What kind of effect does driver "beaming" have on the expected polar response at crossover? I have always seen simulated phase plots calculated with the assumption that each driver is a point source radiator. This is not at all true for some popular drivers, with long ribbons as the extreme example.
I know that there is no definitive general answer, but what about this case: a beamy mid (planar or large cone) crosses over to a long thin ribbon fairly high up. What differences from the perfect driver pictured below will occur in the real world?
I know that there is no definitive general answer, but what about this case: a beamy mid (planar or large cone) crosses over to a long thin ribbon fairly high up. What differences from the perfect driver pictured below will occur in the real world?
An externally hosted image should be here but it was not working when we last tested it.
This is a real can of worms. One of the most common faults of crossover design is exactly what you're saying here- neglect of the effect of actual driver dimensions and placement on polar pattern. What you end up with with is an irregular horizontal dispersion with frequency and at worst, the speaker never sounds balanced no matter how much you jiggle the xover and eq, at best you have a speaker which is HIGHLY critical of placement and room treatment. The ear senses not only the direct sound, but the reverberant field- optimize one and you screw up the other unless driver dimensions are taken into account. And when you factor in break-up modes, it gets even worse.
A common "bad" two-way that can really illustrate the problem is the classic 8" crossed over to a 1" tweeter. If you make the on-axis response flat, the off axis response will almost always have a dip near the top of the woofer passband, then a peak as the tweeter rolls in. That can only be avoided by setting the crossover quite low, like 1200 Hz, which will NOT make a 1" tweeter very happy.
A common "bad" two-way that can really illustrate the problem is the classic 8" crossed over to a 1" tweeter. If you make the on-axis response flat, the off axis response will almost always have a dip near the top of the woofer passband, then a peak as the tweeter rolls in. That can only be avoided by setting the crossover quite low, like 1200 Hz, which will NOT make a 1" tweeter very happy.
Not an easy thing, designing loudspeakers. 
Here is a ripple tank simulator that I tried to use to emulate line sources...a true line source should be a good model for the situation in the vertical axis, right? It looks like the center lobe is not much altered but the off-axis lobes, especially the lower one, are seriously attenuated.
Does this connect with reality?
Here is the ripple tank I used. I didn't play with the wavelength yet.
http://webphysics.davidson.edu/Applets/Ripple/Ripple.html
Still looking for a simulator that supports line sources.
Here is a ripple tank simulator that I tried to use to emulate line sources...a true line source should be a good model for the situation in the vertical axis, right? It looks like the center lobe is not much altered but the off-axis lobes, especially the lower one, are seriously attenuated.
An externally hosted image should be here but it was not working when we last tested it.
Does this connect with reality?
Here is the ripple tank I used. I didn't play with the wavelength yet.
http://webphysics.davidson.edu/Applets/Ripple/Ripple.html
Still looking for a simulator that supports line sources.
Tiroth:
I don't have a definite answer to the question, but as you say the response will be different. My intention with the XDir program was mainly to understand the effects that can occur, in principle, not the details in a specific case. In order to understand something, I find it useful to start with simple models, and then to refine them if needed. I might do that some time. It is not very simple, though, modelling the driver surfaces with a bunch of point sources would be one way, but at frequencies where the cone does not move as a rigid piston simulations become very difficult.
Now the question was not about simulation techniques, so I guess I have answered nothing whit the above. However I still think simplified simulations can help the understanding of more complicated systems. I think you would find some of the effects in the real world similar to what XDir predicts, even if not identical.
I don't have a definite answer to the question, but as you say the response will be different. My intention with the XDir program was mainly to understand the effects that can occur, in principle, not the details in a specific case. In order to understand something, I find it useful to start with simple models, and then to refine them if needed. I might do that some time. It is not very simple, though, modelling the driver surfaces with a bunch of point sources would be one way, but at frequencies where the cone does not move as a rigid piston simulations become very difficult.
Now the question was not about simulation techniques, so I guess I have answered nothing whit the above. However I still think simplified simulations can help the understanding of more complicated systems. I think you would find some of the effects in the real world similar to what XDir predicts, even if not identical.
IMHO, you need a high order crossover to make it work. A good 4+ order XO, around 2 kHz can definitely successfully integrate an 8" woofer an 1" tweeter.Bricolo said:
The premise is to shut off the woofer as quickly as possible, and likewise the tweeter. Use the woofer as high as you possibly can (most 8" woofers are OK - from an off-axis standpoint - up to 2 kHz), then cut in to the tweeter. Roll off the low end of tweeter to reduce distortion. Plus, with a high order XO, you have reduced overlap, so other interference issues are reduced as well.
Dan Wiggins
Adire Audio
Svante said:
Svante, I must thank you for your excellant little apps. I think they are very useful and there is a lot to be said for acheiving an 60% match to reality...especially when the next 20% requires FEA and the final 20% can't be simulated at all.
I think it is a testament to the quality of your programs that I am always wanting more: line sources in xdir, sources of multiple sizes in The Edge, etc. The work is never done.
Wouldn't a ribbon tweeter, if in the long part going vertically up and down beam in the vertical direction and not in the horizontal, which I wouldn't see to be that bad?
I built a Eton 8.1 type 2-way (except with MB Quart tweeter and little different x-over, also sealed enclosure). It has a 6 db/octive on the woofer and 12 db/octive on the tweeter and I'm happy with it although it might beam a little, I'm not sure.
I built a Eton 8.1 type 2-way (except with MB Quart tweeter and little different x-over, also sealed enclosure). It has a 6 db/octive on the woofer and 12 db/octive on the tweeter and I'm happy with it although it might beam a little, I'm not sure.
One "funny" approach to mating an 8" with a 1" was an old British design by (I think) Rogers. The woofer fired through a vertical slot 4" or so wide, which caused it to have a dispersion pattern that looked like a much smaller driver. The cavity resonance was reduced by appropriate eq, albeit imperfectly.
A fine example of a well-done two-way was the cheapo Super Zero from NHT. And of course, the classic LS3/5A. And just about anything from PSB in the last 7 or 8 years.
Yes, that's right. The limited vertical dispersion may or may not be a bad thing, depending on the design goal, but it WILL be noticeable, as will the change in spectral balance with distance in the near field due to the 1/r power law of the ribbon contrasted with the inverse square power law of the woofer.
A fine example of a well-done two-way was the cheapo Super Zero from NHT. And of course, the classic LS3/5A. And just about anything from PSB in the last 7 or 8 years.
Yes, that's right. The limited vertical dispersion may or may not be a bad thing, depending on the design goal, but it WILL be noticeable, as will the change in spectral balance with distance in the near field due to the 1/r power law of the ribbon contrasted with the inverse square power law of the woofer.
SY said:
Yep
So let's face it!
IMO you will only get the possibility of fairly correct reproducing, If you have the same distance to both speakers. In other words, there will always be phase problems, unless you are sitting in "the sweet spot"
So why really bother about off-axis. Stay away from off-axis!
Make speakers that doesn't spread out the sound, because it will only give you problems with reflection. You wil get the best sound, if your own room is not adding anything to the sound. Place your chair in the center - let there be distance from walls to speakers, let the speakers point at your face (ears) and close your eyes and listen.
Stop bothering about off-axis performance - drop it - use dipoles - ESL's, magnetostats and things like that. JMO.
Thanks for all the comments. Let's see if I can stir things up a bit.
There appear to be three schools of thought on this:
1. Directionality is good : eliminates reflections
2. On axis is important, off axis performance should smoothly degrade
3. We need really good off-axis support, maybe 60 degrees of good fidelity in the horizontal and >30 in the vertical
As usual, each seems to have its proponents and detractors. Something to chew on: the Pass Rushmore appears to use the very design I posited initially: big mid, tall ribbon, high xover point. It must work, so why does it?
My guess would be you need to be far enough away that the cone of the sweet spot is reasonably wide. If you figure, even 10 degrees of lobe is 1' at a distance of only 6', which is probably sufficient for a pretty broad range of listener heights (seated).
A mitigating factor is probably that crossover is occurring so high that we are nearly in supertweeter range: losing some sparkle is not as bad as losing some the range of 3-6kHz.
There appear to be three schools of thought on this:
1. Directionality is good : eliminates reflections
2. On axis is important, off axis performance should smoothly degrade
3. We need really good off-axis support, maybe 60 degrees of good fidelity in the horizontal and >30 in the vertical
As usual, each seems to have its proponents and detractors. Something to chew on: the Pass Rushmore appears to use the very design I posited initially: big mid, tall ribbon, high xover point. It must work, so why does it?
My guess would be you need to be far enough away that the cone of the sweet spot is reasonably wide. If you figure, even 10 degrees of lobe is 1' at a distance of only 6', which is probably sufficient for a pretty broad range of listener heights (seated).
A mitigating factor is probably that crossover is occurring so high that we are nearly in supertweeter range: losing some sparkle is not as bad as losing some the range of 3-6kHz.
There also seem to be rooms where one or the other works best.
and don't forget: Everyone's taste is different.
But one thing should definitely be avoided: A linear on axis response only combined with an off-axis response with large peaks and dips. One has to remember that the perception of timbre is not only depending on direct sound but on the reflected one also. So the design with perfect on-axis response and lumpy off-axis response will only sound smooth in a completely anechoic environment.
Regards
Charles
medum said:
What makes you think there will not be phase problems in the sweet spot? Besides the obvious difficulty of correctly aligning both speakers, putting your head exactly in the sweet spot, not moving it, etc.. you also have to consider variation between drivers, finding a crossover that yields perfect acoustic phase, etc...
Quite good advice if you're able to make an anechoic chamber to listen in, as phase_accurate pointed out. However, constructing a chamber that is truly anechoic down to a few hundred hertz is beyond what many people are able and/or willing to do.
In a regular room, you cannot avoid a significant contribution from the off-axis response, quite simply because of reflections that cannot be perfectly controlled. Most people also find it quite disconcerting to listen to sound without these reflections.
I'd recommend reading some of the JBL papers on public address systems, they contain a few important points about how indirect sounds contribute.
You need to have a controlled off-axis response. A simple rule of thumb is to pick a crossover point where the directivity of the drivers match as closely as possible.
By a "speaker that doesn't spread out the sound", I'm assuming you mean one that has limited directivity. The only way of getting this limited directivity is through the use of a horn.
Even so, with the arrangement you've suggested, the off-axis sound is contributing to the reflections, and hence the perceived timbre. In fact, off-axis sound is the main contributor to reflections.
Also, if you want to limit directivity, shouldn't you be limiting it to the same directivity as the microphone used? That's a quickly moving target, or as the case may be, multiple moving targets.
Of course, you could be suggesting having the speakers pointing at you from the sides. That gives poor and chaotic imaging in my experience, and is pretty much equivalent to using a set of headphones, except you can get significantly better headphones for a much smaller price. (Etymologic, Sennheiser and Stax spring to mind)
Panel drivers have a particular off-axis performance. Their main strength in this area stems from the fact that you're using a single transducer to reproduce the frequencies that are supported by the reflections. Hence, you get no irregularities in the off-axis image.
With dynamic drivers, even in a dipole arrangement, you still have to pay attention to the directivity of the drivers, or you'll end up with an "unnatural" off-axis image.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.