That is a great thread,Although, I have not read it through all of the way yet.
I like to enjoy my music in the medium to high range.
So, My design goal is normaly 120db considering the distance from my stack of speakers there is a calculated 10db of loss.
Anything less than this such as the 85db to 95db range I have smaller speakers that I use nearfield for just listening to as to not rock out the house. he,he
Here are the charts that I have been using to try and determine the width of my next drivers.
I am sure that some of you have seen them before and there are several papers with these same graphs.
I don't have the links but they can be found at the JBL site if you google up "speaker directivity".
jer
I like to enjoy my music in the medium to high range.
So, My design goal is normaly 120db considering the distance from my stack of speakers there is a calculated 10db of loss.
Anything less than this such as the 85db to 95db range I have smaller speakers that I use nearfield for just listening to as to not rock out the house. he,he
Here are the charts that I have been using to try and determine the width of my next drivers.
I am sure that some of you have seen them before and there are several papers with these same graphs.
I don't have the links but they can be found at the JBL site if you google up "speaker directivity".
jer
Attachments
I'd like a good 6" or 8" midrange to go with 15TBX40's - they have very little "hit" in a small reflex - maybe a good mid would wake them up -? ~30 years ago I had Cerwin-Vega S1 bookshelf with 6th order assisted order reflex, a 12" , large lappede-cone mid and a "dhorm" - something like that but better blended/less rough could be ok just to make use of those woofers
Last edited:
A couple of examples, one I linked to before:
http://www.tb-speaker.com/detail/1230_04/w8-1808.htm (93dB/2.83v/m)
http://www.tb-speaker.com/detail/1230_04/w8-1772.htm (95dB/2.83v/m)
My vintage Coral Flat 8 II are about 94dB/2.83v/m, 93-95dB is pretty typical of 8" paper cone full range drivers, although more conventional 8" woofers or midbass drivers will tend to be quite a lot less sensitive (heavy cones) as they're designed for useful bass response in smaller enclosures.
High sensitivity and large cone area usually means a large Vas (typically ~100 litres in the above examples) which means a pretty large cabinet for good bass reproduction, (50-60 litres) but if you're using them as a midrange driver you can get away with an enclosure in the 5-10 litre range if it's stuffed well and/or you put a resonance compensator in the network for the fundamental resonance peak caused by the small enclosure. (My Flat 8's free air Fs is 38Hz and rises to ~125Hz in a 10 litre closed midrange cavity, still an octave below where I would cross it over as a midrange driver)
As for "after proper baffle step correction", I'm not sure what you mean by that. Quoted sensitivity for drivers is usually specified for half space, not free space.
So this is the sensitivity you'll actually measure on an infinite baffle, or significantly above the baffle step frequency on a finite baffle. Much lower than the baffle step frequency the response (assuming the drivers near-field response is still flat) will be -6dB.
So a 93dB/w/m specified driver would only actually produce 87dB/w/m well below its baffle step frequency, but likewise a 4" 86dB/w/m driver would only produce 80dB/w/m SPL well below its baffle step frequency. The difference between the drivers is still 7dB.
With a 4" driver on a very narrow cabinet the baffle step frequency will be well up into the midrange, say 600-800 Hz, so at the low end of the midrange (where the high midrange excursion and SPL demands are in music) you will lose up to 6dB SPL and 6dB potential headroom. The driver has to work that much harder, and also requires in-band baffle step correction.
On the other hand a larger driver of necessity requires a wider baffle which makes the baffle step frequency lower. A 39cm wide baffle (which would not be out of place on a large 3 way) puts the baffle step frequency at about 300Hz, so the midrange driver maintains its quoted half space sensitivity over nearly all of the midrange. Compared to the narrow baffle that's an extra 6dB of SPL and headroom in the bottom octave of the midrange essentially for free, along with not needing explicit baffle step correction in the midrange network.
So with a larger driver on a moderately wide baffle you can maintain your half space sensitivity over the full operating range of the midrange driver...
Last edited:
Well, I don't like to place speakers close to walls, so no, I find baffle step correction is still needed overall.
However if your woofer/midrange crossover frequency is equal to your cabinets baffle step frequency you can achieve it largely through attenuation of the midrange and tweeter (relative to woofer) with L-Pads, rather than needing to apply some sort of shelving filter within the midrange drivers passband.
In a 3 way the amount of attenuation needed on the mid and tweeter also depends on the proximity of the woofer to the floor. If the woofer is quite close to the floor it will receive more floor gain, so the total correction needed will be a little bit less than 6dB.
Semi OT, but what is wrong with close to wall placement? Phase cancellation front front wall is minimized if speakers can be placed near wall.
10 inch mid-range
I never would have believed it, had I not tried it for myself. After about a 2 year search for a driver to cover 250- 5,000 Hz ( that I could afford ) I took a wild chance on a guitar speaker. I chose the Eminence Legend 1058. I think it's about $60.00 Granted, right out of the box, not so good, but I equalized it to the reverse curve frequency response. Amazingly enough, other than a
JBL 123 12" mid I had on loan, it's the only driver I have used that can match up to the Stage Accompany Ribbon used above 5K. Before this i used a pair of 5" pro drivers that just can't move as much air/ or sound as natural of a match to the rest of the system. This Legend 1058 is rated from 100- 6K.
It's also the only driver that can push enough air at the lower x-over point of 250, to sound like a match for the 15" mid-bass below it. Just my 2 cents.
Have a nice day.
I never would have believed it, had I not tried it for myself. After about a 2 year search for a driver to cover 250- 5,000 Hz ( that I could afford ) I took a wild chance on a guitar speaker. I chose the Eminence Legend 1058. I think it's about $60.00 Granted, right out of the box, not so good, but I equalized it to the reverse curve frequency response. Amazingly enough, other than a
JBL 123 12" mid I had on loan, it's the only driver I have used that can match up to the Stage Accompany Ribbon used above 5K. Before this i used a pair of 5" pro drivers that just can't move as much air/ or sound as natural of a match to the rest of the system. This Legend 1058 is rated from 100- 6K.
It's also the only driver that can push enough air at the lower x-over point of 250, to sound like a match for the 15" mid-bass below it. Just my 2 cents.
Have a nice day.
Attachments
Yeah, it is somewhat off topic, and it's not a short and simple discussion that I could do justice to in a post or two.
But really quickly, the problem is that placing the speakers really close to the front wall does not minimize phase cancellation, it just moves it higher in frequency to frequencies (lower midrange) where it's more intrusive and less easily solved.
Unless your speaker has a really wide baffle and a low baffle step frequency the notches you then get in the low midrange from the front wall are actually deeper than the notches you'd get in the upper bass if they were a bit further away as the path length differential is much less. (Baffle directivity will tend to counteract this somewhat with a wide baffle, but a narrow baffle will see severe lower midrange notches)
If you bring the speakers out far enough (around a metre or so from front wall to front baffle) this pushes the notches down into the upper bass region below 200Hz, at which point you can then smooth them out with a multi-sub approach. (For example subs closer to the corners than the mains)
As we perceive essentially the steady-state response below about 200Hz this is a valid way to solve the boundary cancellation notches at these low frequencies without harming the midrange.
If you push the speaker close to the wall and raise the cancellation frequencies into the midrange you can't smooth them using multi-subs nor can you EQ them, (since they're caused by comb filtering) so you're stuck with them.
Keeping a certain minimum distance from the nearest walls is also important for imaging, by ensuring a certain minimum time to first reflection, and if the speakers are designed for free-standing use, keeping the right tonal balance.
In my experience, deliberate lack of baffle step correction in the speaker design seldom if ever neatly matches room gain from placing speakers right in the corners or against the front wall.
Although some speakers (especially those with wide baffles) can still sound quite good pushed up close to the front wall with EQ applied to correct the bass/lower midrange, they never sound as good in the midrange or image as well as when they're a minimum of about 0.7m from front wall to the front of the baffle. (Any more than about 1m and I find the bass suffers, if you only have mains with no subs)
Last edited:
Placing speakers at the wall is not a problem if the speaker is designed for the purpose, they just shoud be aimed inwards the room.
This is essentially not true if the purpose is to listen to music where what we perceive is mainly the modulation which is not steady state in most music.
- Elias
Unless they're mounted in wall, flush with the wall, pushing any box speaker up against the wall will still cause a time delayed reflection from the front wall.
Steady-state response in this context means the frequency response as measured without any windowing, (swept sine, RTA+pink noise etc) or with a very long window time. (>500ms)
It doesn't mean the music is steady-state
I'm not talking about any box speaker but the one designed for the purpose. If the speaker is not too deep, say 10cm, and placed against wall facing inwards the room, and have some midrange directivity, such a placement is not really a problem.
Steady-state response in this context means the frequency response as measured without any windowing, (swept sine, RTA+pink noise etc) or with a very long window time. (>500ms)
It doesn't mean the music is steady-state
Yes, music is not steady state ! So why measure with such conditions ?
Steady state measurement will give totally wrong indication of the balance and response when music is played.- Elias
But if you place speaker really close to wall, phase cancellation moves to frequencies where speaker does not radiate sound backwards too much. I mean ~5-20 cm from back wall. Question is, I think, if little midrange suck out more or less meaningfull than suck out in mid bass which usually anyway lacks power because of floor cancellation.
Last edited:
How many full size main speakers do you know of that are 10cm deep ? Stretching the point a bit aren't you ? In the context of the discussion we're talking about taking "normal" speakers and moving them to different distances from the front wall.
You know exactly what I mean by steady-state measurements in the bass region, as you're a participant in at least 2 other threads where this has been discussed exhaustively, (in one of them only a week ago) so I'm not sure why you're trying to confuse the issue here where it is OT.Yes, music is not steady state ! So why measure with such conditions ?
It's well proven by research that at high frequencies (especially over 2Khz) our perception of tonal balance largely follows the direct (reflection free) sound, not the room averaged power response. The integration time of the ear is shorter than the time to first reflection. (This is why the power response of a speaker in upper midrange/treble doesn't really matter that much, and why moderate dips in wide off axis response can go unnoticed - this much is relevant to the discussion of midrange driver size)
At bass frequencies the integration time of the ear is very long (hundreds of milliseconds) and is unable to differentiate between the first arrival and the multiple reflections returning from room boundaries - reflections start returning from room boundaries before even a single cycle of the direct wave has completed, thus the steady-state measured frequency response <200Hz is an accurate measure of the perceived frequency response.
If you don't agree, you are at odds with all of the research that has been done on this. The ear doesn't care how much is direct signal and how much is room reflections, it just perceives the whole lot summed together at low frequencies, and it has no other choice when reflections arrive before a direct path cycle has even completed.
Likewise it is impossible to measure the direct path bass response in a room independent of the reflections for the same reasons - all that can be measured is the summed response - the so called "steady-state" response. You claim this is wrong, but how else do you propose measuring bass response in a room ?
5-20cm from the wall ? So your speakers are only 5-20cm thick ?
You have to count the distance from the front of the speaker to the wall, not from the rear of the speaker...
See, this is where we disagree, and what the thread is really about. You suggest that the polar pattern must be as wide as shown at 1Khz to mate with a wide dispersion tweeter. Why ? -3dB at 70 degrees off axis ? Is that really necessary ? What acoustical research is this based on, or does it just look "good" on a polar response plot ?
How will your dome tweeter (I assume that's what you have in mind for a wide dispersion tweeter) sound crossed over at 1Khz ? Pretty awful, I guarantee.
What is wrong with 2Khz, which is -3dB at 35 degrees off axis and -6dB at 50 degrees off axis ? That's a whole extra octave the tweeter now doesn't have to handle, which will reduce power handling requirements by 3-6dB, and excursion at least 4 fold, whilst causing the midrange driver very little incremental increase in load.
You also can't just look at the polar response of the driver by itself, you have to consider the summed polar response of the two drivers crossed over with each other, which always makes the polar response much better even considerably below the crossover point. Crossed over at 2Khz the horizontal polar response will be much better than -3dB 35 degrees off axis.
I actually cross mine over at 4Khz, 3rd order (which will probably make you gasp I suspect) to a wave-guide loaded ribbon. (which is around 90-120 degrees horizontal from 2Khz-10Khz) I wouldn't want to cross over that high to a dome tweeter with no directivity control, certainly.
I was taking a room measurement recently and was surprised at how little the power response actually sags - for a flat on axis response from 1-10Khz (measured windowed at 1m) the sag in power response below the crossover frequency (measured approximately with a steady state measurement at the other end of the room) is quite modest - between about 2-3dB centred around 2.5-3Khz, with a nice smooth broad droop.
When you consider that an even order L-R crossover has an inherent 3dB sag in the power response with driver spacings of a half wavelength or more and people find that perfectly acceptable, this is really not that big a deal.
The major reason for not crossing tweeters over too low is they aren't good at producing midrange.
The SPL requirements in the treble are quite low compared to the rest of the spectrum, but rise dramatically starting at about 3-4Khz and below due to the nature of music spectrum. Every little bit extra you cross a tweeter lower places large additional demands on it.For example I took measurements of a variety of music once and processed them through different high pass filters, and found that just increasing the crossover frequency from 3Khz to 4Khz, (regardless of whether the slope was 12, 18 or 24dB/oct) reduced the power handling requirements of the tweeter by 3dB. Shifting from 2Khz to 4Khz was considerably more than a 6dB reduction in required power handling, due to the slope of the music spectrum.
I know, there are dome tweeters that claim to be able to cross over at 2Khz or even 1.5Khz. Maybe you will get away with it in a low powered bookshelf system that isn't playing at high SPL levels, but as part of a large 3 way floor standing system, forget it, no matter what the manufacturer claims for the driver. Distortion will be high, the driver will be straining at high volume levels, especially compared to an 8" midrange driver. (You have the same problem as the 12" woofer crossed to an 4" midrange, just at the next decade up)
Because I'm using a ribbon, there's an advantage to crossing over as high as you can, since they're not as capable at low frequencies as even a dome tweeter.
Hmm, such as ? Define successful. Large 3 way system with a dome tweeter crossed at or below 2Khz that doesn't sound raspy at high SPL ?
That is what is preached by so many, but I'm not sure that close directivity matching is really that important. What I suspect may be happening in designs where there is a big mismatch in directivity between, say, a large midrange driver and a dome tweeter, is that there is a discontinuity in the baffle diffraction signature of the speaker at the crossover frequency.
As I was mentioning before, by the time you get to about 2Khz there is more or less no baffle edge diffraction with an 8" mid, however lets say you cross over to a normal dome tweeter at 2 or 3Khz, the dome tweeter will have a large diffraction signature from the baffle even with rounded corners, so you are crossing over from a driver where there is no diffraction signature to one where there is a large diffraction signature. This can't be good for imaging.
By choosing a wave-guide loaded tweeter I have a tweeter that has a very minimal diffraction signature even as low as 2-3Khz. (and I'm crossing it at 4Khz)
Above 2Khz there is no baffle diffraction from the 8" mid, crossing over to a tweeter where the 90 degree radiation is so minimal across its entire range that there is very little baffle edge diffraction, so there is a smooth blend from one point source to another, with no sudden appearance of a large baffle diffraction signature.
I think this is what matters more than trying to match directivity at the crossover frequency. Going the traditional route of using a smaller mid and a lower crossover frequency just "solves" the issue by ensuring that both drivers have a similar (but large and unwanted) baffle diffraction signature.
Last edited:
Sorry DBMandrake, but you need to update your knowledge. Integration time at 2kHz is up to 100ms long for a loudness perception. The claim that it's shorter than first reflection is just ridiculous. Please stop spreading false information in all over the threads as people who don't have much knowledge on psychoacoustics may get false ideas and draw wrong conclusions.
No, exactly it is not accourate measure of perceived response, since first of all, steady state assumes infinitesimally narrow bandwidth which does not correspond human perception in any way. It's just basic signal theory.
I'm not looking for an agreement but for facts !
- Elias
Interesting discussion.
Im coming at this from the custom HT high end speaker side and all my testing/measurements have pushed me to using TD12M drivers as my mid range drivers. Im not working slowly on a new dual JBL 2226 design with Iwata/TAD horns giving me even a higher end performance.
Let just say that 6.5" drivers or smaller need not apply for high performance HT applications. Even my 6.5" PHL 1120s had to be XOed higher to TD12S drivers to get enough output.
The discussion around "attack" and having 15dB peaks in music is almost doubled when it comes to movie performance. Essentially I use 25 to 30dB peaks as a variable in calculating my needs. Not only do we need to worry about driver sensitivity,Xmax, linear performance during peaks we need to ensure our amplication head room exists. I would say that 99% of all HT setups clip the amps and send the drivers way into distortion/compression levels during split second peaks, people do not know the difference until they are exposed to clean peaks using higher end drivers and proper amplification. Its then easy to hear the difference. Once you have a setup that has a "clean attack" you will know it. Everything else just sounds "choked off".
Lets just say that if you like to listen at 85dB @ 12feet then you better have 120dB max SPL and high sensitivity drivers so that even 80Watt/CH AVRs do not clip.
Im coming at this from the custom HT high end speaker side and all my testing/measurements have pushed me to using TD12M drivers as my mid range drivers. Im not working slowly on a new dual JBL 2226 design with Iwata/TAD horns giving me even a higher end performance.
Let just say that 6.5" drivers or smaller need not apply for high performance HT applications. Even my 6.5" PHL 1120s had to be XOed higher to TD12S drivers to get enough output.
The discussion around "attack" and having 15dB peaks in music is almost doubled when it comes to movie performance. Essentially I use 25 to 30dB peaks as a variable in calculating my needs. Not only do we need to worry about driver sensitivity,Xmax, linear performance during peaks we need to ensure our amplication head room exists. I would say that 99% of all HT setups clip the amps and send the drivers way into distortion/compression levels during split second peaks, people do not know the difference until they are exposed to clean peaks using higher end drivers and proper amplification. Its then easy to hear the difference. Once you have a setup that has a "clean attack" you will know it. Everything else just sounds "choked off".
Lets just say that if you like to listen at 85dB @ 12feet then you better have 120dB max SPL and high sensitivity drivers so that even 80Watt/CH AVRs do not clip.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.