Hello,
Long time lurker, first time poster. I was wondering if I could get some advice on where to tackle issues i'm running into with building ported boxes:
The boxes I've been making to WinISD's spec seem to always (8 or 10 attempts) have the port tuned 5-7 Hz lower than expected. I'm using online calculators to proven alignments and drivers recommended in appropriate enclosures as a starting point.
My attempt:
I'm starting with an Eminence Kappalite 3015LF, recommended volume of 3.2 cu ft, modified port size but same tuning frequency (53 Hz). Port calculations seem to be consistent across the different calculators I've tried in spot checking my work; this leaves me with volume/loss/driver calculations being the root of my problem.
I account for internal volumes with a homemade calculator accounting for drivers, ports, braces, adding back any holes made to attach said objects (like driver holes). Recently I've added a 3% fudge factor to reduce the volume to account for losses and still coming up short on frequency targets.
Attached is the impedance graph from the latest design. The box is the recommended size, adjusting for internal displacement, and an attempt at a 53 Hz port. It started out at 46 Hz and shortening the ports (and enlarging the box) got me to 49ish Hz. 5 of the 6 sides sides are braced with truncated-triangular braces and 2 sides padded with 1 in acoustic foam.
The results are ok, but in attempts to pursue perfection (and willing to accepting 95% of it) I'd like your (collectively) take on how to approach this problem to improve my understanding and produce less scrap in the process on future builds. Thanks for the help!
Long time lurker, first time poster. I was wondering if I could get some advice on where to tackle issues i'm running into with building ported boxes:
The boxes I've been making to WinISD's spec seem to always (8 or 10 attempts) have the port tuned 5-7 Hz lower than expected. I'm using online calculators to proven alignments and drivers recommended in appropriate enclosures as a starting point.
My attempt:
I'm starting with an Eminence Kappalite 3015LF, recommended volume of 3.2 cu ft, modified port size but same tuning frequency (53 Hz). Port calculations seem to be consistent across the different calculators I've tried in spot checking my work; this leaves me with volume/loss/driver calculations being the root of my problem.
I account for internal volumes with a homemade calculator accounting for drivers, ports, braces, adding back any holes made to attach said objects (like driver holes). Recently I've added a 3% fudge factor to reduce the volume to account for losses and still coming up short on frequency targets.
Attached is the impedance graph from the latest design. The box is the recommended size, adjusting for internal displacement, and an attempt at a 53 Hz port. It started out at 46 Hz and shortening the ports (and enlarging the box) got me to 49ish Hz. 5 of the 6 sides sides are braced with truncated-triangular braces and 2 sides padded with 1 in acoustic foam.
The results are ok, but in attempts to pursue perfection (and willing to accepting 95% of it) I'd like your (collectively) take on how to approach this problem to improve my understanding and produce less scrap in the process on future builds. Thanks for the help!
Interesting! Me n' other advanced DIYers used the Pro version for years with accurate 'enough' results, though have seen some bad reviews re the latest version. As for during my active DIY speaker design/building 'career', I always empirically cut to length using a frequency generator and o'scope to find Fs, Fb, impedance peak(s).
That said, what driver specs are you using? Factory specs dictates a T/S max flat alignment = 8.2+ ft^3 net box volume (Vb), which seems reasonable since this driver has a ~0.46 Qts + any added series resistance, so will be higher = even bigger box = tuning below Fs.
This all based on a T/S max flat alignment = Vb = Vas, Fb = Fs where Qts' = ~0.403
3.2 ft^3 net is a semi-high Qtc sealed in size, so extremely under-damped if vented unless tuned really low and/or the vent is damped to near aperiodic.
[Qts']: [Qts] + any added series resistance [Rs]: http://www.mh-audio.nl/Calculators/newqts.html
[Rs] = 0.5 ohm minimum for wiring, so may be higher if a super small gauge is used as a series resistor and/or there's other series resistance.
That said, what driver specs are you using? Factory specs dictates a T/S max flat alignment = 8.2+ ft^3 net box volume (Vb), which seems reasonable since this driver has a ~0.46 Qts + any added series resistance, so will be higher = even bigger box = tuning below Fs.
This all based on a T/S max flat alignment = Vb = Vas, Fb = Fs where Qts' = ~0.403
3.2 ft^3 net is a semi-high Qtc sealed in size, so extremely under-damped if vented unless tuned really low and/or the vent is damped to near aperiodic.
[Qts']: [Qts] + any added series resistance [Rs]: http://www.mh-audio.nl/Calculators/newqts.html
[Rs] = 0.5 ohm minimum for wiring, so may be higher if a super small gauge is used as a series resistor and/or there's other series resistance.
Thank you for the reply, I should clarify, I'm not saying winISD is wrong. I'm saying I'm having problems translating the simulation to results. I'm doing all I can think of to get as close as possible and still missing the mark.
The port shortening was my attempt at fixing the port tuning issue while knowingly reducing the acoustic loading on the driver.
I'm aware the Qtc is high, it seems to be consistent with eminence suggesting high Q boxes for bass guitar boxes as this is a practice box for a friend.
Ultimately, the tuning frequency of a box doesn't have anything to do with T/S parameters.
Tuning frequency is just Box volume and port dimensions, and a bit of acoustic voodoo.
If memory serves Richard Small gave:
Length=2350*Dia^2/(Vb*Fb^2)-0.732*Dia
That is for Length, Diameter in mm and Length>50mm, Vb in liters and Fb in Hz
The 0.732 factor is for one flanged and one free end, you can use 0.85 if both ends are flanged.
Sometimes ports used near the floor or near a wall or corner of a room may perform differently.
If you measure Fb and find it to be different from the predicted value, you can figure out how much to add or subtract by calculating:
deltaL = -deltaFb*Dia^2*2350/(Fb^2*Vb), where deltaFb = Fb_desired-Fb_measured.
Tuning frequency is just Box volume and port dimensions, and a bit of acoustic voodoo.
If memory serves Richard Small gave:
Length=2350*Dia^2/(Vb*Fb^2)-0.732*Dia
That is for Length, Diameter in mm and Length>50mm, Vb in liters and Fb in Hz
The 0.732 factor is for one flanged and one free end, you can use 0.85 if both ends are flanged.
Sometimes ports used near the floor or near a wall or corner of a room may perform differently.
If you measure Fb and find it to be different from the predicted value, you can figure out how much to add or subtract by calculating:
deltaL = -deltaFb*Dia^2*2350/(Fb^2*Vb), where deltaFb = Fb_desired-Fb_measured.
Unless your design is at the critical limits of what the driver can take a few Hz means virtually nothing in the grand scheme of things 🤔
The T/S parameters dictate the size of the box, the size of the box dictates size of the ports.
The acoustic voodoo is the part i'm failing to grasp. I do have them a couple inches off the ground; when i tested the box on the edge of a table top it did tune about a .5Hz higher.
I'll have to try the equation. Empirical solutions work but its a bit of a cop out for a real understanding for the task at hand.
I'm still in the learning phases of the mechanical design. Its not that the mechanical system cannot withstand what I'm attempting to do with it, it does sound quite impressive, but I'm attempting to refine my understanding of the design and debug of speaker design.
I am well aware of what T/S parameters are used for.
The problem is that the end corrections are based on idealizations of the mass of one side of either a piston in the end of a long tube, or a piston in an infinite baffle. You can read about this in L.L. Beranek's Acoustics, if you are so inclined. He won't give you tips for calculating end corrections for weird situations, though. The end correction is the acoustic voodoo. Closed form solutions for acoustic phenomena are hard to come by.
Notice in the equation I left you:
Increasing Diameter increases length a lot (and vice versa) because of the squared diameter term
Increasing Fb decreases length a lot (and vice versa) because of the squared Fb term in the denominator
Increasing Vb decreases length and vice versa in the denominator
IF you have more questions, ask away.
The problem is that the end corrections are based on idealizations of the mass of one side of either a piston in the end of a long tube, or a piston in an infinite baffle. You can read about this in L.L. Beranek's Acoustics, if you are so inclined. He won't give you tips for calculating end corrections for weird situations, though. The end correction is the acoustic voodoo. Closed form solutions for acoustic phenomena are hard to come by.
Notice in the equation I left you:
Increasing Diameter increases length a lot (and vice versa) because of the squared diameter term
Increasing Fb decreases length a lot (and vice versa) because of the squared Fb term in the denominator
Increasing Vb decreases length and vice versa in the denominator
IF you have more questions, ask away.
You're welcome!
In retrospect I was trying to respond to the whole post Vs just the tuning issue, so probably should have just posted the late bjorno's vent 'cheat sheet' for a bit of design theory, though while not all inclusive, may give enough for you to understand the sim Vs measured Fb and didn't mean to imply the speaker alignment was relevant, which as Ron noted is totally irrelevant since a vented box is just an acoustic band-stop filter.
Also as Ron noted is needing to use the right vent pipe end corrections (each end has one) based on where these termination(s) in the cab are as well as in relation to any other local boundaries.
Also, from very dim memory, WinISD Pro has the option to choose a few different end corrections, though not nearly enough as we ideally need.
The issue is end correction and port efficiency.
If the ports are slots or triangle, the correction factor available in WinIsd is not high enough.
sharing 2 or 3 walls needs much higher correction factor.
likewise if round ports but relatively close to the walls, same issue they can appear longer.
if the efficiency of the port is low, they will also behave as if lower.
Slot or triangles, even modeled with proper end correction. tend to be lower in real life because of low efficiency.
Round ports with higher efficiency can tend to be higher in tune in real life.
if showing lower, usually because mounted close to walls.
Boxes/ speaker cones also have leakage, and there is absorption loses from absorption material.
3 to 5 Hz of real world error, usually means you built the box right and calculated correctly.
sounds about normal to me.
Its not a problem, but common for people to over analyze speaker behavior in models and twittle with tuning to much.
In real life a reflex is a reflex. With pro audio or bass guitar boxes your usually just trying be 3 to 5 Hz above or below Fs
and call it a day. The actual 1/2 space or Full space frequency response in real life is no where near the model.
lower tuning will keep it from unloading easier. but even then 3 to 5 Hz aint gonna change much.
Bass guitar preamp might have 10 up to 18 dB boost with the EQ. Slight bump of the bass knob does way more
than twittles with modeling software and tuning.
If the ports are slots or triangle, the correction factor available in WinIsd is not high enough.
sharing 2 or 3 walls needs much higher correction factor.
likewise if round ports but relatively close to the walls, same issue they can appear longer.
if the efficiency of the port is low, they will also behave as if lower.
Slot or triangles, even modeled with proper end correction. tend to be lower in real life because of low efficiency.
Round ports with higher efficiency can tend to be higher in tune in real life.
if showing lower, usually because mounted close to walls.
Boxes/ speaker cones also have leakage, and there is absorption loses from absorption material.
3 to 5 Hz of real world error, usually means you built the box right and calculated correctly.
sounds about normal to me.
Its not a problem, but common for people to over analyze speaker behavior in models and twittle with tuning to much.
In real life a reflex is a reflex. With pro audio or bass guitar boxes your usually just trying be 3 to 5 Hz above or below Fs
and call it a day. The actual 1/2 space or Full space frequency response in real life is no where near the model.
lower tuning will keep it from unloading easier. but even then 3 to 5 Hz aint gonna change much.
Bass guitar preamp might have 10 up to 18 dB boost with the EQ. Slight bump of the bass knob does way more
than twittles with modeling software and tuning.
Last edited:
Keep in mind actual port resonance is when there is least amount of cone movement.
Hard to measure, so usually the impedance curve is used. But wont always reflect exactly
the reflex point.
Also keep in mind advanced parameters in WinIsd are
leakage Ql = 10
Absorption Loss Qa =100
Port Efficiency Qp = 100
For vented cabinet Ql is closer to 7.
once your over 50 to 60 liters, or 85 liters in your case Ql can be closer to 6.5 even 6
likewise paper cone accordion edge live sound drivers can have leakage.
Qa of 100 assume no loss, cabinet has light to moderate lining Qa should be 60 to 50
Qp of 100 assumes perfect port, none of them are 70 to 50 more realistic
likely used end correction of .732 assume every flare is different and being close
to cabinet walls end correction is closer to .750 or 800
WinIsd can only be set to the next step up .850
Here is our WinIsd Model
3.2 CuFt , 53 Hz Fb
Using default
Ql =10
Qa =100
Qp = 100
End correction = .732
The tuning frequency of the model is 53 Hz.
Grey impedance curve shows 51.1 Hz at lowest point
even perfect model reflects different tuning by measuring with impedance
curve.
Orange impedance, more realistic
Ql =7
Qa = 50
Qp = 60
end correction = .732
Tuning frequency of model still 53 Hz
Winisd will not adjust tuning frequency for advanced features.
But WinIsd will show effect on Impedance curve
lowest point of impedance curve 50.1 Hz
another shift with losses
using same vent length with higher end correction 47.62 Hz
Your build was built to spec
The model needs to assume leakage, Absorption, and port losses
to match the impedance curve.
Far as actual resonance could still be higher according to cone movement.
Even a perfect model is 3 to 5 Hz different than real life according to the impedance
curve.
Hard to measure, so usually the impedance curve is used. But wont always reflect exactly
the reflex point.
Also keep in mind advanced parameters in WinIsd are
leakage Ql = 10
Absorption Loss Qa =100
Port Efficiency Qp = 100
For vented cabinet Ql is closer to 7.
once your over 50 to 60 liters, or 85 liters in your case Ql can be closer to 6.5 even 6
likewise paper cone accordion edge live sound drivers can have leakage.
Qa of 100 assume no loss, cabinet has light to moderate lining Qa should be 60 to 50
Qp of 100 assumes perfect port, none of them are 70 to 50 more realistic
likely used end correction of .732 assume every flare is different and being close
to cabinet walls end correction is closer to .750 or 800
WinIsd can only be set to the next step up .850
Here is our WinIsd Model
3.2 CuFt , 53 Hz Fb
Using default
Ql =10
Qa =100
Qp = 100
End correction = .732
The tuning frequency of the model is 53 Hz.
Grey impedance curve shows 51.1 Hz at lowest point
even perfect model reflects different tuning by measuring with impedance
curve.
Orange impedance, more realistic
Ql =7
Qa = 50
Qp = 60
end correction = .732
Tuning frequency of model still 53 Hz
Winisd will not adjust tuning frequency for advanced features.
But WinIsd will show effect on Impedance curve
lowest point of impedance curve 50.1 Hz
another shift with losses
using same vent length with higher end correction 47.62 Hz
Your build was built to spec
The model needs to assume leakage, Absorption, and port losses
to match the impedance curve.
Far as actual resonance could still be higher according to cone movement.
Even a perfect model is 3 to 5 Hz different than real life according to the impedance
curve.
As above but also think about how hard this is going to be driven, driver TS will be way off when running into power compression and therefore so will your predicted system response.
When you said advanced features, I thought about the TL model and clicked it. The impedance came out to 1 Hz off measured (47.5 vs 46.7). Would that have anything to do with this? I can't seem to find much information about that setting.
Am I better off checking my sim in hornesp? From my previous experience with both, WinISD has an almost identical output to hornresp in an infinite baffle case. It's just easier with the default alignments to use WinISD.
With all that said, I cut new pvc and installed them in the box after reading your response. The acoustic efficiency is significantly better with the calculated length ports vs the shortened ones. The low end is much more impactful with better transients also so i do believe what you're telling me. I just didn't know why it worked, just that it worked. Thank you!
It is a learning curve with with the model and real world.
just wanted to assure you 3 to 5 Hz was not a big deal.
But the model can show you, why it happened.
leakage, and absorption losses can cause small 1 to 3 Hz jumps.
Far as the ports themselves end correction is the main culprit for big jumps.
If you were building a 1x10 box, it could be more frustrating
smaller boxes are less forgiving with port length.
Smaller 25 to 35 liter boxes will have bigger jumps.
Larger 80 to 170 liter boxes more forgiving.
But you can still expect the usual 3 to 7 Hz error factor.
Your case if using single flare, round ports not sharing walls.
.732 end correction is correct. But it is likely, since it is hard to
predict real world leakage and behavior. If the port is close to
side walls or rear panels. The end correction is likely .800 to .820
but WinIsd will just jump to .850
So to get closer you need to predict length with .732 end correction.
the length is correct. Then change WinIsd to .850 or dual flare.
The number will auto change. Manually enter the previous longer length.
You will likely see a 3 to 5 hz Hz change.
depends on box volume. As mentioned smaller boxes, bigger jump
VirtuixCad allows end correction higher than .850
If your building ports with shared walls, like 2 wall triangle ports or 3 wall
slot ports. End correction goes above the decimal point.
2 wall = 1.728 to 1.850
3 wall = 2.227 to 2.350
.732 is common for single flare or no flare
.614 is common for no flare but find unrealistic.
at least depending how close it is to the side or rear wall.
Usually with common off the shelf cut to length ports.
especially ones with multiple fittings and size variation.
I find .800 to be a reasonable end correction.
Then assume Ql of 7 and Qp / Qa around 50 to 60
or basically leakage and absorption in typical light lined box.
Within reason once your up to 3.8 or 5 CuFt you can assume
even more leakage around 6.5 to 6
just wanted to assure you 3 to 5 Hz was not a big deal.
But the model can show you, why it happened.
leakage, and absorption losses can cause small 1 to 3 Hz jumps.
Far as the ports themselves end correction is the main culprit for big jumps.
If you were building a 1x10 box, it could be more frustrating
smaller boxes are less forgiving with port length.
Smaller 25 to 35 liter boxes will have bigger jumps.
Larger 80 to 170 liter boxes more forgiving.
But you can still expect the usual 3 to 7 Hz error factor.
Your case if using single flare, round ports not sharing walls.
.732 end correction is correct. But it is likely, since it is hard to
predict real world leakage and behavior. If the port is close to
side walls or rear panels. The end correction is likely .800 to .820
but WinIsd will just jump to .850
So to get closer you need to predict length with .732 end correction.
the length is correct. Then change WinIsd to .850 or dual flare.
The number will auto change. Manually enter the previous longer length.
You will likely see a 3 to 5 hz Hz change.
depends on box volume. As mentioned smaller boxes, bigger jump
VirtuixCad allows end correction higher than .850
If your building ports with shared walls, like 2 wall triangle ports or 3 wall
slot ports. End correction goes above the decimal point.
2 wall = 1.728 to 1.850
3 wall = 2.227 to 2.350
.732 is common for single flare or no flare
.614 is common for no flare but find unrealistic.
at least depending how close it is to the side or rear wall.
Usually with common off the shelf cut to length ports.
especially ones with multiple fittings and size variation.
I find .800 to be a reasonable end correction.
Then assume Ql of 7 and Qp / Qa around 50 to 60
or basically leakage and absorption in typical light lined box.
Within reason once your up to 3.8 or 5 CuFt you can assume
even more leakage around 6.5 to 6
How close is close to a wall? I've been told or read 1" in any direction to half the ID of the port. I'm approximately 1.5" away in both directions currently which is between those rules of thumb.
In the case of slot loading, does the correction factor pertain to the larger or smaller dimension?
not related to larger or smaller dimension.
The equation is based on the cross sectional area.
Doesn't matter if it is Circle , square, rectangle.
The port opening has a certain area.
WinIsd and many other models are just convenient because you can
select circular or square, and when you enter the numbers
it is basically a spread sheet and calculates the area for you.
You will notice in the project box in vent section
it will display cross sectional area.
Anyways not to be confusing.
End correction will basically adjust the length of the port.
assuming a slot port with 3 shared walls.
sharing walls will make the port appear longer.
So the correction factor will shorten the port.
winisd has " two flared ends option"
the actual constant is .850
for 3 shared walls the constant is 2.227
In a nutshell if you calculate a slot port
with winisd it will be wayyy to long.
The equation is based on the cross sectional area.
Doesn't matter if it is Circle , square, rectangle.
The port opening has a certain area.
WinIsd and many other models are just convenient because you can
select circular or square, and when you enter the numbers
it is basically a spread sheet and calculates the area for you.
You will notice in the project box in vent section
it will display cross sectional area.
Anyways not to be confusing.
End correction will basically adjust the length of the port.
assuming a slot port with 3 shared walls.
sharing walls will make the port appear longer.
So the correction factor will shorten the port.
winisd has " two flared ends option"
the actual constant is .850
for 3 shared walls the constant is 2.227
In a nutshell if you calculate a slot port
with winisd it will be wayyy to long.
A few pointers I have found invaluable over the past four decades:
Having noted these points, be aware that ported enclosures have awful transient response and high energy storage, so if you want a good sounding sub it should be sealed with a Q of around 0.6, but expect it to require EQ and use heaps of power. Ported remains king for HT rumble though.
- Begin the design with a port area equal to Sd and work backwards until it fits in the box. Size matters.
- Port length should not exceed twice its diameter, and be at least its own diameter away from any enclosure wall.
- Flow should be equalised in both directions to maintain linearity and reduce port rectification.
- Port radius is most important and depends upon application - it's a bit of a 'Hoffman' situation where optimising one aspect detracts from another.
- NO BENDS!
- A single circular port remains optimum; slots are the worst.
- Proper port design is a win-win; increased efficiency and lower distortion.
Having noted these points, be aware that ported enclosures have awful transient response and high energy storage, so if you want a good sounding sub it should be sealed with a Q of around 0.6, but expect it to require EQ and use heaps of power. Ported remains king for HT rumble though.
I appreciate the pointers!
This started as an exercise in validating my calculator/design. This is the third design with this method and the first 2 were almost spot on so I was wondering why this one was so far off knowing this is the largest project I've attempted thus far.
Part of the reason for that driver selection was the ability to be used in a sealed box (based on a .46 Qts) if a ported box would not work.
If I must use a slotted port, do I calculate the theoretical diameter of the round port the cross sectional area as the square and calculate the shortened length?
FYI/FWIW, the pioneers addressed this issue re horn mouth terminations in Chapter 11 beginning on pg. 234: 'Horn Design Data'.
Iam not sure if I understood the question, but if you need to calculate equivalent diameter (called hydraulic diameter) for square or rectangular (slotted) port, here are the formulaes
If your using a rectangle or a square for the slot port.
WinIsd is using the same formulas above in post #19
It is basically a spread sheet.
Select the square port option and enter your slot port dimensions
and it will calculate the area or cross sectional area of the port.
In order to use the 2.227 end correction for 3 shared walls.
You will need to use another Modeling program
where you can enter the 2.227 end correction value.
I use VituixCAD
Free to download/ install.
And use the enclosure tool.
Far as Area of a Rectangle can also use numerous online calculators.
Area of Rectangle
As mentioned, WinIsd will essentially do the same thing
It has options for circle port, or square port.
and will also do rectangle in the square option.
Only other common shape is triangle port
which WinIsd wont do.
SSS Triangle Calculator
Triangle shares 2 walls
K or end correction = 1.728
No shared walls .732
1 shared =1.23
2 shared =1.728
3 shared =2.227
Last edited: