Ok. I've never built a bass horn yet and after experiencing Labsubs and lately the DTS-10's I've gotten the itch to give it a go. The ported 21's are about a week away from being wrapped up and I've finally decided on a horn simulation that I'd like to turn into a pile of sawdust and wood...I think.
I figure I'd better get the ball rolling if I expect to get this built before weather turns bad for the winter.
I've run through a ton of simulations in the last 6 months and there are about 20 possible scenarios that I like and about 8 that it was really tough choosing between for the build. I weeded that down to about 3 possibles by eliminating the ones involving drivers I don't already own. Why buy if you don't need to? I ended up deciding on a 380L net, 30hz FLH using the LMS 5400, over a 380L 30hz B&C FLH and a 30hz 380L B&C TH.
The performance of each is close enough on paper to make it a toss up really.
For the curious here's my thought process. The other contenders were involving dual xxx12's in a TH, TC's PA5100 and the 18sound 21NLW9600. The 18sound 21 and the B&C are VERY similar in performance when using the same enclosure, basically interchangeable IMO with very small differences in modeled performance, so that eliminated the 18sound because I already own the B&C. The PA 5100 also models very well, but differently in the same basic horn parameters as the 21's but obviously needs some changes in the overall form factor due to being a smaller diameter deeper driver. Again since none of them really showed a notable performance advantage on paper and I already own the B&C's...The dual xxx 12 sub I would really like to build (sub bass gargantua), but it's the most complicated, expensive, heaviest and largest. Maybe later. The LMS I decided on over the B&C, because it's the less obvious choice and more of a known quantity at this point. Plus I'm already doing a new build for the B&C's. The build I am doing will be leaving some of the LMS's potential on the table. It'd be better in a larger 20hz FLH or an 18hz TH, but I don't want to build anything so large and difficult to move first off. If I can't easily fit it in the back of my Jeep with the help of 1 other guy it was eliminated this is also why I ended up with a 30hz corner. The LMS doesn't like TH's so small tuned so high.
Anyway. This is what I've got. Nothing is set in stone yet. I need help reviewing the model for mistakes and I could especially use some help with folding the horn up. I've never folded one, so help or comments from more experienced guys would be much appreciated. All opinions are welcome.
BTW the model is intentionally a very basic single expansion rate conical (parabolic now) type to hopefully keep the folding and build as simple and straight forward as possible.
The glitches are phase sourced, and are inevitable with tapped horns. So long as they occur below the passband they're of no consequence, but as you've discovered if you want a wide bandwidth a tapped horn won't give it. The inability of tapped horns to run above 100Hz is why Danley's charts only go to 100Hz. When they first came out he used charts that went to 200Hz, but the ragged response above 100 Hz scared away some, so he changed them in short order.
Yes these have easily handled anything I have thrown at them yet. Right now one is placed under the right side of the pa stack. The other is still unused since it will be 2m gp tested soon. Right now I just have a single channel of a Crest 8002 on the one. Eventually something like a big itech,pl380, or something is in order. I have a 230v line available.
Bentz,
These quick and dirties are with the mic positioned middle of the mouth and coincident with the plane of the side. The cab is rotated tall with the mouth off of the floor and the whole thing is on a furniture dolly rolled outside onto a dock.
Bill,
Yes It is difficult to get useable top end out of a TH. Like other horns it seems that the fuller sized the horn is made the more you can smooth the peaks out. Too bad most bass horns have to be undersized from ideal. I had expected the top end roughness. Actually the fact that I use no smoothing and the graph range is zoomed in quite a bit makes it look a little worse than it is. I had originally thought that the 100hz notch was due to the cab construction and length of the last couple of parallel walls. Now I am leaning towards cancellation from the front and rear radiation of the driver or phase as you say. 22 ft path length and 100hz being about 11ft long roughly seems to have some correspondance. I never really intended these to be used up past 80-90hz after eq but I was trying to get as much extension as possible just because. I didn't do so well in that regard. Now I would like to find out where the real build diverted from the simulation and how to square the 2. Hopefully I can learn some lessons from that.
For what it's worth, if you manage to maintain the assumptions of a model, then you should expect to be able to exactly match the model. If anything, the dips in the model should 'usually' be a lot steeper and more dramatic than in real life. Have you gone back to adjust your model to match what was actually built? Or can you think of any assumptions that may not be valid?
At about 115Hz and 130Hz you can see 2 extra peaks in the response that aren't in the simulation....and i don't think it would be a waste of time to track those down. If nothing else, it could be something learned for next time.
Looking at your build, it appears that there isn't any horizontal expansion down the horn? I'm gonna guess that the dimensions are roughly 2ft, which is about a 1/4 wavelength at the frequencies in question. The general rule of thumb for maintaining a planar wave front is for the wavelengths to be 10x the dimension of the horn, so I'm wondering if you're not seeing the horn break down in the horizontal....in which case, those wavelengths are going to start reflecting back and forth down the horn instead of keeping a coherent wavefront. I've always wondered what would happen in horns if a vertical wall was placed in the middle of the walls for the length of the entire horn...then each channel would be effectively 8x the wavelength which may be close enough to achieve a planar wavefront - I dunno, but it's probably too late to add that now, and it wouldn't be worth the hunch on a guess.
As far as mic position, any chance that the speaker or mic is within ~2ft from a boundary? You might also want to double check that you don't have cabinet resonances at those frequencies - Have you done slow sine sweeps at high SPL looking for buzzes yet?
Yes. I have done some sine sweeps. That is how the measurements were taken. I used a pretty hot drive level for them. Here was one resonance from the cab that was somewhere between 120-180hz. I haven't tracked it down specifically yet just heard it. It was from the access hatch. I may just put a brace in the center of the panel that is forced against the driver motor when the hatch is screwed down.
There is no possibility that boundaries are affecting the measurement really. Once I get some gp measurements from out in the field I suspect they will bear this out. You are right there is no expansion in the width direction. The internal width would be some 22.7" give or take. 100hz is about 11.3 ft or 135.6". That would place the width of the path at about a 5th of a wavelength. It makes sense that there would be an issue related to the width but 100hz seems a little low to me. There is quite a bit of vertical mid wall as you put it in the cab, the braces, but that only covers maybe 30% of the total length.
Going back and tweaking the simulation to match the real result is exactly what I was talking about. I can't get the cab construction related resonances in there but I can get it closer. I think that the biggest discrepancy is probably in the atc and vtc values that I guessed at.
Quote:
Originally Posted by Ricci /forum/post/20520984
Now I am leaning towards cancellation from the front and rear radiation of the driver or phase as you say. 22 ft path length and 100hz being about 11ft long roughly seems to have some correspondance.
That's exactly it, the front and rear waves are going to be 180 degrees out of phase at some point, and that's where you'll get the deepest notch. Part of the design process is to jockey the driver position to place that notch above the passband, and to minimize the less serious peaks and valleys that will occur within it.
There are 4 DSL subs with 100Hz charts, 3 with 150Hz, and 6 with 200Hz charts.
Quote:
Originally Posted by Ricci /forum/post/0
Going back and tweaking the simulation to match the real result is exactly what I was talking about. I can't get the cab construction related resonances in there but I can get it closer. I think that the biggest discrepancy is probably in the atc and vtc values that I guessed at.
I do not believe you will be able to find this there.
It is a compound issue from having so many bends. The "simple" way that was used to get a path length (and horn simulation) loses a little accuracy with every bend. This is compounded with a smaller volume horn per the bandwidth used. Once you get over around a half dozen bends, it doesn't account for the full amount of resistance to volume, and starts drifting away from the sim some with each new bend.
The distance from the horns throat to the rear tap is part of it as well in a TH. If the real acoustic distance is different from the simulation distance, the two will not mesh exactly as the phase is different.
EDIT: In reply to Bill's out of phase suggestion.
Yep. Actually now that I think of it visually that explains it quite nicely. If you consider the radiation from the driver side near the mouth as a single complete 11ft wave cycle in your head and then place another driver 22ft back from the first radiation (the throat side) it will make 2 complete wave cycles until meeting up with the first wave radiation perfectly in sync with it. However the catch here is that the second radiation from 22ft back in the horn throat starts out 180deg opposite phase so when it completes 2 cycles and meets with the radiation from the driver near the mouth it is still reversed phase and creates a huge cancellation at 100hz. Duh.
This is simple stuff but not readily apparent sometimes. See I had been thinking that hey 22ft is almost exactly 2 complete 100hz wavelengths so that should be reinforcing not creating a null. I had forgotten that the rear radiation starts out in completely opposite phase. This leads me to another question...If the above is what is causing the 100hz null should there not also be one at about 50hz?
Soho what do you think about the above. Certainly makes a lot of sense to me intuitively.
Bill,
I know you are a FLH guy but do you have any tips for determining the delay needed for tops to go with a TH? The majority of the output comes from incubating 22ft inside the path but the driver also radiates some from only 2ft inside the cab...Perhaps it should be 3/4's of the full length? Actually everything about a TH alignment screams phase and time domain issues to begin with, but in practice it doesn't seem to be that big of an issue with bass freq's.
This is how TH's work. This is what causes the regular peaks and dips as you go up in frequency. I posted the math to determine the change over point somewhere. This is accounted for in the sims. You can see the in-phase and out of phase crossovers quite easily if you widen the graphs bandwidth.
In your situation the real horns acoustic path is slightly different from the simulations because of the amount bends present in the horn. All it would take is a couple of feet at 100Hz. It may be easier to think of it as friction.
If you really compare the sim and real plots you can see how the phase is off up top. This is what is causing the really deep dip lower in frequency than simed, and the new mini-peak along the way to the real harmonic peak. It is the classic backwards "n" shape. If you did a sweep over a larger bandwith you would see the pattern unfold. You can see the same thing in lilmike's high resolution measurements in all the dip (out-of-phase) sections. They are just magnitudes less noticeable, as the phase of the two sides are better matched, so the volatile transition area is smaller.
Quote:
Originally Posted by Ricci /forum/post/20521709
Bill,
I know you are a FLH guy but do you have any tips for determining the delay needed for tops to go with a TH? The majority of the output comes from incubating 22ft inside the path but the driver also radiates some from only 2ft inside the cab...Perhaps it should be 3/4's of the full length? Actually everything about a TH alignment screams phase and time domain issues to begin with, but in practice it doesn't seem to be that big of an issue with bass freq's.
That's part and parcel of the TH conundrum, you have the front wave and rear wave with different arrival times, so how do you set it? The answer is to the distance between the mains and the exposed cone. The lower you go the less you can hear delay, so there's no need to worry about the rear wave arrival time. The reason it's not a big issue as you note is the period of those low frequencies, which also allows the front and rear waves to combine constructively in the low end despite being sourced so far apart, but as those periods decrease things get hinky. The reason I don't do tapped is that hinkyness (as some users want to run their subs higher than 100 Hz), the higher displacement demands at Fc and the higher distortion of the TH due to the exposed cone. But TH is still a lot better than direct radiating.
I have done some outdoor testing on one of the cabs. I have attached the maximum long term output available using a 10-120hz, 24 second long, ascending sine wave sweep recorded outdoors at 2 meters ground plane. Amplifier is a Powersoft K10 on 240v. There is a good deal of thermal compression in evidence so CEA2010 burst testing should show peak output numbers somewhat higher still. There is no lack of output here. The driver did not grenade, or shred the cone or develop coil rock or any of that. If anything the driver has a much harder time dealing with the power themally. The power compression sweeps are 9 24 second long ascending sweeps back to back with each having the level increased by 5db over the last. This is ended when the sub exhibits clear distress signals. This was immediately followed by 1/48 octave sine burst distortion testing using the STEPS program which takes about 5 min to complete the full measurement. There were 9 of these run back to back at the same input levels as used during the power compression sweep tests. By the end of this the motor was extremely hot and there was the smell of hot voice coil adhesive (Mmmmm.) and some darkening. This was on a visciously hot 95degree day in direct sunlight as well. I ended up running a greatly reduced power 20hz sine wave for a few minutes, which is at the in passband excursion maximum to help cool the motor down afterwards. This is far far more punishment than even the worst real world, long term duty it might ever be asked for.
After testing a DTS10 a couple of times and switching to totally different much beefier drivers and testing that and now testing this tapped horn I have noted that the usual peaks that are seen at low volumes and in simulation get heavily compressed and squashed at the highest playback levels. I am ruling out amplifier clipping due to having plenty of power on hand to either fry or launch the cones at will. It appears that there is either severe compression occuring in the drivers or that there is some sort of acoustic limiting or compression occuring inside the horn itself. i am leaning towards some of the later. Thoughts?
Also after taking this thing a little bit past the limit truthfully I wish that I had beefed the cabinet up a little more and used more bracing. At the very highest sweep levels the cab sounded as if it may blow a seam and developed a buzz in a few narrow frequency ranges near 63hz and about 85-90hz. This isn't evident if you knock the output back to sane levels. I doubt they would be notable with real broadband content either. Outdoor sine waves are pretty unforgiving on exposing rattle and buzz issues. There must be an enormous amount of power being developed inside of this cab. I think I may have underestimated how much.
Now this is a real curiosity. When I first put the cabs together and applied power I took some close mic graphs. These all show a 100hz dip that was not apparent in the simulation. I then ran the pair lightly in a pa set-up for a few days. Fast forward...I get one of them outside to test it and lo and behold the 100hz dip is gone and between 90-110hz the response has changed. The 100hz dip is no longer there. I freak out thinking perhaps I had damaged something in this cab, or the driver, or both, which wouldn't have been out of the realm of possibility considering the beating bestowed on it, so i run inside and take a measure on the other cab and it is now missing the 100hz dip too. That one hasn't been used very hard at all. I thought that perhaps I somehow had EQ engaged during the original measurements but there is no way. The drivers were brand new and had never been above about 2v so perhaps they had a major break in and this is responsible but I find that a hard pill to swallow. The cabs are 13ply BB with pocket screws and are PL'd together and built by a professional cabinet shop who's owner owns DTS-10's and is familiar with speakers so I have a hard time believing that a panel came loose and shifted enough to cause the response change either. I am at a loss to explain it really. I have attached a couple of measurements illustrating this as well. The 100hz notch is easily seen in the original measurement. The other one is the current response where there is no longer a 100hz dip. Very odd.
Quote:
Originally Posted by Ricci /forum/post/20563812
..
It appears that there is either severe compression occuring in the drivers or that there is some sort of acoustic limiting or compression occuring inside the horn itself. i am leaning towards some of the later. Thoughts?
It seems that horns, even bass-horns, are subject to nonlinearities due to the nonlinear nature of air.
This nonlinearity can cause nonlinear distortion (harmonic distortion).
In you case here, it may seem that it can also affect the frequency response of the system, in a non-linear way.
I found some odd behaviour during testing with acoustical damping, and in that case the distortion level was highly influenced by the amount of damping.
Type of damping did not show significant differences.
Since the design was a success, part from some distortion in a narrow frequency band, I did not pursue this any further.
Now, what is actually causing this, I do not know.
Maybe we can find out.
It may also be mechanisms of different origin, causing the same kind of measurable artifacts.
Changes in driver parameters as the power is increased could be worth investigating.
This may be worth some more investigation, seeing as it would be nice to be able to simulate or at least calculate the distortion level earlier in the design phase.
Also, it will be possible to reduce the distortion by making better designs, if one knows what causes it.
Quote:
Originally Posted by Ricci /forum/post/20563812
After testing a DTS10 a couple of times and switching to totally different much beefier drivers and testing that and now testing this tapped horn I have noted that the usual peaks that are seen at low volumes and in simulation get heavily compressed and squashed at the highest playback levels... It appears that there is either severe compression occuring in the drivers or that there is some sort of acoustic limiting or compression occuring inside the horn itself. i am leaning towards some of the later. Thoughts?
Increase in Rdc has no significant effect in simulations.
(Sim I am sitting with now is for a different system, but we can assume that it is the same mechanisms that comes in to effect.)
Same goes for Bl - reduced Bl has no significant effect, and, seeing as this occurs at a relative high frequency, Bl reduction due to excursion should not be an issue.
Impedance curve shows a very narrow peak at the same freq as the peak in the (low level) freq response, indicating a very high q.
Excursion also shows a smaller increase.
This leads me to believe that non-linear damping occurs in the horn, so that the resonance that creates the peak in the freq response is dampened sufficiently at a higher sound pressure level to have a measurable effect.
Other suspects might be mechanical resonances in the walls.
But, what we do know is that when the pressure differences over a small enough distance gets too high, or the air velocity gets too high, then the small-signal model of the propagation of the sound through the horn is not valid any more.
Qe depends on Re and Bl, given that the moving mass does not change.
(A horn is generally not that sensitive to changes in suspension compliance, and thus changes in Qe due to changes in driver fc will not have much effect.)
Changes in Qe will also affect the response lower down in frequency, often more than in the upper range.
Would be interesting to see some distortion figures as well.
The SPL in these measurements are, well, quite high, and many mics will have problems when the spl exceeds 120dB.
Yes the drivers were "virgin" when dropped in. These are some of the most linear and highest power handling drivers around with plenty of test data to back that up, so I do not think it is non linearity on their part causing it. Perhaps break in had something to do with it. Seems like an aweful big change. Could very well be a combo of driver non linearity, air pressure non linearity and cabinet vibration or flex. I am suspecting that at high volume levels there is some sort of point where the loading at the throat is changing or the air pressure itself undergoes some sort of change. If you take the simulation data I provided a few pages back and look at it in Akabak there may be some clues looking at the pressure, velocity, acceleration, etc... Maybe this is the dreaded throat distortion kicked into overdrive? Perhaps we can entice Tom D. to comment since he probably has some good insight. One thing to note is after the response changed it is a lot closer to the simulation now.
I have attached a comparison graph. At least all of the peaks are in the right place. I have already completed the full test work-up on these but it will be a while before I get everything put together and up for consumption. I did have to move the microphone back to 4 meters to get accurate results for CEA-2010 burst testing. The results of that test are...Strong... Shall we say...
Wish I had something to input here but I'm really just ignorant to the whole subject. So I'm just sitting back with some popcorn and lovin' the results.
Sorry, you lost me there. What are you thinking throat compression had to do with the change, or are you talking the difference in the high low voltage testing again?
My question about their chasteness and if you had pulled any numbers on them first was hinting at a break-in scenario.
Drivers are a little tight sometimes from the factory. A normal use heat cycle or two can fix this right up however. It doesn't take days at resonance frequencies.
I'm not saying that this is the only answer either, but it is the simplest one other than a measurement snafu.
As long as the new stuff is repeatable I wouldn't worry about it too much. They were fine before, and are even better now. Actually, they are so good now that you need to stop posting here, and get to uploading the rest of the data. I am really interested is the rest of the info now. Where you able to get the high voltage impedance testing setup done?
Sorry man I was jumbling a couple of different thoughts together. I am happy with the way things turned out broke in. I will leave it at that. Now I am curious about the large changes in the response shape that occur at high power this happens in the dts10 too. My guess is some sort of overload of the air mass in the throat or something.
About the high power impedance tests....I was messing with the VI box and I have been having some irregular results running it into rew. I was trying to determine why the results weren't matching what I thought they should be and ended up frying the high level input. Apparently I am the first to ever do that. I only had it on a single channel of an 8002 running into a nominal 5 ohm load. It is rated at 2500w. I am worried about it being able to handle the tests now. I have another on the way. I really want to see the results for the tapped horns because I feel like it will give us insight into the changes that happen at the limit.
You do a very good job measuring and publishing the results, thank you for that.
As for the compression/high-level freq response - impedance curves at high level may provide useful additional information that can be used to try to determine what causes this, and, more important - the distortion.
Might be that horn distortion due to acoustic overload may also be an issue in bass horns, not only for compression drivers.
Something is causing distortion in the bass horns, and it is not the driver itself, and observations of measurements indicate that it is caused by pressure changes that is large enough in amplitude to change the wave propagation speed significantly - because the distortion is most often low from the 1/4-frequency and up to about 1/2-wavelength, where the distortion often rises rapidly.
Without further objective investigation this is just guesswork, though.
Yes. I think that a high power impedance measurement may be the best shot at determining the cause. I am expecting to see some very large changes in the impedance profile at high power. I really need to get to the same input voltages as used for the last 2 sweeps in my power compression test I just hope the box will handle that much power. I may have to cut the sweep duration down some.
Okv you seem to have a good amount of knowledge and are quite the lurker I see (4 posts in one year). Glad to see I've goaded you into posting.
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Related Threads
?
?
?
?
?
AVS Forum
34M posts
1.5M members
Since 1999
A forum community dedicated to home theater owners and enthusiasts. Come join the discussion about home audio/video, TVs, projectors, screens, receivers, speakers, projects, DIY’s, product reviews, accessories, classifieds, and more!