Port Length Recommendations

[collo]

avaserfi wrote:

.... I have been unable to find any information on this subject.....I will look into building two cabinets with equal volume and port length - the only difference being one will have folds and the other will not. This model will be measured to see if there is any effect on port folds and port resonance.

I built a concentric port, which turned out to be quite a dismal failure.

As part of the testing, I evaluated the effectivenes of inserting damping pads inside the turns as a means of reducing the amplitude of the port resonance. They didn't help...

[SteveCallas]

Mike wrote:

The usual response has been "use passive Radiators for this sub". I'm trying to explore the different possibilities to achieve a decent sub using ports for those who's budget doesn't include the added expense of PR's. I'm hoping the use of longer ports and the resulting lower first port resonance can be dealt with.

Mike, take a look at increasing enclosure volume to 120 liters, using two 4" diameter ports that are each 35" long, raising the highpass to 20hz, and increaisng power to 1000 watts Should be a pretty potent performer.

avaserfi wrote:

Firstly, it seems as if your port measurement is far to flat. Rather it should have a far more narrow passband.

This is because it is nearly impossible to completely block out the output being created from the drivers, even when close micing the port.

Secondly, at no point do I see a 20dB spike in output relative to the rest of the graph, but rather it seems there is approximately an 8dB peak in relation to port response. In comparison to the main signal the resonance has a peak that is no greater than the actual signal of the subwoofers and thus is at reference without any crossover. This means that the port resonance is actually playing at 0dB with regard to the main signal [no higher or lower]. If the resonance was at 120Hz with an 80Hz crossover and attenuated by a 4th order crossover it would thus be -15db from the actual signal.

I'm not sure you are interpretting the graph correctly. There is no combined response of the port and drivers, just each individually. My comments were concerning the output being created from the port only. Even with the output from the drivers artificially increasing the close mic port measurement at ~100hz, there is still a 20db spike from that average output level (65db) to the first resonance (85db). In an ideal measurement, the driver's output wouldn't influence the close mic port measurement and that spike would be even more extreme than 20db. Ricci's simulation shows a spike of 60db. Again, this is not in the overall output created by the subwoofer, but rather just the output created from the port.

That said, if this first resonance peak does reach the same amplitude as the output being created from the drivers - "0db" as you put it, displacement at that frequency is doubled and it would correspond to a 6db peak in your frequency response.

collo wrote:

I do have a 2kw IB now and ready access to 10inch flared ports, so there is the possibility of more work, although my house is over 100yrs old and would not survive 130dB @ 10hz.

That's such a "glass half empty" point of view

[collo]

umm that would be a "House half empty" view...

[avaserfi]

SteveCallas wrote:

I'm not sure you are interpretting the graph correctly. There is no combined response of the port and drivers, just each individually. My comments were concerning the output being created from the port only. Even with the output from the drivers artificially increasing the close mic port measurement at ~100hz, there is still a 20db spike from that average output level (65db) to the first resonance (85db). In an ideal measurement, the driver's output wouldn't influence the close mic port measurement and that spike would be even more extreme than 20db. Ricci's simulation shows a spike of 60db. Again, this is not in the overall output created by the subwoofer, but rather just the output created from the port.

That said, if this first resonance peak does reach the same amplitude as the output being created from the drivers - "0db" as you put it, displacement at that frequency is doubled and it would correspond to a 6db peak in your frequency response.

Where is there a 20dB spike from average? For the spike to be 20dB the average of the port SPL graph would have to be 65dB that is clearly not the case as 65dB is closer to being the minimum value of the port SPL graph. Rather the average of the plot is closer to 75dB (when taking the average you include the resonance peak). This situation is the same in Ricci's simulation - in no way is the average of his graph 60dB which is the only way a 60dB spike could be had if the peak is 120dB.

As far as a combined response, I realize that one was not provided, but rather I noted that the peak representing port resonance reached the average output (perhaps slightly below) of the driver output graphs. This is where the relative port resonance level of 0dB was derived and relates directly to the definitions in the papers I have mentioned.

Either way, this is a moot point as the argument is purely based on your feelings of port resonance and its audibility. I suggest you read the previously linked articles which deal in depth with audibility of resonances such that you have a foundation with regard to what resonances and its audibility.

[Mike P.]

Mike, take a look at increasing enclosure volume to 120 liters, using two 4" diameter ports that are each 35" long, raising the highpass to 20hz, and increaisng power to 1000 watts Should be a pretty potent performer.

There's no doubt it would be. What I'm trying to achieve is the most output in the 18 hz HT range in the smallest box possible. If the debate over the 120 hz first port resonance can be resolved, then the use of longer ports will be a viable option for designs where PR's are not in the budget and size constraints are an issue.

[SteveCallas]

avaserfi wrote:

Where is there a 20dB spike from average? For the spike to be 20dB the average of the port SPL graph would have to be 65dB that is clearly not the case as 65dB is closer to being the minimum value of the port SPL graph. Rather the average of the plot is closer to 75dB (when taking the average you include the resonance peak). This situation is the same in Ricci's simulation - in no way is the average of his graph 60dB which is the only way a 60dB spike could be had if the peak is 120dB.

The chart shows spl vs frequency - yes, there is a 20db peak, as the amplitude ~100hz is down to ~65db, then shoots up to ~85db at the first resonance. The average of the port output over the entire frequency range of the measurement is not significant or related to anything we are talking about. I don't understand why you are using that as a frame of reference in regards to a first resonance peak. And as I stated, take away driver background noise form the port measurement, and you'll see that the peak is even more extreme.

As far as a combined response, I realize that one was not provided, but rather I noted that the peak representing port resonance reached the average output (perhaps slightly below) of the driver output graphs. This is where the relative port resonance level of 0dB was derived and relates directly to the definitions in the papers I have mentioned.

And in a combined measurement, the first resonance peak would result in a peak in the overall FR as well, just not as sharp.

[avaserfi]

Mike P. wrote:

There's no doubt it would be. What I'm trying to achieve is the most output in the 18 hz HT range in the smallest box possible. If the debate over the 120 hz first port resonance can be resolved, then the use of longer ports will be a viable option for designs where PR's are not in the budget and size constraints are an issue.

As of now it seems that nearly all debate with regard to the audibility of port resonance in a design where resonance is at approximately 120Hz with an 80Hz crossover has been complete conjecture based on no credible research. I don't see how see how any resolution will be had if this continues. I have repeatedly cited credible sources that have spent vast resources on defining thresholds for resonances and they are continually ignored. Sadly, the results of these studies cannot be accurately summarized due to the large amount of variables involved. It seems as if this topic will remain in a standstill due to this situation.

At the same time it seems this discussion is departing from audibility of said port resonances to other semi related subjects further reducing the likelihood that a consensus will be formed.

[SteveCallas]

Mike wrote:

What I'm trying to achieve is the most output in the 18 hz HT range in the smallest box possible. If the debate over the 120 hz first port resonance can be resolved, then the use of longer ports will be a viable option for designs where PR's are not in the budget and size constraints are an issue.

You're barely losing any extension by using a 20hz tune in the slightly larger enclosure with the higher highpass. You're gaining more headroom, more port cross sectional area, and raising the first resonance. If you really want to find about the impact of a port resonance, go ahead and use a 45" long port with an 80hz crossover. You can always shorten the port afterwards.

[avaserfi]

collo wrote:

I built a concentric port, which turned out to be quite a dismal failure.

As part of the testing, I evaluated the effectivenes of inserting damping pads inside the turns as a means of reducing the amplitude of the port resonance. They didn't help...

This is unsurprising after my research last night - as you quoted. Hopefully I will have time to test the straight versus folded port design.

SteveCallas wrote:

The chart shows spl vs frequency - yes, there is a 20db peak, as the amplitude ~100hz is down to ~65db, then shoots up to ~85db at the first resonance. The average of the port output over the entire frequency range of the measurement is not significant or related to anything we are talking about. I don't understand why you are using that as a frame of reference in regards to a first resonance peak. And as I stated, take away driver background noise form the port measurement, and you'll see that the peak is even more extreme.

Firstly, it is important to note you have no crossover implemented in your measurements which is a completely unrealistic circumstance and not at all what I have recommended be used. For example, if a typical 4th order Linkwitz-Riley crossover was implemented at 80Hz on your model the port resonance would be attenuated by about 30dB - far from audible.

As far as the references to port output and subwoofer output they are incredibly significant to the audibility of resonances. Resonances thresholds have been developed relative to SPL of the original [non-resonant] source. This is why claiming the maxima of 20dB above average as the the actual amount of created resonance is grossly misleading.

Also, please remember this discussion is not about whether or not these resonances exist, but rather if they are or are not audible.

And in a combined measurement, the first resonance peak would result in a peak in the overall FR as well, just not as sharp.

Again, this is only true with no crossover - a completely unrealistic [and unrecommended] expectation.

I do not see the point in arguing this subject further at no point in time did I ever recommend use of a long port without a crossover yet this is your chosen exemplar. I concede that if no crossover is used there is the possibility* of audible port resonance in a subwoofer system; just as there is the possibility of audible panel resonance as well.

*This is dependent on the natural roll off of the subwoofer system as well as the frequencies involved as audibility of resonance is frequency dependent among other things.

[SteveCallas]

avaerfi wrote:

Firstly, it is important to note you have no crossover implemented in your measurements which is a completely unrealistic circumstance and not at all what I have recommended be used.

Come now, of course I recommend the use of a crossover (no higher than 80hz if using capable speakers), and that will decrease the amplitude of the summed response of the drivers and port above some chosen frequency, however, the spike still remains intact, albeit at a reduced output level. I stated that in my first response in this thread:

Now sure, putting the crossover into play helps bring everything down a bit, but the spike - in relation to the rest of the subwoofer output above 80hz - remains intact, just reduced in magnitude

So you're never going to completely eliminate it, the goal is to minimize it enough so that it stands no chance of being audible or annoying. What's audible? I believe the human ear can pretty easily discern a 3db bump even in the range we're talking about. What's annoying? It will vary from person to person. Rather than test at what point my first resonance will become annoying, I say play it safe. For the sub that I am showing measurements of, the first resonance is not easily discernable with an 80hz crossover. However, with what has been suggested in this thread - a first resonance as low as 120hz being ok with a 4th order 80hz crossover - I doubt that would be the case.

Can I bring up several sources to try and back my position? No, and I'd care not to search. If someone wants to try for themself though, it's a pretty simple test. If you hear an odd sound like blowing over a bottle with bass notes in music, you would be wise to shorten the port - 36" or less and you'll be safe.

[Mike P.]

As of now it seems that nearly all debate with regard to the audibility of port resonance in a design where resonance is at approximately 120Hz with an 80Hz crossover has been complete conjecture based on no credible research. I don't see how see how any resolution will be had if this continues. I have repeatedly cited credible sources that have spent vast resources on defining thresholds for resonances and they are continually ignored. Sadly, the results of these studies cannot be accurately summarized due to the large amount of variables involved. It seems as if this topic will remain in a standstill due to this situation.

No, I will build a sub with the design goals mentioned, measure it wih REW and post the results. Whatever the posted results are, we move forward from there.

[avaserfi]

SteveCallas wrote:

So you're never going to completely eliminate it, the goal is to minimize it enough so that it stands no chance of being audible or annoying. What's audible? I believe the human ear can pretty easily discern a 3db bump even in the range we're talking about. What's annoying? It will vary from person to person. Rather than test at what point my first resonance will become annoying, I say play it safe. For the sub that I am showing measurements of, the first resonance is not easily discernable with an 80hz crossover. However, with what has been suggested in this thread - a first resonance as low as 120hz being ok with a 4th order 80hz crossover - I doubt that would be the case.

As I suspected, complete speculation. My recommendations and comments are completely based on a multitude of credible perceptual research. These research articles very clearly state the threshold levels for which resonance becomes audible in various conditions rather than relying on a beliefs such as: 3dB bumps are audible in at certain frequencies. It seems use of such credible sources that have undergone extreme levels of peer review are far more reliable.

Can I bring up several sources to try and back my position? No, and I'd care not to search. If someone wants to try for themself though, it's a pretty simple test. If you hear an odd sound like blowing over a bottle with bass notes in music, you would be wise to shorten the port - 36" or less and you'll be safe.

I had read the papers in question prior to this threads creation. As I have also mentioned previously in this thread I have done my own testing with a 55" port using a handful of subjects. In all cases there was no audible port resonance with the specified crossover. Even so, my results are meaningless due to the far more credible studies that have been done.

It also seems your recommendation of a 36" or less port recommendation is based on no credible research, but rather a guess at what is safe. I see no reason to heed such a baseless warning.

You have made it very clear at an unwillingness to look at the actual credible data with regard to the phenomenon of port resonance thusly, I see no point in continuing in repeating myself. If there are any new issues raised I will be more than willing to reply, but I have no interest in continuing a debate of credible research versus one mans personal opinion.

Mike P. wrote:

No, I will build a sub with the design goals mentioned, measure it wih REW and post the results. Whatever the posted results are, we move forward from there.

While this is a noble goal how will you know if the resonance levels measured are within audible thresholds or not [without having read the relevant material]? One person listening to a subwoofer is not a statistically valid method of attempting this study.

Either way, if I have time tonight I will try to do this test with the subwoofer I built.

[SteveCallas]

You can link to thousands of pages of research if you like - if the resonance is audible, but the research tells me I shouldn't be hearing it, do I just pretend I'm not hearing it? That's silly. Lighten up.

Mike's proposed experiment is absolutely valid. Use a 55" long port (~122hz first port resonance) with a 4th order crossover at 80hz, take some measurements, and do some critical listening. If he notices an odd bump in the measurements and hears a repetitive one note tone during bass notes, it will be obvious that the first port resonance is within the audible (and annoyable) range.

My 36" length recommendation is based on actual user experiences. People have built ported subs under my direction with port lengths in the low 40" range and have reported hearing a repetitive resonance with bass in music. It went away when they shortened the port. In designs which I have suggested 36" port lengths, no such resonances were heard. Is 36.00" the precise length not to be exceeded? Probably not, but I'd say it's quite close to the threshold, and it makes some logical sense in that it makes the first resonance more than an octave higher than the crossover point, meaning output from the sub will be at least 24db down.

[avaserfi]

I just finished taking the measurements using my Kappa VQ build (port length approximately 55" with cross section of 37.5 square inches).

Maroon is the subwoofer measured from 1" with no crossover.
Blue is the subwoofer measured from 1" with a 4th order Linkwitz-Riley crossover at 80Hz.
Purple is the port measured from 1" from the opening with no crossover.
Green is the port measured from 1" from the opening with a 4th order Linkwitz-Riley crossover at 80Hz.



Clearly the baseline of reference for these measurements is about 110dB with the crossover activated the port resonance [Q value approximately 37*] located at 111Hz is attenuated approximately 20-22dB. With regard to audibility of this resonance the handful of research papers on this topic have come to remarkably similar conclusions. At a Q of this value and frequency in this span the relative [to baseline frequency] level of audibility is approximately 2-3dB above that of the baseline using the Olive et al paper** or using the Fryer paper (summarized thoroughly in Toole's work) and a linear transform of the resulting data it seems with a music source resonance audibility would need to be 2-3dB above baseline while for it to be audible with pink noise it would likely be a few dB below baseline . Please look to my referenced articles for expanded information on this subject.

Clearly the attenuation achieved via the recommended crossover achieves a value far lower than that of defined [by credible research] resonance thresholds.

*Formula used: Q = F0 / (delta F) where F0 is center of resonant frequency and delta F is f2 - f1, where f2 is the upper and f1 the lower cutoff frequency (-3dB point in relation to the resonant peak).

**It should also be noted that these levels are in reference to pink/white noise it has been shown that resonance is typically less audible with actual music signals.

Do note measurements were taken in room with a calibrated ECM8000 microphone situated on a boom mic stand using a Xenyx 802 as a power supply all via a Soundblaster external USB sound card. Pictures of these measurements follow:

[Mike P.]

avaerfi wrote:
While this is a noble goal how will you know if the resonance levels measured are within audible thresholds or not [without having read the relevant material]? One person listening to a subwoofer is not a statistically valid method of attempting this study.

Agreed. But it either works or it doesn't for a HT application. If I can't hear any "problems" while listening to a movie at different output levels, then I won't have a problem recommending the longer port lengths to someone who's goals it suits.

SteveCallas wrote:
My 36" length recommendation is based on actual user experiences. People have built ported subs under my direction with port lengths in the low 40" range and have reported hearing a repetitive resonance with bass in music. It went away when they shortened the port. In designs which I have suggested 36" port lengths, no such resonances were heard

.

I can see longer ports having an effect on music. But my goals are the use of longer ports in HT subs only. Some experimentation should give us more information on the matter. I'm looking forward to getting started on this.

[SteveCallas]

Mike, if it's for movies only, it will be much harder to notice, as the resonance won't stand out as much. I didn't catch that earlier.

[avaserfi]

SteveCallas wrote:

You can link to thousands of pages of research if you like - if the resonance is audible, but the research tells me I shouldn't be hearing it, do I just pretend I'm not hearing it? That's silly. Lighten up.

Due to the large amount of extraneous variables involved with such a method there is a large probability that other issues such as room resonance are to blame for this ringing. It is impossible to separate bias and these extraneous variables from the possibility of audible port resonance. This is where the reliance on credible research papers that use a variety of listeners in varying situations such that they are generalizable to real world situations.

Mike's proposed experiment is absolutely valid. Use a 55" long port (~122hz first port resonance) with a 4th order crossover at 80hz, take some measurements, and do some critical listening. If he notices an odd bump in the measurements and hears a repetitive one note tone during bass notes, it will be obvious that the first port resonance is within the audible (and annoyable) range.

My 36" length recommendation is based on actual user experiences. People have built ported subs under my direction with port lengths in the low 40" range and have reported hearing a repetitive resonance with bass in music. It went away when they shortened the port. In designs which I have suggested 36" port lengths, no such resonances were heard. Is 36.00" the precise length not to be exceeded? Probably not, but I'd say it's quite close to the threshold, and it makes some logical sense in that it makes the first resonance more than an octave higher than the crossover point, meaning output from the sub will be at least 24db down.

Please see post #39

Mike P. wrote:

I can see longer ports having an effect on music. But my goals are the use of longer ports in HT subs only. Some experimentation should give us more information on the matter. I'm looking forward to getting started on this.

I only have one interest music if there is any coloration at all to the signal other than something I intentionally introduce via equalization I will be unsatisfied...

[Mike P.]

Thanks for doing the measurements and posting the results. This is a step forward for my original intension's. I have no doubt my own experiments with HT applications will verify your findings. The goal of small box, long port is looking good.

[avaserfi]

Mike P. wrote:

Thanks for doing the measurements and posting the results. This is a step forward for my original intension's. I have no doubt my own experiments with HT applications will verify your findings. The goal of small box, long port is looking good.

No problem at all. I see you have modeled the IXL-12 in a small box with a round port [posted in this thread] have you begun to model the unit with a slotted port yet? Before you begin construction I believe it would be beneficial if I overlook your plans. I am about to head to work, but after I am done I will model the driver and make my recommendations for ideal cabinet/port size as well as a small box solution.

[Mike P.]

The IXL-12 in a small box with a round port was just preliminary modeling to get things going here. I will have to draw up plans with a slot port as that would save more space over a round port. I'll try to model with a maximum airspeed of around 28 m/s at the tuning frequency to avoid most of the effects of port compression.

[SteveCallas]

avaserfi wrote:

With regard to audibility of this resonance the handful of research papers on this topic have come to remarkably similar conclusions. At a Q of this value and frequency in this span the relative [to baseline frequency] level of audibility is approximately 2-3dB above that of the baseline using the Olive et al paper** or using the Fryer paper (summarized thoroughly in Toole's work) and a linear transform of the resulting data it seems with a music source resonance audibility would need to be 2-3dB above baseline while for it to be audible with pink noise it would likely be a few dB below baseline

When the response of the port and driver are summed, there will be a peak ~6db centered at the first port resonance. When that response is then summed with the speakers', you are correct in that the effect will be minimal, I am calculating a peak less than 0.5db. I guess what remains to be noted from Mike is whether or not the specific tone of the resonance can be percieved - I know your sub will be for HT only Mike, but for the sake of this experiment, can you also use music?

[Mike P.]

Not a problem Steve. I'll gladly test both HT and music.

[avaserfi]

Mike P. wrote:

The IXL-12 in a small box with a round port was just preliminary modeling to get things going here. I will have to draw up plans with a slot port as that would save more space over a round port. I'll try to model with a maximum airspeed of around 28 m/s at the tuning frequency to avoid most of the effects of port compression.

I would recommend having maximum airspeed in the passband below 25 m/s if you are trying to sufficiently minimize port compression and the possibility of chuffing. I will try to model such a situation later tonight.

SteveCallas wrote:

When the response of the port and driver are summed, there will be a peak ~6db centered at the first port resonance. When that response is then summed with the speakers', you are correct in that the effect will be minimal, I am calculating a peak less than 0.5db. I guess what remains to be noted from Mike is whether or not the specific tone of the resonance can be percieved - I know your sub will be for HT only Mike, but for the sake of this experiment, can you also use music?

It seems as if your calculations are not in reference to the baseline SPL as there will be absolutely no peak above baseline with the subwoofer crossover in place and mains activated. It seems as I must repeat myself, but this fact is extremely important as resonance thresholds have been established with reference to relative level compared to the actual [uncolored] signal. Simply put having a resonant peak of 0.5dB containing a high quality factor [as is the case] within a cross over region that is even -10dB* below the baseline at the frequencies in question is inaudible in situations involving music reproduction as well as pink/white noise.

As far as perception I have used myself, others and far more credible sources such as peer reviewed JAES articles related to actual real world measurements to prove my point. I fail to see why you are unwilling to accept that this resonance is not audible if properly crossed over as I recommend.

*It is clear that a -10dB attenuation is a much higher value than that actually obtained from my measurements. Furthermore, it is likely that if resonance was only attenuated by -5dB it would be inaudible in the same circumstances as shown by credible perceptual research.

[avaserfi]

Mike P. wrote:

There are subs that do model well in such an enclosure. The new IXL-12 is what got me started on this path. Works in a 3.5 cu.ft. box tuned to 18hz. Handles 800 watts, SPL is 108db at 18 hz, the only drawback being the port air speed. Here is the saved file for the IXL-12. Simply download it, click on the file and it will open In WinISD. All the modeling is done.

I have briefly modeled the IXL-12 with a slot port. In a 2.75 cubic foot enclosure tuned to 18Hz with a slot port 9.5" x 2" x 53.5" with first port resonance at about 125Hz. The unit models at 106dB at 18Hz with 800 watts*. Furthermore, port speed never exceeds 23 m/s with a second order highpass Butterworth centered at 18Hz. This highpass filter also eliminates the chance of over excursion near tuning.

*It is important to note that modeling is only accurate with the assumption that the driver and system as a whole will act in a linear fashion. This is a highly unlikely case with most drivers once moderate to high power/excursion is achieved due to lack of linearity with the drivers motor [suspension and motor force].

With regard to this linearity relating to air velocity modeled in your attached file I would expect port compression near tuning which would result in less overall output at 18Hz as compared to my offered model despite modeled results showing the opposite.

[hot grits]

Whats up fellas, I am also using a 55 inch slot port on my current build. http://www.hometheatershack.com/foru...v-2-build.html
Unfortunately, this project has to take a backseat for about a month. I am hoping my crossover takes care of any issues. If not, can someone post a link to that notch filter that was mentioned in one of the first posts. I did a google search, but i dont know exactly what I am looking for. Also, why wouldnt you be able to tame down the peak at the first port resonance with an EQ?