Flared ports and turbulence

[Collo]

How much difference do flared ports make?
I have scoured the net for empirical data on this subject to no avail.

Finally I decided to make a project of it and do my own testing.

Using 90mm ports with a range of flare sizes, I determined the velocity at which "chuffing" occurs.

Unflared - chuffed at all useable speeds
10mm flare - chuffed at 8 metres per second
25mm flare - chuffed at 13 metres per second
35mm flare - chuffed at 17 metres per second
50mm flare - chuffed at 23 metres per second

Along the way, I had to learn different techniques for producing flares - using a router, using a heat gun over a mold, and making "one off" flares with "bondo" type automotive body filler.

Hopefully these results will be useful to the DIY community.

To put my methods under the spotlight, see the writeup at
http://www.users.bigpond.com/bcolliso/flare-testing.htm

All constructive criticism is welcome

regards
Collo

[GregoryD]

Very impressive work so far! I do have some questions though.
It looks like all your ports are the same length and do not take into consideration that the tuning frequency of the box changes slightly with the flare difference. That is, the box is being tuned at a different frequency for each flare radius. If this tuning frequency changes and you are always measuring at the same frequency (30Hz.) then the results may be skewed due to decreased/increased output from the port. Did you verify the port tuning for each of the test ports? An easy way to get close is to looks for the frequency that provides the minimum cone motion for a constant signal in. If this has been already been accounted for...then I missed it in your write-up. Keep up the good work!
GregoryD

[Collo]

Thanks for those comments.

Near the start of the article is a link to "more on construction of test ports" which takes you to another page with pictures and a table of the physical length for each port.

After asking around on the web, I found that the generally accepted approach is to assign half the length of the flare to the port.

All the ports have an acoustic length of 300mm, so, for example, the port with a 50mm flare each end has a physical length of 350mm.

[Ron E]

I would suggest you measure the output voltage of the amp at the onset of chuffing, then you could get an estimate of velocity and SPL by plugging in this number into your program.

Measuring within a foot or so of the outlet will allow you to neglect room effects for the most part and more easily compare the results. You could even record the sound for each case, notch out the fundamental and have a more objective measure of chuffing.

[Svante]

Great work! The community sure needs more empirical data!

[Collo]

Originally posted by Ron E
I would suggest you measure the output voltage of the amp

RonE....Thanks for that suggestion. I had considered this approach, but wasn't sure if picking the impedance from the WinISD graph and applying ohms law was valid.

I've attached the impedence graph, showing 4.099ohms @ 30hz, the test frequency.
Using a multimeter would give me the RMS voltage.

Can I simply use P=(E^2)/R to calculate the power?


The reason I conducted the testing in-room was to combine the outputs of the front mounted port with the side mounted drivers.

If I can use your method, life would be much easier. At least for the single frequency part of the tests.

As for detecting the "chuffing", it is quite easy to be consistent. As you approach the "chuff zone", the nature of the sound changes noticably when you increase the level by 1dB

[paulspencer]

Bill, that's great work! A very useful result. It just goes to show that the standards that are used aren't very helpful - ie 17m/s preferable and 34 m/s absolute limit. In most cases, 17m/s will be inadequate, however from your results I suspect that it may be possible to get acceptable performance at 34m/s if the flare is larger, the subwoofer is far enough away with an undisturbed vent exit and the overall SPL level is sufficient to have a masking effect.

My main question is will others find that they can get away with higher vent velocities? Are your figures based on the ideal chuff free port which will not chuff unless you put your ear right up to it?

Another area of investigation would be dimpled ports B&W style. It's a pity you aren't in Melbourne as I have some dimpled vents which I haven't gotten around to testing properly. I'd like to do some similar testing and see what numbers I come up with. My flares are 18mm, 36mm and 72mm radius with a 100mm vent, although I'd probably cheat a little and use the same vent length since my vents tuned lower would get expensive with the required length of PVC!

Again, some very interesting results. It's actually surprising that more diyers don't focus more on getting vents right, rather than just using a passive radiator at extra expense. This is in the spirit of diy ie. put in time and innovation and spend more on drivers rather than throwing money at the problem. This illustrates what I have pointed out many times - vent flares make a huge difference. They are probably even worthwhile going up to 100mm flare radius.

[Collo]

Thanks for those comments Paul.

re your query on an acceptable level of chuffing
The test results list the level at which chuffing was first detected. At the seating position, you could probably go an extra 3dB, but not more than that. Re-working the figures with 3dB added for the upper limit gives:

for a 90mm diameter port
10mm flare min 8m/s max 11m/s
25mm flare min 14m/s max 20m/s
35mm flare min 17m/s max 24m/s
50mm flare min 24m/s max 35m/s

So yes, you can go to 34m/sec but only with a 50mm flare or larger each end, and if you're sure you won't occasionally exceed this speed on peaks. An even larger flare would give more improvement, although it is a case of diminishing returns and fitting it in could be a challenge.


Dimples obviously make a difference. I have heard Rabbitz's sub with the B&W "flowport" and can confirm that it seems to work quite well. It would be good if you were able to do a comparison of "dimpled" vs "non-dimpled" for the same flare radius.
With a 100mm dia port, you'll need to push 25% more air than with the 90mm dia that I used. I'm sure your setup would have no worries in that department.

The "masking effect" you speak of would depend on the composition of the signal. For example a passage with most of its spectral content well above the tuning frequency would mask out some chuffing. Another passage with most of its content at the tuning frequency would not mask the "chuff". I think if you designed this way, you would get caught out.


Having only tested one diameter of ports, it becomes an educated guess for other sizes. I would say that the required flare size depends on the ratio of the flared / unflared areas. This logic would give:

for 8m/sec a 90mm port needs a 10mm flare, but a 100mm port would need an 11mm flare
for 14m/sec a 90mm port needs a 25mm flare, but a 100mm port would need a 27mm flare
for 17m/sec a 90mm port needs a 35mm flare, but a 100mm port would need a 40mm flare
for 24m/sec a 90mm port needs a 50mm flare, but a 100mm port would need a 61mm flare

If I was in the US and using a 100mm "Precision Port" (which is about 35mm flare), I would know that my sub would be noise free below about 16m/sec and I would design to not exceed about 22 m/sec

[paulspencer]

Bill, whether or not the radius needs to be scaled up as a ratio remains to be seen. I'm not sure that you really do need to. I think it could go either way. I'm hoping that with my 4" vents with 72mm flare radius I can get away with 35m/s as in some simulations it's been difficult to get the velocity below this with my crazy drivers!

One thing I've looked into is designing a box a bit larger than normally optimal, to allow for the effect of the rumble filter, which turns the F3 into an F6! This increases the vent velocity a little.

[morbo]

excellent post and tests, thanks for doing this!