Cab building question about shelf port length...

[J-wall]

You may remember my last thread:

TL606 or fEARful with 3015 non-LF?

I have learned a lot since then and have settled on basically cloning (or trying to clone) a BareFaced Compact 115.

Internal volume is 3.18 ft3 (before bracing is added), tuning I want is 50hz, speaker used is a 3015 non-LF. Target weight is 30lbs. Bracing will be fEARful-esque on the top half around the woofer and the shelf port vertical supports will run all the way to the back, but there will be dowels front to back and side to side at the bottom half.

My question now concerns port shelf length as it relates to port air velocity.

I have been told that you want port air velocity to be 17m/s or less, and WinISD I have been told needs the length of rectangular ports reduced by 30% because it is set up with a focus on round ports not rectangular shelves. Now, with the tuning and volume listed above and an overall port area of 17.5" wide and 2.125" high the program calculates a port length of 7.51" and a port air velocity of 16.6 m/s which is great.

However, when that length is reduced 30% to 5.25", the program calculates that the tuning of the box goes up to 55hz, and the port air velocity goes up to 17.4 m/s.

Which do I trust?

Also, the baffle is recessed from the front of the cab, so to calculate port length, do I start from the front/rear of the recessed baffle board or from the actual front of the shelf which will be 7/8" longer than the front of the baffle?

Thanks in advance!

Josh

[billfitzmaurice]

Port velocity is basically determined by port area. You choose the port area to be large enough to keep the port velocity tolerable, then adjust the duct length to reach the desired tuning. But changing the tuning will alter the port velocity somewhat, as it's tied to cone excursion, which is also tied to tuning, so you may have to play with the dimensions a bit to get an 'ideal' result. Also be sure that the port velocity is calculated with a high power input, not the 1 watt program default.

As for the difference in the duct length for shelf versus other shapes, unless the width to height ratio of the port is quite high, as in at least 10:1 and/or the port is less than 1.5 inches on one dimension, there's little to no difference.

[J-wall]

Yet another reason Bill is worthy of an appreciation thread. I may not have the answers to my questions specifically from that answer, but I am smarter overall and I feel empowered to proceed. Thanks, Bill!

[bobcruz]

The port length includes the extension through the baffle to the actual opening at the front of the cab.

I'd go with the 7.5 inch port length. Unless your port is backed up against the rear wall of the cabinet, I don't think it's something to worry about.

[will33]

If you're unsure of the tuning and don't want to have to saw out stuff you glued in permanently, you can build the shelf in more than one piece. For example, glue in a 5.5" shelf and have a 2" extension, or a pair of one inch extenstions. Try various lengths before gluing them permanently. I find an actual gig is the best place to test this stuff. Biggest differences will be heard in boomy or bad sounding rooms.

When you do glue it permanent, just make sure everything lines up straight so the whole port path is smooth, nothing to cause air turbulence.

[rpsands]

Winisd cannot be trusted to calculate shelf ports correctly. Do not trust it

* Also be sure to calculate the amount of space the woofer displaces. It can be a sizable chunk (up to around .1cf or so on the high side).

[will33]

Originally Posted by rpsands

Winisd cannot be trusted to calculate shelf ports correctly. Do not trust it

* Also be sure to calculate the amount of space the woofer displaces. It can be a sizable chunk (up to around .1cf or so on the high side).

Ya, I had so much headache with one, I patched it all in and stuck pipes in it, hence the shelf extensions and testing by ear.

OP, as far as your input on the model, I use the max power graph. For example, if your speaker hits xmax at 200 watts, set the input to at least 200 watts when looking at the air velocity. Maybe a little more just to be sure. Just looking at your port dimensions, you shouldn't have any chuffing problems, it's just getting the depth right.

[J-wall]

Thanks, everyone.

I guess the moral is to make it adjustable. I will re-cut accordingly, construction has yet to begin.

Originally Posted by rpsands

Winisd cannot be trusted to calculate shelf ports correctly. Do not trust it

Does this mean what you said elsewhere, that the overall length needed to be reduced by 30% - and if that is done, will the target tuning be achieved?

I have no idea how to measure tuning.

Thanks!

[billfitzmaurice]

Originally Posted by J-wall

Thanks, everyone.

I guess the moral is to make it adjustable. I will re-cut accordingly, construction has yet to begin.



Does this mean what you said elsewhere, that the overall length needed to be reduced by 30% - and if that is done, will the target tuning be achieved?

I have no idea how to measure tuning.

Thanks!

IME there's no reduction of length with shelf ports unless they're very narrow, an inch and a half or less. There's no specific formula to calculate the reduction in those cases. To measure tuning the easiest method is with an impedance sweep, but you need gear for that. An easy method is to put the cab face up, sprinkle sugar on the cone, then use a sine wave generator into your amp to run a frequency sweep. At the tuning frequency the cone will have the least possible excursion. You'll see just above and below the tuning frequency that the sugar 'dances', but it takes a rest at Fb.

[rpsands]

IIRC, the number that Winisd generates needs to get divided by 1.5 (approximately). then you refactor including the volume behind the port in the cabinet volume. Repeat as many times as you like.

The sweep is an excellent idea. I've found the car audio port calculator on the web to be fairly accurate (e.g. I've used it to reverse engineer the tuning frequency on fEarfuls which've been measured to be 47hz or so and come within 5%-ish).

Car Audio - PORT Size Calculations and Formulas for WOOFER and Subwoofer BOXES

From what I can tell the port divider makes it act like two separate ports; so in the case of a fEarful, you do two 3.0 x 9.5" ports.

Hope that helps. Also hope someone who knows for sure on the port numbers can weigh in there It seems empirically correct, but I've only tried it with one box I actually built (and its tuning frequency came out within 2% or so of what I had calculated with the car audio calc.).

[J-wall]

I am getting similar numbers for shelf length from WinISD and that online calculator.

I am a bit confused...

[J-wall]

Awesome - I got it to be withing a tenth of an inch using the dividend factor of 1.5 with WinISD and the online port calculator! Woo Hoo! I need to lengthen my port shelf about 1/4" to be spot on and reduce the slot from 3 vents (like a TL606) to two (like a fEARful).

Thanks again for everyone for your help, I really appreciate it! Pics to come within a week or so, I hope!

[parapentep70]

2 comments:

I learnt the hard way it is much easier to add than to cut sections of the shelf. And I was dry-fitting the box! But I glued the shelf to the port divider and removing a couple of cm was difficult enough. Then I started with a short shelf for my 2nd speaker.


I used as well calculators for port length estimation. But I actually tuned the box by measuring the impedance of the speaker in the box. I have high end equipment at work... not required at all, so I did it at home.

I used a computer, a soundcard, a DAW SW (I used Reaper, I paid a license but you can evaluate the SW for as long as you honestly have to for free) and some free plug-in (BlueCat spectrum analizer, not required but handy).

I wired the soundcard out (limited impedance, 10K if memory serves) directly to the driver. Then voltage is proportional (well, ALMOST proportional) to driver impedance, since the source impedance is some thousand times higher than the driver impedance. I used an input to the soundcard wired in parallel. Then, driving tones at constant amplitude and different frequencies I measured the impedance of the driver in the box by measuring the level in the soundcard input. If your soundcard has a much lower output impedance you can add a reasonable resistor in series, something in the order of 1K to 10K will be OK. It it is less your reading will not be proportional to the impedance (it's still OKK since you are interested in the position of the minimum). If it is a lot higher, your level in the soundcard input will be too low.

I tuned to get a minimum in the desired freq. (47 Hz if I remember well). First time my tuning was too low, I cut some 2cm from the shelf and tried again. In the second cab (slightly different) I started with a short shelf. It was much easier to add, it can be done even when the cab is finished.

The freq at which you tune the box is usually close to the resonant freq. of the driver, but they are different things. Just in case I could be wrong I broke-in my first Kappalite 3015. I observed a clear reduction of resonance freq. to where it should be. A lot, something like 4 Hz or so. But the frequency at which I tuned the box (for the theoretical 3015 model used in WinISD) did not change at all when I tried with the broke-in driver. Which makes perfect sense, but I was not 100% comfortable being all values so close.

Now I am confident that my boxes are tuned properly to meet my goals.

[parapentep70]

Originally Posted by J-wall

Awesome - I got it to be withing a tenth of an inch using the dividend factor of 1.5 with WinISD and the online port calculator! Woo Hoo! I need to lengthen my port shelf about 1/4" to be spot on and reduce the slot from 3 vents (like a TL606) to two (like a fEARful).

Thanks again for everyone for your help, I really appreciate it! Pics to come within a week or so, I hope!

I also used this 1.5 factor I read somewhere and I ended with a shelf some 2 cm longer than needed. So my tunning was very near what I wanted.

I must say that my ports for the 3015-non-LF are on the small side and also narrow. I think this could explain the "increase of effective length" respect to the 1.5 factor. That's why I wanted to measure the tuning of the box.

[billfitzmaurice]

Originally Posted by parapentep70

I also used this 1.5 factor I read somewhere and I ended with a shelf some 2 cm longer than needed. So my tunning was very near what I wanted.

I must say that my ports for the 3015-non-LF are on the small side and also narrow. I think this could explain the "increase of effective length" respect to the 1.5 factor. That's why I wanted to measure the tuning of the box.

FWIW the extra virtual length of narrow ports results from air friction against the duct walls. Said friction takes place roughly within 1/4 inch of the duct walls, so if the friction zone makes up a substantial percentage of the duct area the effect is noticeable. Ports with too small an area exacerbate the situation.

[parapentep70]

Bill, this maybe sidetracking but... I noticed my impedance plot was also sligthly "flatter" than predicted by simulations. My intuition on this is that this "air friction" (i.e. losses) reduces the Q of the box near its tuning frequency.

If this is correct it does not look that bad to me: slightly above tuning freq I think I have reduced chance of overexcursion, and the low freq "bump" or "peak" in the response that is provided by the port will be a bit less than what it should... Kind of playing on the safe side in the trade off between extra bass and risk of over-excursion above and below the box tuning freq.

I did not do it on purpose, I simply was inexperienced and stingy when sizing the port area (I think I used the criteria not to exceed 1/10th of soundspeed at 400W, worst freq.).

Now I regret what I did because I cannot tune lower by simply closing a section of the port. The port would have an area too small for this.

[Bass Unique]

Originally Posted by parapentep70

Bill, this maybe sidetracking but... I noticed my impedance plot was also sligthly "flatter" than predicted by simulations. My intuition on this is that this "air friction" (i.e. losses) reduces the Q of the box near its tuning frequency.

If this is correct it does not look that bad to me: slightly above tuning freq I think I have reduced chance of overexcursion, and the low freq "bump" or "peak" in the response that is provided by the port will be a bit less than what it should... Kind of playing on the safe side in the trade off between extra bass and risk of over-excursion above and below the box tuning freq.

I did not do it on purpose, I simply was inexperienced and stingy when sizing the port area (I think I used the criteria not to exceed 1/10th of soundspeed at 400W, worst freq.).

Now I regret what I did because I cannot tune lower by simply closing a section of the port. The port would have an area too small for this.

If air friction was making a difference, it would only be at high power. I am assuming you did the impedance curve at low power and therefore low excursion so the air velocity is much lower in the port. You should not really detect a measureable difference between the simulation and reality due to this. Adding some box stuffing will make a bigger difference and flatten slightly (if you put a lot in).

It is more likely that the measurement method was not robust - although you detect a peak at roughly the correct frequency, the flattening occurs when your true impedance peak rises at resonance and starts to mean you no longer have a "virtual infinity" between actual (8 ohms ish rising to 60 ohms or so) and soundcard impedances (10k ohms ish). This will show up as a softening of the curve.

Incidentally, this electronic damping effect will also tend to reduce the apparent tuning frequency by around 1 to 2 % so actual may be a little higher. (1 to 2 Hz will have little audible effect but on the limit could breach Xmax).

The friction effect within the port starts to first become a problem for air velocity (above 20 ish m/s) which then increases by more than you predict. However, under extreme levels, there is some compression of the air within the port and some energy is lost so you get a power compression effect within the port itself meaning less sound energy escapes.

To be honest, you still can reduce the area slightly but do it in the longest dimension of the port to minimise the effect. If you think the port velocity is getting a little high, slightly flare the ends of the port to minimise turbulence.

[parapentep70]

Bass Unique, yes, I did not consider that the effect of air friction could be strongly non-linear .

About the method to tune the box, at the box tuning freq. Z is minimum (althoug it is at or near the peak of the speaker in free air). And the location of min. and max. impedances is not changed by the fact I drive the loudspeaker with finite impedance, provided it is not reactive. Peaks are the order of tens of ohms, no more. The 10K of my source reduces marginally the "Q" in peaks, but must have zero effect in location of peaks (or valleys in the case of port tuning).

And I agree, I cannot hear a 2 Hz difference... but at full power I might have excursion problems slightly above the box tuning freq (in the peak of displacement vs. freq). Now I see that this point may not be exactly in the optimum calculated with small signal due to the non-linearities associated with port turbulence .

I regret not having provided a more generous port but at least I flared the inner (lower) side of the shelf thinking it could do more good than harm.

Thanks!

[will33]

I'm sure all this stuff you guys are discussing makes sense/is valid, etc. but couldn't all that be avoided by just using sufficient port area?

[koobie]

Great thread, thanks to all for sharing your knowledge.