The Perfect Port?

[augerpro]

So I'm reading the Harman paper on ports, here is a link: https://www.dropbox.com/s/u18zq95ko9...Study.pdf?dl=0. They test various theories regarding flare radius, power compression, harmonic distortion, surface roughness, etc. They come up with a Normalized Flare Radius relating the flare radius to the port length: NFR = port length/ 2(flare radius), 0 < NFR < 1 where 0 is a straight pipe and 1 is a full flare. After a number of tests of the previously mentioned properties, they find a NFR = .5 seems to be the best balance of properties. This greatly simplifies the equation to basically saying flare radius = port length. Essentially there is never a straight a section, the profile is always an arc with a radius equal to the port length and some minimum diameter in the middle that the designer would select based on max air velocity (or Reynolds number if you can figure it out). You can visualize as the port length grows so does the port exit size. For any given length and minimum diameter I came up with this equation which defines the diameter of the port exit: Ed = .268L + Md, where Ed is port exit diameter, L is port length, and Md is minimum diameter. For example if my port were 8" long and I needed a 3" minimum diameter to keep air speed low, then the port exit diameter would be 5.144". This is on both ends BTW. Two other provisos in the article: 12mm radius to blend the flare to baffle, and a 15mm flange on the inside flare to sort of mimic a little flat baffle. Apparently this reduced odd order harmonic distortion in their tests.

Threw this together in Fusion 360 to give you an idea:

[ergo]

Interesting.

For getting closer to 'perfect' see also KEF-s take here

https://www.kefdirect.com/media/wysiwyg/documents/ls50/ls50_white_paper.pdf

They made the middle of the port 'flexible' claiming a further improvement.

[wkhanna]

having a constant, relatively large radius along the entire axis of the port seems intuitively to make sense to me.

the idea to use a flexible (energy absorbing) wall to mitigate naturally occurring resonance is quite interesting, too.

neat stuff, guys!

[augerpro]

Thanks Ergo for the paper!

This Harman paper has me a bit confused. Instead of doing what most of us would do and take a straight pipe and add ever larger radius flares to the end, they make the entire pipe a flare. Going from NFR = 1 which is a half circle, all the way down to NFR = 0 which basically means the radius has reached infinity, in other words a straight pipe. Other than the actual straight pipe they use, the only thing that they test that might get close to a straight pipe w/ flared ends is the elliptical profile, which fares pretty well (go figure). I'm not sure if the provide the equation for that, so not sure how close it would be.

My other concern is in their attempt to compare apples to apples, they keep port length the same, but change the diameter to keep the Fs the same also. Given how important diameter is, I'm not sure the differences can be attributed to the flare. However the more highly flared ports are the ones with the smallest diameter, and they are the better performers vs a straight pipe (which had the largest diameter) so I guess the flares did their job despite being handicapped. Now they also ultimately "choke" a bit harder than the straight pipe, I wonder if keeping the diameter the same (and Fs the same so letting length float) would have resulted in a similar choke point to the straight pipe? BTW even the straight pipe had a 12mm radius at the baffle.

[TEK]

Are you going to take this from the drawing board and test how such a design work?
It would be very interesting to see if this type of port makes any measurable differences.

Producing such a port will be a interesting challenge for your reqular DIY'er ;-)
Any ideas avout how it could be done? I do see that something like the trns-lam design nethod mith work

[augerpro]

It would be very interesting to see if this type of port makes any measurable differences.

Well there are measured results in the paper I don't know yet if I'll test it out. I have a design with a builder right now (all my tools are in storage) and we discussed it but haven't made a decision. If we do it he has access to a 3D printer that we would print it on. You could also CNC each half and glue it together. Those aren't easy for a typical DIY'er, but it is becoming more common.

[augerpro]

Probably the main concern for the DIY'er is surface smoothness. You can see in the paper they test that and I was surprised that even a 1mm surface texture was measurably worse than smooth.

[Steve Manning]

Probably the main concern for the DIY'er is surface smoothness. You can see in the paper they test that and I was surprised that even a 1mm surface texture was measurably worse than smooth.

I wonder what B&W has figured out since they dimple the inside of their ports?

[TEK]

According to commercial/product leflet from like 15-20 years ago (when I bought my 805 that I had a lot of years) the dimpels are there to reduce air turbolence - at least that whats I recall from those days....
3D print - of course, I did not think of that. Jup, more and more available!

[augerpro]

I wonder what B&W has figured out since they dimple the inside of their ports?

Dunno. Like many things maybe nothing but a hypothesis. In the Harman paper they were a bit surprised too, but figured that things like golf balls have dimples that work well in purely turbulent air flow. But you never get to purely turbulent air flow in a port. Looking at their chart you'll see the port has completely choked before then, in the transition zone which is a mix of laminar and turbulent flow. I think on the Reynolds chart turbulent is over 100k, while ports are having real problems by around 50k.

[TEK]

"Bowers & Wilkins’s Flowport™ minimises turbulence in the same way as a golf ball. Dimples on the surface generate tiny eddy currents over which air can flow smoothly and, above all, silently."

http://www.bowers-wilkins.net/Discover/Discover/Technologies/Flowport.html

Wondering about other parts of B&W tech? Go here:

http://www.bowers-wilkins.net/Discover/Discover/Technologies

[augerpro]

Thanks TEK! So it is interesting B&W only use the dimples at the flare, which is exactly where we get the formation of eddies, so pretty turbulent. But is turbulent enough to matter, given the results in the Harman paper? Still they didn't put it inside so maybe they got some benefit without the penalty shown in the Harman paper which applied the beads to the entire port?

OTOH there are no measurements provided, and his claim of 6dB seems hard to believe when they already have a very large flare, something similar in the Harman paper would be good enough that the flare was no longer the SPL limit, but the port tube (mostly diameter) itself.

[wkhanna]

these ports are perfect for our printer at work.
i can get a V good surface finish on a radius like they have.

[Steve Manning]

Thanks TEK! So it is interesting B&W only use the dimples at the flare, which is exactly where we get the formation of eddies, so pretty turbulent. But is turbulent enough to matter, given the results in the Harman paper? Still they didn't put it inside so maybe they got some benefit without the penalty shown in the Harman paper which applied the beads to the entire port?

OTOH there are no measurements provided, and his claim of 6dB seems hard to believe when they already have a very large flare, something similar in the Harman paper would be good enough that the flare was no longer the SPL limit, but the port tube (mostly diameter) itself.

As I recall, either from a video or paper, B&W looked at golf balls to come up with the dimple concept. Found this on their website. http://www.bowers-wilkins.com/Discov.../Flowport.html

[---k---]

On the subject of ports. Does anyone remember Polk having what they called a Power Port. It was an inverted cone inside the port exit. I remember seeing this about 10 - 15 years ago. I'm not sure the thinking behind it.

[augerpro]

I remember that. Don't recall exactly how it looked but in that Harman paper there is something very close cited. I don't think it was a very new idea. Had to do with directing and smoothing air flow instead of abrupt transition from port environment to free space. Something like that anyway.

[NyxOne]

Power port with drawing from an older thread (post #6)

Chuck

[JonMarsh]

Probably the main concern for the DIY'er is surface smoothness. You can see in the paper they test that and I was surprised that even a 1mm surface texture was measurably worse than smooth.

That's not something i would have expected... Interesting.

[augerpro]

That Polk power port is actually the one cited in the Harman paper. Thought it looked familiar.

[JonP]

Thanks Ergo for the paper!

This Harman paper has me a bit confused. Instead of doing what most of us would do and take a straight pipe and add ever larger radius flares to the end, they make the entire pipe a flare. Going from NFR = 1 which is a half circle, all the way down to NFR = 0 which basically means the radius has reached infinity, in other words a straight pipe. Other than the actual straight pipe they use, the only thing that they test that might get close to a straight pipe w/ flared ends is the elliptical profile, which fares pretty well (go figure). I'm not sure if the provide the equation for that, so not sure how close it would be.

My other concern is in their attempt to compare apples to apples, they keep port length the same, but change the diameter to keep the Fs the same also. Given how important diameter is, I'm not sure the differences can be attributed to the flare. However the more highly flared ports are the ones with the smallest diameter, and they are the better performers vs a straight pipe (which had the largest diameter) so I guess the flares did their job despite being handicapped. Now they also ultimately "choke" a bit harder than the straight pipe, I wonder if keeping the diameter the same (and Fs the same so letting length float) would have resulted in a similar choke point to the straight pipe? BTW even the straight pipe had a 12mm radius at the baffle.

Very interesting paper, with interesting implications. Like, the large differences in THD, with even moderate changes, for instance.

Here's some work done a while ago, not quite the Harmon tour de force (Force?) as the paper, but some good work done on straight pipes with different end flare radi and the effects on limits... in the general direction of your questions. He even boils down his findings into a calculator..

https://www.subwoofer-builder.com/flare-testing.htm

Wow... been 10-12yrs already... 8O

[mdjamesd]

Not to awaken a dead thread, but I stumbled across this one. When I was building my sub last summer, I finally hunted down the power port spreadsheet. It works fantastically for enabling the use of a shorter port to reach desired tuning. If anyone wants it, drop me a pm.

[augerpro]

Can you upload it here, or link to it? Was this a third-party-built software from Polk's patent, or?

[Zvu]

One more thing, suited for this thread.

Is there a way to predict how severe port resonances will be ?

There are a number of simulators that tell us where the port resonance will occur but, as can be seen from measurements from stereophile, there are a lots of ports that influence, otherwise perfect frequency response, in a rather bad fashion. Lots of loudspeakers have been made and engineered by a knowledgeable bunch of guys yet, severe port resonances are present and are easily heard. Clearly, back firing ports that have resonances higher in frequency will impact the sound less but some of this loudspeakers have front-firing ports and with very low frequency resonances (carrying a lot of energy). Here are some measurements from stereophile:

Morel Octave 6



Joseph Audio Pulsar



Sjofn HiFi



Canalis Anima



Opera Callas



Dynaudio Excite X14



Usher BE718



Aerial Acoustics 5T



Snell Illusion A7




There are a number of speakers that perform well in this regard and the price isn't part of the equation. Examples are given:

Pioneer SP-BS22-LR



Elac Debut B6



Harbeth Super HL5



Technics Premium Class SB-C700



Revel Performa3 M106



DALI Zensor 1




I'm posting this because i'm truly and utterly ashamed of how bad my port measurements are on my last 2-way bookshelf. I just couldn't believe they were true until a friend measured it with his measuring equipment and it measured the same. Since i did front firing slot-port, it's now integral part of the cabinet and bracing, and i don't want to invest a dime in a new cabinet until i fully understand what is happening here and how to mitigate or prevent it. Port resonances on my 2-way are at about 300Hz and 600Hz.

[Alaric]

Probably the main concern for the DIY'er is surface smoothness. You can see in the paper they test that and I was surprised that even a 1mm surface texture was measurably worse than smooth.

That's not something i would have expected... Interesting.

After playing with, and actually improving a few, intake ports on engines, that doesn't surprise me at all. The turbulence creates a bearing surface for the moving air, reducing friction.

[augerpro]

Zvu> my measurements are pretty poor there too. I wonder what made the better examples better? Although I would note all are tuned fairly high except the Harbeth. Perhaps their higher resonances were just easier to absorb with stuffing/lining? Is their a difference between the resonances generated by a straight port versus a fully curved port as is the example in this thread? Definitely something worth investigating.

[mdjamesd]

Can you upload it here, or link to it? Was this a third-party-built software from Polk's patent, or?

There was a spreadsheet made available many moons ago when Matthew Polk was interviewed regarding the power port. Polk removed it from their webpages, but the internet never forgets.

I am out of town at the moment, but when I get back, I will fire up my old laptop and upload it here

[fatmarley]

IIRC there was a commercial speaker that had some kind of material lining the inside of the port. Not sure what that was supposed to do though.

[Zvu]

Zvu> my measurements are pretty poor there too. I wonder what made the better examples better? Although I would note all are tuned fairly high except the Harbeth. Perhaps their higher resonances were just easier to absorb with stuffing/lining? Is their a difference between the resonances generated by a straight port versus a fully curved port as is the example in this thread? Definitely something worth investigating.

There are few in the "bad" department that have high tuning that still suffer from strong port resonances so no correlation there that i've noticed. Curved vs straight would be interesting comparison.

IIRC there was a commercial speaker that had some kind of material lining the inside of the port. Not sure what that was supposed to do though.

Joseph Audio Perspective has it optional. But i think that's more in the service of loudspeaker tuning rather than anti-resonance role, but i might be wrong.

Anyway, some kind of software that can simulate performance of different kinds of ports would be great.

[Alaric]

Calculating Port Velocity

If you know the cross-sectional area of a given port you can calculate the port velocity based on the bore diameter and the piston speed at any given RPM using the following formula.

Port Velocityfps = (Ps ÷ 60) x (B2 ÷ Ap)

Where:

Ps = piston speed in feet per minute

B = bore diameter in inches

Ap = area of port in square inches

The first part of the formula converts the piston speed to feet per second while the second half relates the bore area to the port cross section. Consider the following example: A 3-inch-stroke 302-ci engine running at 4,400 rpm (torque peak) achieves a piston speed of 2,200 feet per minute at that point. The bore is 4.00 inches and the port cross section is 2.44 square inches

It should be relatively easy to adapt the math. Piston=driver cone, etc. I don't know at what speed the air flow becomes audible ("chuffing") but the math is out there for port flow. I mentioned "cross pollination" of technology and information in a recent rant in Homebrew, and this is a perfect example.

Pvel = (2,200 ÷ 60) x (4.002 ÷ 2.44) = 240.4 feet per second

[Zvu]

Hi Alaric.

Calculation of air velocity through port is something that most of the calculators do, calculating resonant frequency of the port also... The thing no calculator known to me does is how bad that resonance is going to be.

EDIT:

I'm playing with Seas L15RLY-P right now. Unibox says there's 830Hz resonance, and even estimates how bad is it. We'll see if it's any good in test cabinets for my new project.

https://s27.postimg.org/vlx8hgsab/2_kom_Seas_L15_RLY-_P.gif

Unibox prediction of port resonances pretty much matches frequencies i measure them at in my current 2-way:

https://s28.postimg.org/61c973nkt/SB17_NRXC35-8.gif

[Alaric]

The final part of the calculation would be the audio specific bits (resonance, audible flow noise, etc.). The crossover engine math hits a wall when you realize the object is to increase port velocity. So you have to work backwards from that.

[keith_correa]

Is this Power Port spreadsheet what you're looking for?

powerport.zip