One thing that puzzles me about recommendations for the port size is that it is focused on the air velocity.
As far as I understand it, the problem with port noise ("chuffing") occurs when the flow in the port becomes turbulent. As I understand it from the little I know about flow mechanics the border between turbulent and laminar flow is largely determined by the port cross-sectional geometry and Reynold's number. For different port diameters these factors will result in different air velocities.
Is the fixation with the air velocity a result of that most people use vent diameters of approximately the same size? Or am I completely wrong about the Reynold's number?
Oops, did I get OT now again?
As far as I understand it, the problem with port noise ("chuffing") occurs when the flow in the port becomes turbulent. As I understand it from the little I know about flow mechanics the border between turbulent and laminar flow is largely determined by the port cross-sectional geometry and Reynold's number. For different port diameters these factors will result in different air velocities.
Is the fixation with the air velocity a result of that most people use vent diameters of approximately the same size? Or am I completely wrong about the Reynold's number?
Oops, did I get OT now again?
Hi there Svante,
I was tempted to post earlier in this thread, but the jostling of egos didn't leave much room. Anyway, your query has tempted me out.....
You are correct about the Reynolds number, but the maths quickly becomes complicated, particularly for flared ports.
Most design packages will give you either a peak velocity figure or a graph of velocity vs frequency. This can be used to select your port if you know the rules....
Turbulence can occur as the air in the boundary layer (next to the port wall) exits the port too quickly. Adding a correctly sized flare causes a controlled expansion which can prevent "chuffing"
A second source of turbulence occurs when the air in the "core" of the port becomes turbulent due to excess velocity. This cannot be fixed with flares and requires moving to a larger diameter.
I've found that these two processes have their own rules about how usable velocity varies with port diameter, flare size and frequency.
This was written up and briefly tossed about in an earlier thread:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=75510&highlight=
or go direct to the research:
http://www.users.bigpond.com/bcolliso/flare-testing.htm
regards
Collo
I was tempted to post earlier in this thread, but the jostling of egos didn't leave much room. Anyway, your query has tempted me out.....
You are correct about the Reynolds number, but the maths quickly becomes complicated, particularly for flared ports.
Most design packages will give you either a peak velocity figure or a graph of velocity vs frequency. This can be used to select your port if you know the rules....
Turbulence can occur as the air in the boundary layer (next to the port wall) exits the port too quickly. Adding a correctly sized flare causes a controlled expansion which can prevent "chuffing"
A second source of turbulence occurs when the air in the "core" of the port becomes turbulent due to excess velocity. This cannot be fixed with flares and requires moving to a larger diameter.
I've found that these two processes have their own rules about how usable velocity varies with port diameter, flare size and frequency.
This was written up and briefly tossed about in an earlier thread:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=75510&highlight=
or go direct to the research:
http://www.users.bigpond.com/bcolliso/flare-testing.htm
regards
Collo
While velocity has something to do with turbulence, I don't think there is any way you could ever call any port flow anything but turbulent. Reynolds number in pipes typically applies more to fully developed, steady flow. Port airflow is unsteady and not fully developed, so not laminar even in so-called "laminar" velocity ranges. It has been a while since fluid mechanics, but these dimensionless ratios do apply in some form to more complicated situations, but the relations are usually raised to some power that is empirically determined.
The determining factor in noise generation is not so much turbulence, but separation of flow and eddy production - which may be related to reynolds number, but it is not a trivial matter to predict. This is why there are rules of thumb. Avoiding flow separation is the ticket, not avoiding turbulence. That is how the flare works. That is why a golfball has dimples....
Certainly you can limit turbulence by making a bunch of straws and dropping the "reynolds number", but that isn't a pure mass as modeled in T/S theory, it is a mass with distributed resistance due to all the viscous stuff (since when is c.r.a.p. a bad word) going on.
BTW, collo has some interesting data up on his site. I wouldn't exactly call it rigorous, but it is better than almost anything else available...
The determining factor in noise generation is not so much turbulence, but separation of flow and eddy production - which may be related to reynolds number, but it is not a trivial matter to predict. This is why there are rules of thumb. Avoiding flow separation is the ticket, not avoiding turbulence. That is how the flare works. That is why a golfball has dimples....
Certainly you can limit turbulence by making a bunch of straws and dropping the "reynolds number", but that isn't a pure mass as modeled in T/S theory, it is a mass with distributed resistance due to all the viscous stuff (since when is c.r.a.p. a bad word) going on.
BTW, collo has some interesting data up on his site. I wouldn't exactly call it rigorous, but it is better than almost anything else available...
There is zones between laminar and turbulent air flow. Till Re=2320 its laminar, from Re=2320-4000 its something in between, a more than Re=4000 is turbulent.
Reynolds number Re=v*g/Mi , v - air velocity ; g - acceleration of gravity ; Mi - cinematic viscosity of the air
Air velocity depend of frequency and power & X - excursion. At same frequency but more excursion - bigger velocity of the cone - bigger velocity of the air. Approximate: Sd*Vc=So*Va, Sd - cone area; Vc - average velocity of the cone ; So - tunnel area ; Va - avarage air velocity in BR tunnel
When there is flow of fluid near any walls (like tunnel) there is a specific flow near walls called border flows. And it is different (makes friction) than in rest of the tunnel.
Really complicated for simulation.
Point is to take a port with diameter which cant throw to v=34 m/s (1/10 or 10% of velocity of the sound) with nominal power (or max. SPL you need).
If port lenght is to big, there is way to go with smaller diameter. Just put the straws in, across all tunnel area (and lenght). That is the way to reduce air velocity, because of much bigger area of close walls and border flows (much bigger wall friction).
Reynolds number Re=v*g/Mi , v - air velocity ; g - acceleration of gravity ; Mi - cinematic viscosity of the air
Air velocity depend of frequency and power & X - excursion. At same frequency but more excursion - bigger velocity of the cone - bigger velocity of the air. Approximate: Sd*Vc=So*Va, Sd - cone area; Vc - average velocity of the cone ; So - tunnel area ; Va - avarage air velocity in BR tunnel
When there is flow of fluid near any walls (like tunnel) there is a specific flow near walls called border flows. And it is different (makes friction) than in rest of the tunnel.
Really complicated for simulation.
Point is to take a port with diameter which cant throw to v=34 m/s (1/10 or 10% of velocity of the sound) with nominal power (or max. SPL you need).
If port lenght is to big, there is way to go with smaller diameter. Just put the straws in, across all tunnel area (and lenght). That is the way to reduce air velocity, because of much bigger area of close walls and border flows (much bigger wall friction).
Ron E said:
A resistance is rather easy to model and while it can be a problem if one uses "straws" in the port, using multiple small diameter ports would IMO not make the resistance a problem. I have this figure in one of my textbooks, it shows the frequency at which the resistance and the impedance of the mass are equal, for different port diameters. Assuming that fh is a few tenths of hertzes (?) and that the Q of the resonator also should be a few tenths it seems as if any diameter larger than a centimeter (which is off the scale, but you can realize it anyway) would do just fine.
However, seen in the light of the text below, maybe it won't work anyway...
An externally hosted image should be here but it was not working when we last tested it.
Ron E said:
Collo said:
Interesting that you both look at the turbulence inside the pipe and the vortices at the port opening as different phenomena, I never thought of doing that, but it makes perfect sense. That would speak against using Re as a (simple) predictor of port noise. Hmm.
Collo said:
Thanks, I'll have a look at that!
Edit: now I have had a look at your webpage, and I am impressed. At first I had some doubts about doing a perceptual evaluation of when "chuffing" starts, but since this is an empirical study, and since the end goal is to know when the noise becomes audible, I am convinced that it is a good way to go.
Thanks for the good work!
Beranek states that a tube is a mixed mass resistance element for a>.01/sqrt(f) and a<10/sqrt(f)
a is the radius.
f=30
a=1.8mm to 1.8 meters.
While this can be lumped, it is not simply a mass and resistance, the resistance is distributed - a transmission line type of situation. Earl Geddes describes ports in his 1988 JAES article on bandpass enclosures as a mass with a distributed resistance. At even higher frequencies the pipe impedance obviously becomes an issue.....
Grab Bullock's old bandpass boxmodel and see the dramatic differences between regular ports (not filled with straws) and "ideal" ports, when used in BP enclosures.
------------------------------------------------
As far as the "central flow" being considered separately from the flare, it is only due to separation of flow at the entrance/exit. There is a subtle difference between the way I and collo think about this - I am considering pipe flow vs. entry and exit flow. I interpret collo as considering the flare as separate from the extension of the pipe which is attached to it. Port compression is not somehow due to higher turbulence in the central flow Turbulent air actually flows better (more mass transfer for a given average velocity) than laminar because the velocity profile is flatter. As I said before, the flow in ports is not steady or fully developed so it lies clearly in the turbulent regime, no matter what the reynolds number.
BTW, the air is not necessarily being compressed in the port and released in the flare. Air in subsonic flows is essentially incompressible. The typical velocity used in engineering calcs for truly compressive flow is ~1/2 the speed of sound. I believe the answer of why we use the value of 1/20 of the speed of sound for port velocity calcs is really just a standard (times ten- or order of magnitude) engineering approximation.
a is the radius.
f=30
a=1.8mm to 1.8 meters.
While this can be lumped, it is not simply a mass and resistance, the resistance is distributed - a transmission line type of situation. Earl Geddes describes ports in his 1988 JAES article on bandpass enclosures as a mass with a distributed resistance. At even higher frequencies the pipe impedance obviously becomes an issue.....
Grab Bullock's old bandpass boxmodel and see the dramatic differences between regular ports (not filled with straws) and "ideal" ports, when used in BP enclosures.
------------------------------------------------
As far as the "central flow" being considered separately from the flare, it is only due to separation of flow at the entrance/exit. There is a subtle difference between the way I and collo think about this - I am considering pipe flow vs. entry and exit flow. I interpret collo as considering the flare as separate from the extension of the pipe which is attached to it. Port compression is not somehow due to higher turbulence in the central flow Turbulent air actually flows better (more mass transfer for a given average velocity) than laminar because the velocity profile is flatter. As I said before, the flow in ports is not steady or fully developed so it lies clearly in the turbulent regime, no matter what the reynolds number.
BTW, the air is not necessarily being compressed in the port and released in the flare. Air in subsonic flows is essentially incompressible. The typical velocity used in engineering calcs for truly compressive flow is ~1/2 the speed of sound. I believe the answer of why we use the value of 1/20 of the speed of sound for port velocity calcs is really just a standard (times ten- or order of magnitude) engineering approximation.
Hi there Ron,
I take your point about the air in a port as being always turbulent . With the air moving in and out all the time, I imagine it doesn't have a chance to settle into laminar flow.
I guess it's a matter of just how turbulent.
My dividing the airflow into "core" and "boundary layer" flows is a concept I got from one of the AES papers. I'm on the road for a few days, so I'm guessing it was the one about "Maximising Port Performance". It's a shame that you have to purchase those papers - it makes it difficult to share the ideas around.
In that paper, they measured the velocity of the air inside a port at different distances from the centreline.
I suspect they were using a constant flow of air for these tests.
The velocity in the central portion (what I call the core) was essentially uniform out to within a short distance of the port wall. As they approached the wall, the velocity dropped. This is the zone I refer to as the "boundary layer".
As more air was pushed through the port, the "boundary layer" grew in size.
I adopted the terminology as a convenient model to explain the results I had achieved with my tests. In reality, I have no way of knowing if my explanation is correct.
The main thing is that the results can be used to predict whether a given port / flare combination will exhibit audible turbulence above a given velocity. The nice thing about experimentally confirmed results is that they work even if the rationale is a bit wide of the mark.
I agree with you about "port compression"- The air isn't compressed at all.
It encounters increasing resistance causing the acoustic length of the port to change. The subsequent detuning causes a loss in output. It's a shame that whoever coined the phrase didn't call it "port detuning". - it would have conveyed a truer picture.
regards
Collo
I take your point about the air in a port as being always turbulent . With the air moving in and out all the time, I imagine it doesn't have a chance to settle into laminar flow.
I guess it's a matter of just how turbulent.
My dividing the airflow into "core" and "boundary layer" flows is a concept I got from one of the AES papers. I'm on the road for a few days, so I'm guessing it was the one about "Maximising Port Performance". It's a shame that you have to purchase those papers - it makes it difficult to share the ideas around.
In that paper, they measured the velocity of the air inside a port at different distances from the centreline.
I suspect they were using a constant flow of air for these tests.
The velocity in the central portion (what I call the core) was essentially uniform out to within a short distance of the port wall. As they approached the wall, the velocity dropped. This is the zone I refer to as the "boundary layer".
As more air was pushed through the port, the "boundary layer" grew in size.
I adopted the terminology as a convenient model to explain the results I had achieved with my tests. In reality, I have no way of knowing if my explanation is correct.
The main thing is that the results can be used to predict whether a given port / flare combination will exhibit audible turbulence above a given velocity. The nice thing about experimentally confirmed results is that they work even if the rationale is a bit wide of the mark.
I agree with you about "port compression"- The air isn't compressed at all.
It encounters increasing resistance causing the acoustic length of the port to change. The subsequent detuning causes a loss in output. It's a shame that whoever coined the phrase didn't call it "port detuning". - it would have conveyed a truer picture.
regards
Collo
Interesting reading..
I've always thought there were 2 problems here:
1. Port "chuffing", and
2. Port "whine" or "whistle"..
..and that problems with overall high air speed directly related to #2, NOT #1. (..and this is why Mach is spec'ed in most programs with a value of anywhere between .15 to .2 as a threashold.)
Port "chuffing" on the otherhand I thought was almost exclusivly related to that region imeadiatly below the port's tunning freq. where the driver unloads and effectively results in acoustic clipping. "Hit" a bass reflex with some reasonably high spl secondary (or worse fundamental) freq. and this occurs. Of course when the driver unloads here it starts pumping air like mad, (..depending on the nature of the tunning and the driver's excursion capability), and the port is not able to accomidate both its resonant "mass" and the added release and intake of air for the driver's "pumping". Here the addition of flares (and better - dimpled flares), simply reduce air flow resistance for that additional release and intake of air.
Am I wrong?
I've always thought there were 2 problems here:
1. Port "chuffing", and
2. Port "whine" or "whistle"..
..and that problems with overall high air speed directly related to #2, NOT #1. (..and this is why Mach is spec'ed in most programs with a value of anywhere between .15 to .2 as a threashold.)
Port "chuffing" on the otherhand I thought was almost exclusivly related to that region imeadiatly below the port's tunning freq. where the driver unloads and effectively results in acoustic clipping. "Hit" a bass reflex with some reasonably high spl secondary (or worse fundamental) freq. and this occurs. Of course when the driver unloads here it starts pumping air like mad, (..depending on the nature of the tunning and the driver's excursion capability), and the port is not able to accomidate both its resonant "mass" and the added release and intake of air for the driver's "pumping". Here the addition of flares (and better - dimpled flares), simply reduce air flow resistance for that additional release and intake of air.
Am I wrong?
Ron E said:
Shurely, one must be able to lump the resistance into one, if the size of the tube is considerably less than a wavelength? For higher frequencies, one will have to chop the tube into several shorter tubes, and also consider the acoustic capacitance of each segment. But maybe that is just what you said...?
Collo,
"I agree with you about "port compression"- The air isn't compressed at all.
It encounters increasing resistance causing the acoustic length of the port to change. The subsequent detuning causes a loss in output. It's a shame that whoever coined the phrase didn't call it "port detuning". - it would have conveyed a truer picture."
I don't think there's anything acoustic about port operation at Fb; even the longest ports are a tiny fraction of a wavelength.
It's purely a mass/spring/(damper, if you want to include losses) system, where the oscillating mass happens to be beating on the air around it and making sound.
If the effective length of the port did change with length, it would just move Fb; any output compression would be due soley to resistance, IMO.
Scott,
"Port "chuffing" on the otherhand I thought was almost exclusivly related to that region imeadiatly below the port's tunning freq. where the driver unloads and effectively results in acoustic clipping...Am I wrong?"
I think so.
Not sure what acoustic clipping is supposed to mean, but air velocity peaks at Fb and drops rapidly below, while driver excursion skyrockets.
Any distress below Fb is almost certainly driver distress.
"I agree with you about "port compression"- The air isn't compressed at all.
It encounters increasing resistance causing the acoustic length of the port to change. The subsequent detuning causes a loss in output. It's a shame that whoever coined the phrase didn't call it "port detuning". - it would have conveyed a truer picture."
I don't think there's anything acoustic about port operation at Fb; even the longest ports are a tiny fraction of a wavelength.
It's purely a mass/spring/(damper, if you want to include losses) system, where the oscillating mass happens to be beating on the air around it and making sound.
If the effective length of the port did change with length, it would just move Fb; any output compression would be due soley to resistance, IMO.
Scott,
"Port "chuffing" on the otherhand I thought was almost exclusivly related to that region imeadiatly below the port's tunning freq. where the driver unloads and effectively results in acoustic clipping...Am I wrong?"
I think so.
Not sure what acoustic clipping is supposed to mean, but air velocity peaks at Fb and drops rapidly below, while driver excursion skyrockets.
Any distress below Fb is almost certainly driver distress.
I got some equations for port diameter. Or simple, my correction of Xmax to Xe.
First, I calculate Xe (real excursion, depend od SPL, Sd etc.)
Looks simple, but was long time to get it. Mixed from many formulas.
Xe=k1*Sd*sqr(Re)/(BL) where k1=0.68/(0.68+Qts)+0.28
Fb=(0.383/Qts+0.142*Qts)*Fs
Dt=a*sqr(Fb*Sd*Xe) where a=(Qts/0.28)^0.8
Last is well known formula, but in original its Xmax, which is wrong, because Xmax is physical dimension, not real excursion. Xe is real excurion at some usual max. RMS power and around frequency of max. excursion (above Fb) which is 1.4-1.45*Fb. And correction factor a.
First, I calculate Xe (real excursion, depend od SPL, Sd etc.)
Looks simple, but was long time to get it. Mixed from many formulas.
Xe=k1*Sd*sqr(Re)/(BL) where k1=0.68/(0.68+Qts)+0.28
Fb=(0.383/Qts+0.142*Qts)*Fs
Dt=a*sqr(Fb*Sd*Xe) where a=(Qts/0.28)^0.8
Last is well known formula, but in original its Xmax, which is wrong, because Xmax is physical dimension, not real excursion. Xe is real excurion at some usual max. RMS power and around frequency of max. excursion (above Fb) which is 1.4-1.45*Fb. And correction factor a.
noah katz said:
"acoustic clipping" was just an abreviated attempt for my description..
The reason I asked was that I have not heard port chuffing except when the driver is unloaded or is still loaded but going through visible excursion because of very high spl's. (..obviously low freq. dependent.)
Considering a bass reflex system with moderatly high spl's *and* with a fairly steep highpass (prinicipally "positioned" to effect the passband imeadiatly below fs) - I haven't heard actual chuffing noise.
This suggests to me that driver "pumping" is significantly disturbing the resonance at port termination. Moreover, (though I'm by no means "certain"), I don't think it has much to do with port velocity - rather I think it significantly impairs the resonant behaviour of the port dynamically by oscilating the natural oscilation of the port tunning - effectivly "stunting" or "cliping" the intended tunning frequency. The mechanical result would presumably dramatically increase the turbulance at the port's termination and effectivly magnify the problem and audible behaviour.
Is that more clear?
noah katz said:
Didn't I already say that?
noah katz said:
Resistance changes acoustic length? No.
noah katz said:
Absolutely nothing acoustic about ports
Bullock has a simple and a complex port simulation in his software - the results are quite different.noah katz said:
WTF does that mean? Of course the length changes with length
"Considering a bass reflex system with moderatly high spl's *and* with a fairly steep highpass (prinicipally "positioned" to effect the passband imeadiatly below fs) - I haven't heard actual chuffing noise."
I think I see what you're trying to explain now - driver excursion increases below Fb but port velocity decreases, so where is the chuffing coming from.
Still, your explanation sounds very fanciful to me and not based on any physical phenomena that I'm familiar with.
I'd guess it has something to do with the much higher port air displacement changing the flow regime.
"Originally posted by noah katz
It encounters increasing resistance causing the acoustic length of the port to change. The subsequent detuning causes a loss in output. It's a shame that whoever coined the phrase didn't call it "port detuning". - it would have conveyed a truer picture."
Resistance changes acoustic length? No."
That was not originally posted by me, I was quoting someone else.
"Originally posted by noah katz
If the effective length of the port did change with length, it would just move Fb; any output compression would be due soley to resistance, IMO.
WTF does that mean? Of course the length changes with length "
It means I made a typo _ should have been "If the effective length of the port did change with resistance".
I think I see what you're trying to explain now - driver excursion increases below Fb but port velocity decreases, so where is the chuffing coming from.
Still, your explanation sounds very fanciful to me and not based on any physical phenomena that I'm familiar with.
I'd guess it has something to do with the much higher port air displacement changing the flow regime.
"Originally posted by noah katz
It encounters increasing resistance causing the acoustic length of the port to change. The subsequent detuning causes a loss in output. It's a shame that whoever coined the phrase didn't call it "port detuning". - it would have conveyed a truer picture."
Resistance changes acoustic length? No."
That was not originally posted by me, I was quoting someone else.
"Originally posted by noah katz
If the effective length of the port did change with length, it would just move Fb; any output compression would be due soley to resistance, IMO.
WTF does that mean? Of course the length changes with length "
It means I made a typo _ should have been "If the effective length of the port did change with resistance".
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.