You‘ve got multiple 1/4 wave resonances all in very close succession from 120 to 135hz and it all works for you. Each resonance would be the 1/4 wavelength of the depth of the slot along multiple vectors…………you didn’t do a round slot to match the profile of the woofer so the diagonal vectors from the corners are a little longer……around 27” or so?…….and the primary vector around 22”?
After that, it’s a steep acoustic low pass by nature. You could put some dense foam in the back of the slot’s corners carved to follow the profile of the drivers…..that will smooth out the primary resonance peak and likely a few more that are the multiple harmonics……it might help to smooth out the 200-400hz range here and there. Would take around 10 minutes to do and test….well worth the input
Yes….,you accomplished your goal and achieved nearly the full vocal range through one point source……and you enjoy the presentation so that’s all that really matters, right? For the rest of us, who haven’t experienced what you’ve done…..well….we remain skeptical to accept your subjective observations based on what we see in the data……and that’s not a fair assessment of what you’ve created…….it’s sighted listening without the listening part! Lol.
At this point, best to stop beating your head against a wall……i don’t think you’re going to convince this pack of objectivists of anything. Sadly EVEN IF you managed to get a few of these folks to actually listen to this system, having seen the data their observations would be tainted and they’d likely point out all the flaws their visual intake would convert to aural.
If this would have been my goal…..a high power full bandwidth 2way, I likely would have built a line source with 8” woofers crossed to AMT’s capable of a 1khz crossover instead. Multiple AMTs wouldn’t struggle with distortion on the low end of their passband and ten 8” woofers could get me effective LF in room response down to 20hz with some mild DSP shaping. It would have certainly been an expensive project though……likely $3-$4k in drivers alone. Only a fairly large room could effectively house such a performance though……and I’m not a big in room headphone kinda guy when I’m enjoying music instead of working with it…….I like to hear the room as part of the overall presentation. Since I don’t have a room that large, I don’t see myself building something like this any time soon.
After that, it’s a steep acoustic low pass by nature. You could put some dense foam in the back of the slot’s corners carved to follow the profile of the drivers…..that will smooth out the primary resonance peak and likely a few more that are the multiple harmonics……it might help to smooth out the 200-400hz range here and there. Would take around 10 minutes to do and test….well worth the input
Yes….,you accomplished your goal and achieved nearly the full vocal range through one point source……and you enjoy the presentation so that’s all that really matters, right? For the rest of us, who haven’t experienced what you’ve done…..well….we remain skeptical to accept your subjective observations based on what we see in the data……and that’s not a fair assessment of what you’ve created…….it’s sighted listening without the listening part! Lol.
At this point, best to stop beating your head against a wall……i don’t think you’re going to convince this pack of objectivists of anything. Sadly EVEN IF you managed to get a few of these folks to actually listen to this system, having seen the data their observations would be tainted and they’d likely point out all the flaws their visual intake would convert to aural.
If this would have been my goal…..a high power full bandwidth 2way, I likely would have built a line source with 8” woofers crossed to AMT’s capable of a 1khz crossover instead. Multiple AMTs wouldn’t struggle with distortion on the low end of their passband and ten 8” woofers could get me effective LF in room response down to 20hz with some mild DSP shaping. It would have certainly been an expensive project though……likely $3-$4k in drivers alone. Only a fairly large room could effectively house such a performance though……and I’m not a big in room headphone kinda guy when I’m enjoying music instead of working with it…….I like to hear the room as part of the overall presentation. Since I don’t have a room that large, I don’t see myself building something like this any time soon.
I've found much the same. Massively multiway systems alleviate some filter and performance constraints at the cost of complexity.
Totally fair. Also better aligns to a useful output for more people. I couldn't give my very large system to very many people due to the size requirements alone.
Yeppa.
Much of the work or trying to make wise compromises with bandwidth spans, xover points, directivity matching....get's much easier with a higher multi-way count ime.
But as you say......at the cost of complexity.
It's funny though, because in today's world, if one can handle the complexity of increased multi-way count...... and tolerate latency....
crossovers become the easiest, least consequential element in speaker design.
(I'm of course referring to complementary linear phase crossovers via FIR)
I say it's funny, because xovers are clearly one of the most agonized over elements of speaker design,..... and apparently complex to most folks.
But when even high count muti-way xovers are not complex xover wise... which complexity is worse? lol
Very nice airvoid! It's so damn easy isn't it?
A full 5-way like this takes me but an hour or two....and it crosses to a sub too,....... which only takes another 15 min .
Kerf bending has taken my interest. I could cup the OD of the woofer frame from behind, at least a half moon, and then flare it out to the height of the slot.
The group delay here, After EQing, to neutral would change. The decay length of a pipe that the walls do not resonate, is negated by path length. The largest resonance is shows a peak of 4ms at ~147hz.... That isn't a very large resonance. Cycle time of 147hz is ~ 6.8ms.
OK there's some information I must disclose, sorta embarrassing, but whatever
.... I knew about this, just haven't dealt with it. Thats the threaded rod singing ~172hz.... I am working on it! I've got plans for all the things!I think really this has been a lesson for to learn how to interpret measurements. I am still learning.
What if I convince myself to take the woofers out of the slot, I'd have enough woofers to build 2 more subs for the rear. I am not sure about doing 2 18"s side by side on one baffle. I'd have to take a look at Vituixcad again, but by 200hz I think directivity starts to set in.
The boxes are still undamped inside, theres some tailoring of the slot to be done.... Theres still the concept that the room is going to swamp these small issues. The rooms decay will be much longer than the issues seen here.... the same concepts thought of towards well built BR's and TL's
Last edited:
I was thinking that the room resonates below a certain frequency which in most homes, includes 200hz and lower? The room is generally resonating more than a subwoofer, unless the subwoofer has some super gross decay issues. The air in the room... not the walls and windows.
The acoustics in my basement aren't as bad as up stairs
1" measurement
At my ear at the listening position
I have the 48db high pass on these measurements but its definitely not near a finished crossover. If the blue areas where anywhere near detrimental in the first shot, they are buried in room decay at the listening position.
The acoustics in my basement aren't as bad as up stairs
1" measurement
At my ear at the listening position
I have the 48db high pass on these measurements but its definitely not near a finished crossover. If the blue areas where anywhere near detrimental in the first shot, they are buried in room decay at the listening position.
"Concern yourself with decay to -20 dB; beyond is audibly less critical."
"-15 dB to -20 dB is where perception is notably affected."
"Focus on decay to -30 dB for thoroughness, but -20 dB is typically sufficient for perceptual transparency."
Comments from my 3 AI consultants... If I tell you who I've prompted the AI to emulate, you'd be tickled.
"-15 dB to -20 dB is where perception is notably affected."
"Focus on decay to -30 dB for thoroughness, but -20 dB is typically sufficient for perceptual transparency."
Comments from my 3 AI consultants... If I tell you who I've prompted the AI to emulate, you'd be tickled.
I think that it's naive to think that flattening the axial response is an optimal solution to the problem. I used the whole polar field when I did crossover optimization and flat axial response was seldom optimal.
Strongly agree. There's alway been a bit of compromise in finding the best balance between direct on-ax response and a good set of polars, in anything I've done at least.
That said, once I've found that good compromise for a particular DIY build, there is usually a certain off-axis reference angle, both in terms of H and V,
that I then feel confident using as a surrogate for a reference axis, for further tuning experiments.
In the case of the traces I posted in #13,409, that was about 12 deg H, and 5 deg V. It's polars will look pretty good too.
Maybe Airvoid is doing something similar, dunno do we.
Hi all, i have reviewed this statement closely.Your Axi2050 is not built to use a 200Hz crossover. Try using some the manufacturer's and professional's advice. The Jubilee uses it to 400Hz. I've heard them online and they sound like a giant set of headphones. 400 Hz is a very reasonable limit. Obviously you can't currently experiment with a 400Hz crossover because your woofer cabs won't allow it.
Build a test cab with some of that cheaper patched up plywood.
A c-c spacing of 34", crossover at 400Hz, with a front firing TD18, might just be the ticket.
One of the few tests done from some german HiFi online magazine
https://www.lowbeats.de/klipsch-jubilee-zur-privat-audienz-auf-der-high-end-2023/
do not mention anything about this giant headphones listening effect
but i think that the JBL specs already states that the minimal recommended listening distance is greater 4 meters
btw cutoff frequency of the Klipsch Jubilee is 340 Hz
and if you check the dimensions of the speakers itself one can figure out that for the huge cabinets a very large room is required anyway
so my questions for the setup of @camplo are
A )
listening space
B )
listening distance
C )
average listening levels as of SPL at the listening distance
- therefore an ordinary SPL meter for the headphone is good enough
so far - so good, Stefano
P.S.:
here the expected behaviour of the big Jubilee Horn simulated with the Celestion Horn Wizard
Last edited:
I don’t know what made you jump to that conclusion, measurements for this particular speaker were taken 0-90 degrees at 10 degree increments in the Seas anechoic chamber in Moss, Norway https://www.seas.no/index.php?option=com_content&view=article&id=405&Itemid=269.
They use a Klippel system with B&K mikes, the pic I posted just happened to be showing the filters and response at 20 degrees. This was just to illustrate how modern FIR filters, such as this particular automated system when fed with noise free and accurate measurements, do the job of flattening response in literally seconds.
Of course there is a bit more to it than that; looking at just one axis is like looking at the situation through a straw. But again, phase and magnitude can be manipulated very quickly, below an example of a two way (12” coax with a 1”cd), about 2400 taps used, latency set at 20 msec, all calculated in less than a second. Like you and Mark pointed out it’s only the beginning but at least this part of the job, again given good measurement data, is becoming trivial.
Last edited:
With bendy plywood, no kerfs would be required to bend to the radius of an 18" driver.
After mentioning that ringing is one (of the many..) problems with threaded rod between drivers years ago, glad you that wasn't just BSOK there's some information I must disclose, sorta embarrassing, but whatever
There are lots of recent posts regarding the use of the Axi2050 down to Fc, and using the Klipsch Jubilee K-402 conical/tractrix expansion horn crossover of ~340Hz as an example of why you shouldn't.
Comparing the K-402 to your (mostly) undersized mouth exponential horn is like comparing a pancake to a pineapple, but lets do it:
As can be seen in the green trace, your horn has a sharp drop of >24dB/octave below it's Fc (cutoff frequency) of ~160Hz, while the K-402/Axi2050 (red trace) drops off at about -12dB per octave below 500Hz, similar rate to a cone driver in a sealed box.
The TAD TD 4002 (blue trace) lighter diaphragm has a higher Fs than the Axi2050, so drops off at a steeper rate below 500Hz.
The -8dB dip ~200Hz in the green trace is evidence of its undersized horn mouth relative to the low Fc.
These "features" would be enough for most to heed the usual advice for exponential horns to "cross over ~2x Fc", ~320 Hz.
That said, you have proven your huge, beamy, oval exponential(ish) horn has plenty of output for you at your one meter (!?!) listening distance at Fc.
With "brick wall" filters and enough EQ, it can be crossed to your slot loaded bandpass 18s" that low and still exhibit flat on axis response.
But as Airvoid says: "looking at just one axis is like looking at the situation through a straw"
At any rate, no amount of filtering or EQ will make a HF compression driver sound "clean" (compared to a cone driver capable of smooth response) when it's forced to cover such a wide range.
I wish I could back up that subjective statement (shared by most listeners and designers) with some extensive measured evidence.
The best I have is this simple two-tone test using 523 +932Hz), using the lowest distortion HF compression driver I'd tested, the EV DH1A (yellow trace) compared to a 3.5" "full range" cone driver (TC9FD, blue trace) on the same horn (250Hz Fc):
In retrospect, the drivers should have been equalized to the same response prior to the tests.
The 532Hz level is nearly identical for both drivers, the DH1A is louder at 932Hz.
If the drivers were perfect, those two frequencies would be the only ones seen.
The test shows both IMD (intermodulation sum and difference distortion) and HD (harmonic distortion) products.
Easy to see how much more IM the compression driver has generated.
It was also very easy to hear a large difference when playing music when equalized to similar response.
Subjectively, even order HD is fairly benign, odd order less so, but I think IMD may be what is behind the negative sound quality difference we have always been able to easily hear when comparing cone drivers to compression drivers going "too low".
Or it's something else we haven't yet been able to measure
Art