Hi guys, I thought I would pick your big brains before I go off researching this all day, and not do any of my work again lol.
May someone please tell me how high I can go crossing over a 15" Faital Pro 15400 in a compact H frame minimally constructed to the dimensions of the woofer? The woofer being inset centered back about 6 inches into the H frame. I know my 15400's play nicely in the upper bass frequency's, but I don't know how it would affect off axis response there being inset and all.
These would be actively crossed/EQ'd 3 way and mated MTM OB with Heil AMT 1 and Faital pro 10" 10FE200's. Also I was thinking for the Tweeter my Tymphany DFM-2535R00-08 and Dayton H812 as I love that combo! I don't know what would be better.
Thanks for your thoughts!
May someone please tell me how high I can go crossing over a 15" Faital Pro 15400 in a compact H frame minimally constructed to the dimensions of the woofer? The woofer being inset centered back about 6 inches into the H frame. I know my 15400's play nicely in the upper bass frequency's, but I don't know how it would affect off axis response there being inset and all.
These would be actively crossed/EQ'd 3 way and mated MTM OB with Heil AMT 1 and Faital pro 10" 10FE200's. Also I was thinking for the Tweeter my Tymphany DFM-2535R00-08 and Dayton H812 as I love that combo! I don't know what would be better.
Thanks for your thoughts!
Last edited:
You're right that the H frame will get in the way at some higher frequency. I'd suggest you take a few shots at various angles and plot them together.
You need to calculate or measure the cavity resonance and make sure it's sufficiently suppressed in the crossover.
Electro-acoustic models
Frontiers
Electro-acoustic models
Frontiers
You are correct that an H-frame will have a "tunnel" or cavity resonance on both the front and back sides. This resonance will cause a null in the response at some frequency. That's the bad news.
The good news is that for your particular H-frame, the depth (you mentioned 6" and I assume you mean this is on each "side" so it is a total of 12" front to back) is less than the width (I guesstimate for your 15" driver the width will be on the order of 18"). It turns out that the Q of the dip is related to the aspect ratio of the tunnel formed by the H-frame. The longer the tunnel WRT the width, the higher the Q and the more pronounced will be the dip. Conversely, when the depth << width the Q is low and the dip will be very mild. In fact in your case it might be as little as 3-6 dB, which in the midbass is not a huge issue and you might not even need to EQ it.
The frequency of the dip will be at a wavelength that is equal to four times the depth. Following Linkwitz: resonance frequency = 340m/s / (4*depth in meters) = 340/ (4*6"/40") --> 560 Hz.
The greater the ratio of width to depth in an H-frame the more the H-frame looks like a good ol planar open baffle, which has no resonance at all. The more the depth > width (e.g. Ripole) the more there will be a strong peaking tunnel resonance (and ringing of the time domain response). Choose wisely.
The good news is that for your particular H-frame, the depth (you mentioned 6" and I assume you mean this is on each "side" so it is a total of 12" front to back) is less than the width (I guesstimate for your 15" driver the width will be on the order of 18"). It turns out that the Q of the dip is related to the aspect ratio of the tunnel formed by the H-frame. The longer the tunnel WRT the width, the higher the Q and the more pronounced will be the dip. Conversely, when the depth << width the Q is low and the dip will be very mild. In fact in your case it might be as little as 3-6 dB, which in the midbass is not a huge issue and you might not even need to EQ it.
The frequency of the dip will be at a wavelength that is equal to four times the depth. Following Linkwitz: resonance frequency = 340m/s / (4*depth in meters) = 340/ (4*6"/40") --> 560 Hz.
The greater the ratio of width to depth in an H-frame the more the H-frame looks like a good ol planar open baffle, which has no resonance at all. The more the depth > width (e.g. Ripole) the more there will be a strong peaking tunnel resonance (and ringing of the time domain response). Choose wisely.
Last edited:
You could try flaring the 'wings' out at something like 30 - 45* too and use each 'wing' of different lengths to ….
I realized that I made a mistake in my post, above, about the frequency of the "dip". The frequency that I had calculated is actually the frequency of the first resonant PEAK in the response of an H-frame (due to its self or "tunnel" resonance). The dip happens above that frequency. Here is a link to a pic from SL's web site that shows the typical low frequency response taken at either mouth (front or back) of an H-frame and showing the peaks from the resonance:
https://www.linkwitzlab.com/images/graphics/hframe1.gif
This is a nearfield response. You want to know the far field response, which is the sum of front and rear outputs.
When you add the front and rear output (e.g. when they combine in the room) the response changes as a function of both frequency and off axis angle, and at some combinations of these there are nulls in the response, as shown in this pic:
https://www.linkwitzlab.com/images/graphics/Uframe2f.gif
It's important to note that the frequency at which the peak and dip occur moves up in frequency as you move off axis.
If you try to incorporate the dip into the system EQ (e.g. to flatten it) based only on the ON-AXIS frequency response and then include that region in the passband you will also be boosting the response off axis and the sound may end up a bit boomy in the midrange. I found this to be the case with SL's LX521 as well as some of my own systems that use H-frames for the bass section. To avoid this problem, cross over at or below the peak.
If the H-frame depth is small compared to the width of the mouth, the SPL difference between peak and dip will be small and you don't have to worry about it - you can cross over wherever you want. On the other hand, shallow H-frames do not load the bass much, and you will get a drooping low end due to dipole cancellation.
https://www.linkwitzlab.com/images/graphics/hframe1.gif
This is a nearfield response. You want to know the far field response, which is the sum of front and rear outputs.
When you add the front and rear output (e.g. when they combine in the room) the response changes as a function of both frequency and off axis angle, and at some combinations of these there are nulls in the response, as shown in this pic:
https://www.linkwitzlab.com/images/graphics/Uframe2f.gif
It's important to note that the frequency at which the peak and dip occur moves up in frequency as you move off axis.
If you try to incorporate the dip into the system EQ (e.g. to flatten it) based only on the ON-AXIS frequency response and then include that region in the passband you will also be boosting the response off axis and the sound may end up a bit boomy in the midrange. I found this to be the case with SL's LX521 as well as some of my own systems that use H-frames for the bass section. To avoid this problem, cross over at or below the peak.
If the H-frame depth is small compared to the width of the mouth, the SPL difference between peak and dip will be small and you don't have to worry about it - you can cross over wherever you want. On the other hand, shallow H-frames do not load the bass much, and you will get a drooping low end due to dipole cancellation.
Last edited:
Thanks for all the great information! The only reason I wanted to try a shallow H frame was to lower the F3 compared to the flat baffle I am using now playing up to 500hz crossed to a 1.5" CD. Then again maybe that's not a good idea.
You see, I switched from a pair of Alpha 15A 15" per side to the single 15400 per side. I learned that the system F3 is significantly higher for the 15400 on the same baffle size. So I want to try to lower the F3 somehow physically because of my low watt setup.
I just wondered what the consequences are of running a shallow H frame does to the frequency's approaching 500hz on and off axis before I build it. Like you said I can try a wider H frame or angled sides(great ideas thanks). Then again I probably just need a bigger flat baffle but I hate to do that lol. That's the beauty of the challenge of speaker building i guess, trying to find the perfect compromise 😀
You see, I switched from a pair of Alpha 15A 15" per side to the single 15400 per side. I learned that the system F3 is significantly higher for the 15400 on the same baffle size. So I want to try to lower the F3 somehow physically because of my low watt setup.
I just wondered what the consequences are of running a shallow H frame does to the frequency's approaching 500hz on and off axis before I build it. Like you said I can try a wider H frame or angled sides(great ideas thanks). Then again I probably just need a bigger flat baffle but I hate to do that lol. That's the beauty of the challenge of speaker building i guess, trying to find the perfect compromise 😀
Last edited:
Matt, congratulations, you are learning first hand about some of the tradeoffs of dipole system design.
You likely had a lower F3 with the Alpha 15s because their Qts is might higher than the Faital driver. Remember, the response at low frequencies is the sum of the losses from the front-to-back acoustic short circuit AND the driver's own response curve. A open baffle is like "free air" so the MFG Qts value from the datasheet is a relatively good predictor of the driver response. The dipole losses are then a function of the length of the front-to-back path.
In my own experiences with dipole systems, I have found that there is a point around 100-150Hz, below which it becomes increasingly difficult to impossible to mitigate the total SPL loss for a planar OB or short H-frame. As a result, I now design my systems as slim planar or nude dipoles down to about 150 Hz and then I use a large H-frame for dipole bass. This can be a single, central H-frame (makes a great stand for amps, etc.) or you can have stereo H-frames outside of the main speakers and toed in slightly.
To get the mains to play down to 150 Hz you might use a nude 12" or 15" pro driver, or even a pair of 8" 8 Ohm home audio drivers in parallel in a 12-15" wide baffle. Both of these approaches help to provide relatively high voltage sensitivity that helps to offset the dipole losses. I aim for at least 96dB@2.83V sensitivity for this band.
By using a long, wide H-frame and a higher Qts driver the losses at the lowest frequencies are minimized. As a result, the voltage sensitivity of the sub driver can be a more reasonable 87dB. You are therefore able to use a more traditional "subwoofer" driver. Look for one designed for "infinite baffle" or "free air" applications for which Qts is 0.7 or even higher if possible. You will still need high Xmax, since it is a subwoofer. There really are not too many drivers that meet these goals but they are out there.
Of course the other option is to throw in the dipole towel at 150Hz and transition to a sealed sub (or two). This is IMHO a better option unless the listening space is large enough to support low frequency dipole bass (meaning a large to very large residential space).
You likely had a lower F3 with the Alpha 15s because their Qts is might higher than the Faital driver. Remember, the response at low frequencies is the sum of the losses from the front-to-back acoustic short circuit AND the driver's own response curve. A open baffle is like "free air" so the MFG Qts value from the datasheet is a relatively good predictor of the driver response. The dipole losses are then a function of the length of the front-to-back path.
In my own experiences with dipole systems, I have found that there is a point around 100-150Hz, below which it becomes increasingly difficult to impossible to mitigate the total SPL loss for a planar OB or short H-frame. As a result, I now design my systems as slim planar or nude dipoles down to about 150 Hz and then I use a large H-frame for dipole bass. This can be a single, central H-frame (makes a great stand for amps, etc.) or you can have stereo H-frames outside of the main speakers and toed in slightly.
To get the mains to play down to 150 Hz you might use a nude 12" or 15" pro driver, or even a pair of 8" 8 Ohm home audio drivers in parallel in a 12-15" wide baffle. Both of these approaches help to provide relatively high voltage sensitivity that helps to offset the dipole losses. I aim for at least 96dB@2.83V sensitivity for this band.
By using a long, wide H-frame and a higher Qts driver the losses at the lowest frequencies are minimized. As a result, the voltage sensitivity of the sub driver can be a more reasonable 87dB. You are therefore able to use a more traditional "subwoofer" driver. Look for one designed for "infinite baffle" or "free air" applications for which Qts is 0.7 or even higher if possible. You will still need high Xmax, since it is a subwoofer. There really are not too many drivers that meet these goals but they are out there.
Of course the other option is to throw in the dipole towel at 150Hz and transition to a sealed sub (or two). This is IMHO a better option unless the listening space is large enough to support low frequency dipole bass (meaning a large to very large residential space).
If you want to increase the path length for a given baffle width but don't want to mess up the highs, consider a toroidal baffle.
I guess he meant to mount the driver in the center of a thingie that looks like an inner tube of a tyre.
BTW: This is a version of a flared H-Frame, shaped like a real LF horn:
Cabinets | Wolf von Langa
But I guess it would be too large.
Rergards
Charles
BTW: This is a version of a flared H-Frame, shaped like a real LF horn:
Cabinets | Wolf von Langa
But I guess it would be too large.
Rergards
Charles
Judging by Charles' link, the toroid would be the spherical version of this, or maybe an end treatment. Earl Geddes once recommended this to Siegfried Linkwitz.
I'm not sure how you can make this claim (regarding the HF performance):
There does not seem to be any measurement data posted, or any mention (in German, but you can translate it) about the response... Did I miss it, or is it posted somewhere else?
Great info Charlie thanks! I still haven't learned how parameters (besides high QTS Which is more favorable for OB) effect response on OB.
I wanted to ask you. When you put 2 of the same woofer on a single OB, does the f3 then roll off higher like it would if you were to put 2 of the same woofer sharing the same space in a closed system. Meaning does the same physics apply in a sense? Or do 2 open baffle woofers have the same low frequency roll off as a single driver would on the same baffle? Sorry that sounds confusing.
I want to try this approach using naked drivers you are speaking of with the 10" 🙂 Do you think Qts matters for mid range drivers above say 200hz? Or maybe when does Qts start becoming a major factor? Sorry for all these subjective questions, but I can't find these answers anywhere. I love this forum. You guys are a huge wealth of knowledge! I appreciate you, I don't post that enough reading this forum over the years.
I wanted to ask you. When you put 2 of the same woofer on a single OB, does the f3 then roll off higher like it would if you were to put 2 of the same woofer sharing the same space in a closed system. Meaning does the same physics apply in a sense? Or do 2 open baffle woofers have the same low frequency roll off as a single driver would on the same baffle? Sorry that sounds confusing.
I want to try this approach using naked drivers you are speaking of with the 10" 🙂 Do you think Qts matters for mid range drivers above say 200hz? Or maybe when does Qts start becoming a major factor? Sorry for all these subjective questions, but I can't find these answers anywhere. I love this forum. You guys are a huge wealth of knowledge! I appreciate you, I don't post that enough reading this forum over the years.
Last edited:
'Claim' is a strong word, maybe I haven't chosen the right words on this occasion.
Qts is only a parameterized description of the response in the vicinity of the resonance frequency. It says nothing about what happens at higher frequencies...
The loading of a driver in a planar OB is similar to putting the driver in an infinitely large box (and I mean infinitely large) AKA "free air". This does not depend on how many drivers are on the baffle, each one sees the same loading. There are minor differences in the low end response depending on WHERE on the baffle a driver is located, but I don't think that is what you were asking about.
Qts is only relevant to the response around Fs. So for example if the driver Fs is 50Hz the response at 200Hz and above is not influenced by it. This is why you can use many different woofers (including low Qts ones) down to about 150-200Hz and only when you want that passband to extend below 150Hz do you need to start to take into account Qts, front to back pathlength, etc.
As a result, I use a driver in a very narrow minimal baffle (just wide enough to mount the driver and not break) or nude (no baffle) starting around 150Hz. The smaller the baffle, the better (more CD like) the off-axis radiation up to and through the dipole peak. Linkwitz mentions how baffle diameter vs driver diameter in OB systems influences the off axis response here:
Electro-acoustic models
Look at the figures that accompany the text A3-5, slightly down the page.
Using a driver nude is the b=a case. You can see that the responses for on and off-axis angles all follow each other through the dipole peak. As the baffle radius (b) grows larger than the driver radius (a) this pattern breaks down. I have confirmed this behavior with measurements on my own prototypes. The downside is that for b=a the dipole peak is highest in frequency, or another way to look at it is that the low frequency losses from dipole cancellation is greatest for b=a. This can amount to 12dB or more of loss at 150Hz, even with a 10" driver. This makes the effective sensitivity of the driver low, meaning you will need a lot of power to raise up the level if you want to use a nude driver over a wide bandwidth. This is why good OB systems often are 4-way: this keeps the bandwidth of each driver low and it can be operated around the dipole peak where on axis SPL is high (up to +6dB compared to the same driver in a closed box). This means you can have a slim or nude OB system that uses low power amps, but you will need 4+ bands.
To give you an idea of using a driver "nude", I have attached measurements I took on a pro audio 10" driver. I didn't end up using this particular driver, but was evaluating it for potential as a nude OB driver. The measurements shown were taken from the front side on axis and every 11 degrees until the final one at 90 degrees off axis. Note that you don't see the classic "dipole peak" in this data because with a nude driver (which is definitely not like a point source) you don't get that kind of response pattern. This was shown by the sims in the page on SL's web site that I linked to above. Here we have real data to back it up.
The pattern narrows (lines get farther apart) starting around 800-1k Hz, and sort of hangs together until the breakup peaks starting around 2kHz, followed by the ultimate rolloff above breakup.
It would really be a stretch to use this driver nude down to 150Hz because the losses there are pretty high and that translates into a very low effective sensitivity around that frequency. It's around 20dB down at that point, which is too much to compensate for by increasing the power input WRT the 1kHz power level.
The solution would be to use another larger driver (like a 15-18" nude or a 12-15" in a medium width baffle) and cross over around 500Hz or so. When I have played around with these kind of design questions I have found that a system with a 15" nude or in a narrow baffle from 150Hz-700Hz, then a 6" nude from 700Hz to 2kHz, and then a planar tweeter (e.g. Neo3PDR) is a good approach. Then I have to fill in below 150Hz with a sub like an H-frame or CB sub depending on the room.
The pattern narrows (lines get farther apart) starting around 800-1k Hz, and sort of hangs together until the breakup peaks starting around 2kHz, followed by the ultimate rolloff above breakup.
It would really be a stretch to use this driver nude down to 150Hz because the losses there are pretty high and that translates into a very low effective sensitivity around that frequency. It's around 20dB down at that point, which is too much to compensate for by increasing the power input WRT the 1kHz power level.
The solution would be to use another larger driver (like a 15-18" nude or a 12-15" in a medium width baffle) and cross over around 500Hz or so. When I have played around with these kind of design questions I have found that a system with a 15" nude or in a narrow baffle from 150Hz-700Hz, then a 6" nude from 700Hz to 2kHz, and then a planar tweeter (e.g. Neo3PDR) is a good approach. Then I have to fill in below 150Hz with a sub like an H-frame or CB sub depending on the room.
Attachments
Wow thanks for breaking it all down. I didn't know SL posted these findings.
I haven't been on his site in years, fascinating. This gives me much to consider.
I guess I will have to rethink the 10" if it's going to perform that way. Doubling them up will not fix the issue either I guess. Back to the drawing board!
I haven't been on his site in years, fascinating. This gives me much to consider.
I guess I will have to rethink the 10" if it's going to perform that way. Doubling them up will not fix the issue either I guess. Back to the drawing board!
Regarding the cavity resonance. Linkwitz data wasn't intended to show higher frequency performance in an H frame like the original question asks. What it shows is only the effect of the cavity resonance which could be calculated as high as you like, assuming the driver fits the cavity as a source which it wont. At higher frequencies there will be higher order modes/reflections due to the shape of the frame.