Empirically I have found the same 1dB per octave fall in an in room response is a very good starting point. 0.8 to 1.3 seems to be the range depending on directivity and taste. Variations away from that at certain points do seem to make a real improvement but all very room and diffraction dependent.
Is this behavior the cause of the off-axis response at around 1 kHz being close to the on-axis instead of showing a reduction in response?
That makes a lot of sense, and I should have made that connection. The anti-phase motion of a well damped rubber surround will cause a complex pattern of low level peaks and nulls at the various angles off axis. And it will be different with each driver design.
I hope you're recuperating well @hifijim .
I wonder if I may take a side quest? I was taking some measurements recently, and I noticed what @fluid had noticed.
Here's my farfield (1m) measurements taken recently of a midwoofer with a 10ms window:
There's definitely some bunching up at 500Hz.
I think it's driver +/- cabinet related.
Or is it an observation effect...
TBC...
Last edited:
My hand is back to 100%, thanks! I had to learn a lesson that seems very obvious: When handling a heavy piece of sheet metal with sharp edges, use leather gloves.
I have not posted in a while because my wife and I went on vacation. But now I am ready to get back to my projects.
That is interesting. With a 10 ms gate window, you should be able to get 1/3 octave resolution down to 300 Hz, so your far field results in the 500 Hz region are (very likely) real.
Here are my horizontal polar measurements of the two drivers in the two cabinets. These are 0-90, 15 degree increment.
First MW16TX
MW16TX Normalized
Now SB17CAC35
SB17CAC35 Normalized
I don't see as much "bunching up" at 500 Hz as you have with your system, but I do have some of that effect.
VituixCad does not simulate the effects of cabinet shape beyond the baffle shape, so the differences in cabinet depth/shape/profile are not accounted for. @Patrick Bateman (and others) have shown a measurable difference in polar response based on the size/shape of the cabinet behind the baffle. @mabat and @fluid have shown us simulations that have differences in polar responses based on the size/shape of the cabinet behind the baffle. So what you are seening is probably real, but I don't have the tools to explain what is causing it or to simulate it...
j.
Here is the ABEC model (horizontal response) for the TM I'm building and posted about in my waveguide thread. There are minor tweaks you can make to it, but other constraints can keep you from doing a whole lot. I don't see it as much of a problem. It is interesting that Jim is seeing so much difference between similar cabinets with different drivers, especially at low frequencies like this. Still, I don't see much problem when you get down to the Schroeder frequency when the room will dominate the response here anyway.
I tend to agree. But this is an area where I am careful to keep an open mind. Many people have gone to a great deal of effort to manage the directivity between 100 - 1k. A cardioid mid-bass system is a big effort, and it can increase the DI by 3 to 4 dB in this region. Same with a multi-entry horn, a lot of effort for a 3 - 4 dB increase in directivity in the 100 Hz - 1kHz range. To some people, this relatively small increase in DI is well worth it, so I am very open minded on the subject.
I really think that driver-to-driver differences in directivity can explain a lot. Until a few weeks ago, I basically ignored the off axis response of mid-bass drivers below 1k... I am referring to drivers in the 5 - 7 inch size range. But now I am looking more closely at it.
For example, look at the 100 - 1k normalized response of these four drivers:
ScanSpeak 18w/8531g00
SB17NBAC35-8
Scanspeak 18m4631t00
Purifi PTT6.5X04-NAA-08A
Thank you @HiFiCompass for these plots which were generated under very consistent conditions.
Even in the exact same cabinet, all four drivers would have +/- 2 to 3 dB differences in DI and ER at various frequencies below 1k. It seems to be the surround or the cone edge/surround interface that causes the differences.
j.
Jim, thanks for that post. I have been well aware of those for a long time and wasn’t sure what to make of them since they were taken at only 5 angles- on axis/15/30/45/60. I was a bit reluctant to make any conclusions. Though in retrospect I suppose that’s enough to investigate 5 angles is half of 10 (0-90)
@HiFiCompass do you have any thoughts on @hifijim ‘s post above?
Both @augerpro and I have had our driver/ cabinets measured by @bikinpunk on his near field scanner.
I still have reconcile some… errr… interestin things… but I’d like to hear from Brandon, what his measurements were like compared to the NFS.
Did you have any bunching up in your measurements @augerpro
@HiFiCompass do you have any thoughts on @hifijim ‘s post above?
Both @augerpro and I have had our driver/ cabinets measured by @bikinpunk on his near field scanner.
I still have reconcile some… errr… interestin things… but I’d like to hear from Brandon, what his measurements were like compared to the NFS.
Did you have any bunching up in your measurements @augerpro
I probably should have elaborated in post #130 a bit more... Here is my interpretation of the off axis behavior of the four drivers, and what it probably means to the directivity.
Scanspeak 18W/8531G: Omnidirectional at 100 Hz, with an increase in off axis energy at 300 Hz (negative directivity), followed by some positive directivity at 450 Hz, then negative directivity at 700 Hz, then behavior above 1k as we would expect.
SB17NBAC35: Some small directivity at 100 Hz, Omni at 300 Hz, quite a bit of directivity at 450 Hz, omni at 700 Hz, then increasing directivity above 800 Hz. Directivity from 800 Hz – 2kHz is more than we would expect for this size driver.
ScanSpeak 18M4631T: Omnidirectional until 200 Hz, then negative directivity at 300 Hz, quite positive directivity at 450 Hz, Negative directivity at 700 Hz, and behavior above 1k exactly as we would expect.
Purifi PTT6.5X04-NAA: Omnidirectional to 250 Hz, then slightly negative directivity at 300 Hz. From 400 Hz to 1.2k, the driver is omnidirectional.
No surprise that the Purifi has excellent performance. The best explanation for these driver-to-driver differences is the cone edge/surround interface, and the kind folks at Purifi have put a lot of effort into engineering a good surround.
j.
Scanspeak 18W/8531G: Omnidirectional at 100 Hz, with an increase in off axis energy at 300 Hz (negative directivity), followed by some positive directivity at 450 Hz, then negative directivity at 700 Hz, then behavior above 1k as we would expect.
SB17NBAC35: Some small directivity at 100 Hz, Omni at 300 Hz, quite a bit of directivity at 450 Hz, omni at 700 Hz, then increasing directivity above 800 Hz. Directivity from 800 Hz – 2kHz is more than we would expect for this size driver.
ScanSpeak 18M4631T: Omnidirectional until 200 Hz, then negative directivity at 300 Hz, quite positive directivity at 450 Hz, Negative directivity at 700 Hz, and behavior above 1k exactly as we would expect.
Purifi PTT6.5X04-NAA: Omnidirectional to 250 Hz, then slightly negative directivity at 300 Hz. From 400 Hz to 1.2k, the driver is omnidirectional.
No surprise that the Purifi has excellent performance. The best explanation for these driver-to-driver differences is the cone edge/surround interface, and the kind folks at Purifi have put a lot of effort into engineering a good surround.
j.
I think this is quite significant, and effect can be compounded if left and right drivers are not identical. It is also well into the lateral direction detection range of our hearing. And given it is about 4dB difference in some cases it must be audible. Hmm , will check my accuton c90-6-078 midranges on its behaviour in this region. Never did this type of analysis before.
Reminds me of very distant findings, be it then(~45 years ago) by listening and stroboscopic examinations of cone behaviour. Our conclusion then was to avoid midranges with non-conical cones. Those tested then had all this "flapping" of cone edge part, some even never recovered after first occurrence at a given signal level well within normal range.
Last edited:
Just checked my midrange (Accuton C90-6-078) in sealed box config but different baffle shapes (part of my GAYA2-Final project see f.i. https://www.diyaudio.com/community/...unfinished-after-15-years.391827/post-7288277 ) where my focus was and still is the response between ~700Hz and 5kHz, Measurements in 10' degrees from 0 to 90' off axis at 100cm:
Original Baffle:
With some felt concoctions:
With another change to the baffle:
Regardless of the baffle-shape variations, the range below 1kHz does not show these "ondulations" as @hifijim reported with his selection of drivers.
Original Baffle:
With some felt concoctions:
With another change to the baffle:
Regardless of the baffle-shape variations, the range below 1kHz does not show these "ondulations" as @hifijim reported with his selection of drivers.
Last edited:
Let me contribute. This is a modest RS180S in a 39x21x34cm (hxwxd) enclosure. Measurements 0-90 at 15 deg intervals, Vituixcad interpolations. Gate only was a mediocre 5ms. There seems to be some kind of bump at 400Hz.
Some comment on the measurements of HifiCompass, the freq axis wasn't constant, I got puzzled a bit. But after correcting, all the drivers but the Purifi show a quite common response. Surround? Cone depth?
Some comment on the measurements of HifiCompass, the freq axis wasn't constant, I got puzzled a bit. But after correcting, all the drivers but the Purifi show a quite common response. Surround? Cone depth?
Personally i think it is the physical stiffness drop towards the edge in non-conical cone shapes, in combination with or further worsened by the surround, which relative to the mass of the outer part of the cone is in most cases quite heavy.
There is a good reason for Purifi to add a stiffening to the cone edge, if you think about it it is a huge increase in stiffness relative to the unstiffened cone.
The accuton cones are also very stiff.
There is a good reason for Purifi to add a stiffening to the cone edge, if you think about it it is a huge increase in stiffness relative to the unstiffened cone.
The accuton cones are also very stiff.
The bump frequency goes higher with angle, which means the path length difference of the interfering sounds decreases with increasing observation angle.
Bump at 400Hz means two (or more) sounds interfere constructively at 400Hz, which means their path length difference is multiple of wavelength at 400Hz, which is about 85cm.
Other option is, edge diffraction related back wave which is in opposite polarity to direct sound and would show constructive interference when path length difference is multiple of half wavelength, ~42cm.
Both path lengths are so long indicating the cone shape has nothing to do with it with less than 40cm diameter cone. Its most likely sound diffracting around the box. If baffle is 40cm wide, path length from cone center of it to the further edge and back towards the off-axis we observe at is about 40cm when at 90deg angle. If baffle is only 20cm wide, and perhaps 30cm deep box, then path length from the cone center around the front edge to back edge on the side we observe at is about 40cm and you'd have it. Also, with 20cm wide, 30cm deep cabinet sound all around the box from center of the baffle to center of the back wall would take about 50cm, depending on angle where you observe at its more or less, but this is in the same phase as direct sound and would make destructive interference ~400Hz as its roughly half wavelength difference.
So its probably some error in the measurements due to windowing or something else, or its diffraction that happens behind the baffle. If baffle is large, it could be diffraction on the baffle edge, but in this case there should be more diffraction even lower in frequency due to the backside explained above.
If driver is large, then it could be from the driver. If driver is 15cm in diameter and 5cm deep, there should not be peak of constructive interference below 1wl, below ~2kHz. There is destructive interference to 90deg off-axis when the diameter is half wavelength, and some even below, about 1kHz.
Interference peak at 400Hz makes sense when its from secondary sound source other than the driver, when the driver is small.
Since the bump gets higher frequency with increasing angle it indicates its the back edge. Reduce it by rounding the back of the box. Get it lower in frequency by making the box deeper, get it higher in frequency by making the box shallower.
What comes to hificompass measurements, It looks like edge diffraction of ~2 meter baffle. As the graphs are normalized to on-axis, we get boost at off-axis when there is dip on-axis. On-axis dip around 300Hz is about meter long, so ~0.5 meter path length difference, or 1m if secondary sound making it is in opposite polarity. edit. Gotta confirm this
a second
Bump at 400Hz means two (or more) sounds interfere constructively at 400Hz, which means their path length difference is multiple of wavelength at 400Hz, which is about 85cm.
Other option is, edge diffraction related back wave which is in opposite polarity to direct sound and would show constructive interference when path length difference is multiple of half wavelength, ~42cm.
Both path lengths are so long indicating the cone shape has nothing to do with it with less than 40cm diameter cone. Its most likely sound diffracting around the box. If baffle is 40cm wide, path length from cone center of it to the further edge and back towards the off-axis we observe at is about 40cm when at 90deg angle. If baffle is only 20cm wide, and perhaps 30cm deep box, then path length from the cone center around the front edge to back edge on the side we observe at is about 40cm and you'd have it. Also, with 20cm wide, 30cm deep cabinet sound all around the box from center of the baffle to center of the back wall would take about 50cm, depending on angle where you observe at its more or less, but this is in the same phase as direct sound and would make destructive interference ~400Hz as its roughly half wavelength difference.
So its probably some error in the measurements due to windowing or something else, or its diffraction that happens behind the baffle. If baffle is large, it could be diffraction on the baffle edge, but in this case there should be more diffraction even lower in frequency due to the backside explained above.
If driver is large, then it could be from the driver. If driver is 15cm in diameter and 5cm deep, there should not be peak of constructive interference below 1wl, below ~2kHz. There is destructive interference to 90deg off-axis when the diameter is half wavelength, and some even below, about 1kHz.
Interference peak at 400Hz makes sense when its from secondary sound source other than the driver, when the driver is small.
Since the bump gets higher frequency with increasing angle it indicates its the back edge. Reduce it by rounding the back of the box. Get it lower in frequency by making the box deeper, get it higher in frequency by making the box shallower.
What comes to hificompass measurements, It looks like edge diffraction of ~2 meter baffle. As the graphs are normalized to on-axis, we get boost at off-axis when there is dip on-axis. On-axis dip around 300Hz is about meter long, so ~0.5 meter path length difference, or 1m if secondary sound making it is in opposite polarity. edit. Gotta confirm this
a second
Last edited: