the port lenght resonance appears again.
also it is quite obvious that adding a simple resonator introduces more problems and is not the simple solution as I was thinking some time ago when starting this thread. that idea was to absorb the pressure maximum at half the lenght (here about 10,25 cm) with an absorber closed at one end of that same length (10,25 cm).
have a look at the response at 11 cm extension - it just does not work like that.
Last edited:
This test was with the port internal to the cabinet, so the felt covered holes in the port 'leak' directly in to the cabinet volume. The cabinet itself was decently air-tight, I think.
Very much agree. The cause of turbulence is the change in air speed, not the air speed itself. The more abrupt the change, the more turbulence generated, generally.
This occurs when the reasonably constant air speed of the straight port meets the large air volume of the room and abruptly slows down (pressure rises). In practice, the wave front does not slow down uniformly and we get turbulent flow. When we flare the port ends we gradually reduce the air speed towards the ends of the port, essentially trying to reduce how abrupt the change is at the boundary to the room or cabinet internal.
For your interest, here is a CFD simulation showing air speed of a 4cm diameter straight port with air flow going from left chamber to right chamber. The pressure rise in the left chamber is approx. equal to a 6" woofer with 5mm pk-pk movement. Clearly air speed is highest in the middle. When it slows down to the right we see concentric streams of differing speed that will dissipate energy in sheer friction as the boundaries rub past each other.
By contrast here is the TKE (turbulent kinematic energy). It is concentrated around the concentric / parallel streams of differing air speed, not at the highest region of air speed in the middle of the port. We don't see a lot of turbulence in the middle where the speed is highest because the 'wave front' is uniform and does not cause sheer friction.
I want to show a flared port too, but something is weird with the meshing.
So, I'll follow up on those CFD simulations.
I didn't post the graphic of the flared port because it showed higher maximum turbulence than the straight port and the result was not symmetrical to the port axis. I figured something was not right on my simulation but actually it was simply the intake pressure was too high. It's very interesting because this shows how a flared port can actually perform worse than a straight port at very high air speeds / output levels.
So here is the turbulence energy of straight vs. flared port at a low pressure level:
Here is the same thing but at a higher pressure level. Note the scale has changed:
This is supportive of the research by Roozen.
I didn't post the graphic of the flared port because it showed higher maximum turbulence than the straight port and the result was not symmetrical to the port axis. I figured something was not right on my simulation but actually it was simply the intake pressure was too high. It's very interesting because this shows how a flared port can actually perform worse than a straight port at very high air speeds / output levels.
So here is the turbulence energy of straight vs. flared port at a low pressure level:
Here is the same thing but at a higher pressure level. Note the scale has changed:
This is supportive of the research by Roozen.
Thanks for these very interesting results!
Did you simulate an oscillation or are these constant air movement simulations?
I found out the main issue is the interrupted air flow at the port ends due to high consecutive acceleration/deceleration and the "stalling" effect of abrupt flow cross section change.
Did you refer to the resonance absorber?
I tried stuffing with polyfill (vibrating fibres probably absorb bass output, not recommended) and basotect (more rigid open cell melamine foam with very good results).
I will post measurement results in the next days!
Are you sure that it is not the solver numerical stability problem, but real physical phenomena ?
That doesn't make a lot of sense?
Any round-over will result is less turbulence because the jump in impedance isn't as bad as just straight.
It's only how we define "flaring", because in your pictures the flares goes gradually but all of a sudden very abrupt when it's towards the end. (it has a very small radius there)
In that case we get a sudden jump in impedance again, so in that sense it doesn't surprise me dat we are getting some turbulence again. The radius (in this case it's a compound radius) has to do with this as well.
If you would make that radius at the end much bigger (so I mean on the edge), this would also perform better in sense or turbulence.
The downside is that the tuning frequency will become a lot less predictable and sometimes even start to change depending on the amount or airflow.
What are you simulating this with btw?
Last edited:
That is correct, however the turbulences are starting inside the port, at the big radius curvature, as far as i see it.
I suppose (as i mentioned before) that the simulation shows a constant one-directional air flow and air speeds will get very high (if i am not mistaken and the air flow is actually simulated as oscillating).
Most of the research I've read says that a rounded over end of the port is beneficial - as in the radius between the port and the cabinet. However the question about over-all flare of the port is something that depends on the air-speed / output level because flow separation will happen differently at different air speeds. You can't have one ideal flare for low and high air-speeds. In the 1mbar sim my flared port is 'over-flared' for the air-speed. We might think of it intuitively as a car not being able to take such a tight corner at high speed.
"This study shows that for high exit velocities a slower taper may be required"
https://pearl-hifi.com/06_Lit_Archive/15_Mfrs_Publications/Harman_Int'l/AES-Other_Publications/Maximizing_Performance_from_LS_Ports.pdf
The sim is done in SimScale. It is only a constant air stream from left chamber to right chamber, not a resonant system. It was only intended to show that turbulence occurs at the end of the port, not at the centre where air-speed is highest, but I accidentally showed the same result as Roozen
There is a Comsol paper that shows why optimising for the velocity contour, not the max velocity, is how we maintain low turbulence.
https://www.comsol.com/paper/download/679321/bezzola_acoustics_presentation.pdf
"This study shows that for high exit velocities a slower taper may be required"
https://pearl-hifi.com/06_Lit_Archive/15_Mfrs_Publications/Harman_Int'l/AES-Other_Publications/Maximizing_Performance_from_LS_Ports.pdf
The sim is done in SimScale. It is only a constant air stream from left chamber to right chamber, not a resonant system. It was only intended to show that turbulence occurs at the end of the port, not at the centre where air-speed is highest, but I accidentally showed the same result as Roozen
There is a Comsol paper that shows why optimising for the velocity contour, not the max velocity, is how we maintain low turbulence.
https://www.comsol.com/paper/download/679321/bezzola_acoustics_presentation.pdf