The use of "cone compensation" in THs seems to be a bit controversial, so I think an actual test case that involves measurements showing its impact, or lack thereof, may be a good step.
For those not in the know, "cone compensation" is just the restricting of the cross-sectional area at "S2" in a TH to "compensate" for increase in cross-sectional area at that point due to the driver's cone, the idea being that if this isn't done, the actual compression at that point is not going to be the same as that predicted by HornResp sims, and therefore the response of the TH is also going to be different. Some TH designs include it, some don't. Most people report success with their TH builds, but what's missing from most of those reports is specific data comparing the as-built results to the simulations used for the creation of the TH that was built.
Famous commercial designs that appear to include cone compensation are the TH115 and TH118 from Danley Sound Labs. Danley is basically the "father" of modern THs, so this could be taken as a good sign that cone compensation is a good idea.
But what do actual measured results show?
Well. that's the intent of this exercise.
For those not in the know, "cone compensation" is just the restricting of the cross-sectional area at "S2" in a TH to "compensate" for increase in cross-sectional area at that point due to the driver's cone, the idea being that if this isn't done, the actual compression at that point is not going to be the same as that predicted by HornResp sims, and therefore the response of the TH is also going to be different. Some TH designs include it, some don't. Most people report success with their TH builds, but what's missing from most of those reports is specific data comparing the as-built results to the simulations used for the creation of the TH that was built.
Famous commercial designs that appear to include cone compensation are the TH115 and TH118 from Danley Sound Labs. Danley is basically the "father" of modern THs, so this could be taken as a good sign that cone compensation is a good idea.
But what do actual measured results show?
Well. that's the intent of this exercise.
"cone compensation" - the TH to be tested
I'm going to use my POC3 TH for this exercise. It's a TH based on the Dayton Audio PA310 driver, and it was designed without taking the impact of cone compensation into consideration. While the build came out pretty decent, I've done a few post-build modifications, including doubling up on the thickness of the top panel and reinforcing the front of the bottom panel.
I'm planning one or two more enhancements (like one to deal with a pesky panel resonance just above 200 Hz), and if cone compensation results in improved performance, that will be added to the list as well.
Attached is the current impedance curve of the POC (in blue) compared to the HornResp prediction (green). From my tests, this is what I've come to expect - the lowest peak always comes in at a slightly higher frequency, all of measured peaks are lower than what the sim predicts, but their location (particularly the middle one) are as predicted and the minimums occur at the expected points. This impedance curve also shows the impact of the resonance above 200 Hz, which appears as a blip in the impedance curve.
I'm going to use my POC3 TH for this exercise. It's a TH based on the Dayton Audio PA310 driver, and it was designed without taking the impact of cone compensation into consideration. While the build came out pretty decent, I've done a few post-build modifications, including doubling up on the thickness of the top panel and reinforcing the front of the bottom panel.
I'm planning one or two more enhancements (like one to deal with a pesky panel resonance just above 200 Hz), and if cone compensation results in improved performance, that will be added to the list as well.
Attached is the current impedance curve of the POC (in blue) compared to the HornResp prediction (green). From my tests, this is what I've come to expect - the lowest peak always comes in at a slightly higher frequency, all of measured peaks are lower than what the sim predicts, but their location (particularly the middle one) are as predicted and the minimums occur at the expected points. This impedance curve also shows the impact of the resonance above 200 Hz, which appears as a blip in the impedance curve.
Attachments
"cone compensation" - simulation
HornResp doesn't really provide a means of including cone compensation in a TH sim. Probably the closest approximation would be to increase Vtc by the same amount as the volume contained in front of the cone to approximate a TH WITHOUT cone compensation. I did this for my POC3 sim and included the results below - the compensated version in red, the uncompensated version in gret. The sim suggests that there should be a noticeable change in the FR between 100~130 Hz of 2~3dB, with the uncompensated version showing the deeper response notch. If I had to choose between the two, I'd choose the compensated version - the less notch the better.
HornResp doesn't really provide a means of including cone compensation in a TH sim. Probably the closest approximation would be to increase Vtc by the same amount as the volume contained in front of the cone to approximate a TH WITHOUT cone compensation. I did this for my POC3 sim and included the results below - the compensated version in red, the uncompensated version in gret. The sim suggests that there should be a noticeable change in the FR between 100~130 Hz of 2~3dB, with the uncompensated version showing the deeper response notch. If I had to choose between the two, I'd choose the compensated version - the less notch the better.
Attachments
"cone compensation" - pre-compensated measurements
Well, here's the raw FR of my POC3 TH, before I apply cone compensation.
Kidding.
That's the FR with some DSP to remove some of the response peaks, and extend the bottom end a bit. The measurement was also taken in-house, so it's a little bit wobbly.
If cone compensation changes the response of the TH, then this should show up as a difference in the FR, as long as I duplicate the measurement process exactly, which is something I intend to do.
Well, here's the raw FR of my POC3 TH, before I apply cone compensation.
Kidding.
That's the FR with some DSP to remove some of the response peaks, and extend the bottom end a bit. The measurement was also taken in-house, so it's a little bit wobbly.
If cone compensation changes the response of the TH, then this should show up as a difference in the FR, as long as I duplicate the measurement process exactly, which is something I intend to do.
Attachments
Hello Brian, interesting experiment. I have seen this hourglass-shaped HR contour somewhere. How much of this redundant space is actually reclaimed by the moving cone in action ? it can move up to 20 mm in a large woofer. second, what about the back-side ? the expansion rate becomes irregular, the driver is not simply moved forward or backward : the expansion rate for the first and last bit of the TH change, how does this affect the sound ? Particularly the exit opens less than the rest of the horn. Would it make a difference if the driver would be mounted more inside the throat by doubling the board-thickness on that side ? How do you mount two or more drivers in parallel ? do you change the expansion rate several times like an accordeon ? could you measure the difference of a bottleneck vs simply a thinner throat S1 and S2 ? Interesting Marc
"cone compensation" - implementation
I've decided to use a simple "wedge" inserted at S to implement the cone compensation. The volume of a 90 degree wedge that's 19.5cm x 17.5cm x 8.8cm high works out to 3 liters, and it will be small enough to fit through the cutout for the driver.
I considered using a series of concentric circular cutouts, but I decided that would likely end up being too tedious and wasteful - the plywood I have on hand can find better use elsewhere. The wedge should be simple to make and will only take four pieces to create.
Target Volume = 3.0 liters
Cutout Diameter = 28.3 cm
Panel thickness = 1.7 cm
Wedge
Diagonal Length = 28.3 cm
Width = 17.5 cm
Height = 8.8 cm
Length = 19.5 cm
Volume = 3.0 liters
Panel 1
Length = 19.5 cm
Width = 12.4 cm
Panel 2
Length = 19.5 cm
Width = 10.7 cm
I've decided to use a simple "wedge" inserted at S to implement the cone compensation. The volume of a 90 degree wedge that's 19.5cm x 17.5cm x 8.8cm high works out to 3 liters, and it will be small enough to fit through the cutout for the driver.
I considered using a series of concentric circular cutouts, but I decided that would likely end up being too tedious and wasteful - the plywood I have on hand can find better use elsewhere. The wedge should be simple to make and will only take four pieces to create.
Target Volume = 3.0 liters
Cutout Diameter = 28.3 cm
Panel thickness = 1.7 cm
Wedge
Diagonal Length = 28.3 cm
Width = 17.5 cm
Height = 8.8 cm
Length = 19.5 cm
Volume = 3.0 liters
Panel 1
Length = 19.5 cm
Width = 12.4 cm
Panel 2
Length = 19.5 cm
Width = 10.7 cm
Attachments
"Cone Compensation" - installation
Well, here we are, one wedge, installed and ready for testing.
Hmm... this thing worked out to be pretty high - is the driver going to touch it at high excursion? I think it's safe...
Well, here we are, one wedge, installed and ready for testing.
Hmm... this thing worked out to be pretty high - is the driver going to touch it at high excursion? I think it's safe...
Attachments
Last edited:
"Cone compensation" - Results: Frequency Response
Well, here are the measured frequency response results, which are a bit interesting.
1. There's a noticeable increase in output between 120 Hz and 150 Hz of 2 to 3 dB, similar to that predicted by HornResp.
2. There's a small increase in output at the low end of the TH's passband HornResp suggests that there may be a very slight *decrease* in output at this point.
3. The null around 280 Hz is significantly filled in, which is also predicted by HornResp
4. There's a broadband decrease in distortion, not predicted by HornResp
Well, here are the measured frequency response results, which are a bit interesting.
1. There's a noticeable increase in output between 120 Hz and 150 Hz of 2 to 3 dB, similar to that predicted by HornResp.
2. There's a small increase in output at the low end of the TH's passband HornResp suggests that there may be a very slight *decrease* in output at this point.
3. The null around 280 Hz is significantly filled in, which is also predicted by HornResp
4. There's a broadband decrease in distortion, not predicted by HornResp
Attachments
Last edited:
Cool - wonder if that would also work in this horn by Bill Woods ? http://usr.audioasylum.com/images/2/21168/2Bwith_reflector.gif
"Cone Compensation" - Results: Impedance
Well, here are the measured impedance curves, before and after the installation of the "cone compensation".
If you look at the graph closely, the peaks seem to have moved up very slightly in frequency. the peak around 129 Hz has dropped noticeably, but everywhere else the change is very minor.
Well, here are the measured impedance curves, before and after the installation of the "cone compensation".
If you look at the graph closely, the peaks seem to have moved up very slightly in frequency. the peak around 129 Hz has dropped noticeably, but everywhere else the change is very minor.
Attachments
"Cone Compensation" - Conclusion
The results I obtained with deploying cone compensation in my POC3 suggest that yes, HornResp can be used to sim the possible impact on a TH by adjusting Vtc accordingly i.e.
1. If you do NOT include the volume of air within the driver's cone in the value for Vtc, then the resulting HornResp sim will be closer to a TH with cone compensation in place than one without, which means that the results you see at the upper end of the TH's passband may not be a match for what you're expecting from the sim.
2. To sim the impact of NOT using cone compensation, just add the volume of air contained within the driver's cone to Vtc. HornResp includes a tool that allows you to calculate this quite quickly and easy, just by providing a few dimensions.
There are some other things that showed up in the measurements. Like the broadband decrease in distortion when the cone compensation was added. Hard to draw a definite conclusion that this would happen to all THs if CC is added however, as other things can impact the distortion measurements. The testing would need to be done across a few different THs to see if this decrease in distortion is really a repeatable characteristic.
The results I obtained with deploying cone compensation in my POC3 suggest that yes, HornResp can be used to sim the possible impact on a TH by adjusting Vtc accordingly i.e.
1. If you do NOT include the volume of air within the driver's cone in the value for Vtc, then the resulting HornResp sim will be closer to a TH with cone compensation in place than one without, which means that the results you see at the upper end of the TH's passband may not be a match for what you're expecting from the sim.
2. To sim the impact of NOT using cone compensation, just add the volume of air contained within the driver's cone to Vtc. HornResp includes a tool that allows you to calculate this quite quickly and easy, just by providing a few dimensions.
There are some other things that showed up in the measurements. Like the broadband decrease in distortion when the cone compensation was added. Hard to draw a definite conclusion that this would happen to all THs if CC is added however, as other things can impact the distortion measurements. The testing would need to be done across a few different THs to see if this decrease in distortion is really a repeatable characteristic.
"Cone compensation" - POC4
Using the information I learned from this test, I used HornResp to sim the impact of cone compensation on my POC4 design, which is based on the B&C 18TBX100 driver.
From the attached graph, it looks like HornResp suggests that the response dip between 90 Hz and about 110 Hz will be eliminated if cone compensation is used. Good thing I included it in that design 🙂.
Using the information I learned from this test, I used HornResp to sim the impact of cone compensation on my POC4 design, which is based on the B&C 18TBX100 driver.
From the attached graph, it looks like HornResp suggests that the response dip between 90 Hz and about 110 Hz will be eliminated if cone compensation is used. Good thing I included it in that design 🙂.
Attachments
Well, if you have the HornResp sim for that particular build, we can include the effect of cone compensation in the sim to find out... 🙂
Wow, that's a lot of questions 🙂.
I won't worry about the air displaced while the driver is moving. The idea here for using cone compensation is simple to ensure that the expansion through the horn as it passes the driver as simulated is as close as possible to the built version. In other words, if you do NOT include cone compensation in the built version, you likely need to up the value Vtc in the sim accordingly, to reflect the additional volume of air in the driver's cone. For the sim to be even more accurate, the volume across a small cross-section at S3 or s4 will also have to be reduced, but the net reduction is usually pretty small compared to the total volume there, so the impact is likely to be very minor.
Shown in the attachment is the volume that needs to be reflected as "Vtc" in the TH sim, particularly if the driver is a large one.
Attachments
Mr. Steele,
Here is a comment about tapped horn design by Ian Beaver, an engineer at Danley, that I have never seen anyone post on diyAudio. I think the cone compensation might be the "magic" element he is referring to, as that is the only real difference I see between the majority of subs designed here, and actual Danley tapped horns.
DIY SUBs design suggestion
Ian Beaver: "Obviously I can't speak for all of them, but most (if not close to all) of the DIY designs for a Tapped Horn are not getting one crucial part of the design correct. In many cases "simple theory" gets you in trouble.
There have been a number of Toms designs that did not perform as we wanted-so they are not available. We have pretty much narrowed down what it takes, and we are not telling anybody.
The way Tom designs a Tapped horn is he uses a couple of different prediction programs-rolls it around in his head-mixed with lots of EXPERIENCE of all sorts of tapped horns and comes up with a design.
It is NOT a simple "enter the numbers into the computer and get a result" that most people use."
Here is a comment about tapped horn design by Ian Beaver, an engineer at Danley, that I have never seen anyone post on diyAudio. I think the cone compensation might be the "magic" element he is referring to, as that is the only real difference I see between the majority of subs designed here, and actual Danley tapped horns.
DIY SUBs design suggestion
Ian Beaver: "Obviously I can't speak for all of them, but most (if not close to all) of the DIY designs for a Tapped Horn are not getting one crucial part of the design correct. In many cases "simple theory" gets you in trouble.
There have been a number of Toms designs that did not perform as we wanted-so they are not available. We have pretty much narrowed down what it takes, and we are not telling anybody.
The way Tom designs a Tapped horn is he uses a couple of different prediction programs-rolls it around in his head-mixed with lots of EXPERIENCE of all sorts of tapped horns and comes up with a design.
It is NOT a simple "enter the numbers into the computer and get a result" that most people use."
You can sum up this whole section of this post by simply saying:
Accurate sims match measurements, sloppy sims don't.
Part of an accurate sim is accurately reporting the volume of the throat chamber.
The bold part here is very important, something as small as changing the orientation of the mic or mic placement by as little as a couple of inches, or ambient noise floor changes or a few dozen other things could show this same difference in distortion. Unless you did these measurements in an anechoic chamber without moving anything, these distortion measurements are not infallible, and as such don't mean much.
I have serious doubts about these results. How do you explain the fact that there's a sizable difference in distortion all the way down as low as the measurement goes, down to 30 hz? That little triangle really has no effect at those frequencies. Also, the triangle is symmetrical so it doesn't do anything to resolve the pressure differential across the cone, so pressure isn't an issue here.
No it isn't. None of his designs use "cone correction". Arguably you could say that the TH 115 and 118 (same basic design) does, but I seriously doubt that "cone correction" was his intent, there are a few other very attractive features to the layout he used, I seriously doubt he had "cone correction" on his mind when he laid it out, it just ended up looking similar to the "cone correction" concept.
It may be, but I have a suspicion that he may have found out inserting a "pinch" in the horn at S2 may also produce beneficial results, similar to what I discovered with my "dog food duct", which inserts a pinch between S2 and S3 which can eliminate the first major notch in the FR of a single-expansion TH.
I aim to test something similar in my next "test case". HornResp cannot sim the effect of a pinch at S2 however (unless David modifies it to allow negative Vtc, LOL), so I'll basically be "flying blind" with that test.
I took a lot of steps to ensure the measurement conditions were exactly the same before and after inserting the cone correction, including all measurement levels and the position of both the TH and the microphone. The only thing out of my control was temperature - the first measurement was done just after mid-day, the last one in the afternoon, when it might have been a bit cooler.
I don't. I performed the second measurement a few times, because the decrease in distortion did take my by surprise. The question really is what is contributing to the difference? Maybe the extra stiffness at S1 caused by the "cone correction" triangle being screwed into the panel opposite the speaker?
The thing here is that the decrease in distortion is broadband, which has piqued my curiosity. I've seen differences in my distortion measurements before, but they've usually been over small bandwidths and turned out to be caused by simple things, like standing the TH on an uneven floor for example (the bottom panel seems to contribute a bit - I'll be addressing that with a minor "upgrade" to the design in the future - extra bracing to the bottom panel at its point of maximum flex to increase its stiffness).
The 115/118 is the only one with a "pinch" at S2, Beaver talked about experience and a factor that all the diy'ers got wrong. If this "pinch" was the secret ingredient, by extension, all Danley's other tapped horns are also wrong. This can't be what he was talking about. And the 115/118 don't really have a pinch either, it's just a constant csa through the cone, in other words no throat chamber.
No such thing as negative Vtc, there's either a chamber or there's not. Hornresp can sim a pinch just fine, leave the chamber out and reduce the csa. If you need a more complex shape use Akabak. There's no need to do anything blind. There's always tools.
- Status
- Not open for further replies.