Yeah, usually more info is better than less 
However, initially I'll go as most people do for on-axis optimization and then listening trial & error.
Anyway, the key principle for this xover is more impulse timing optimization than FR. I find timing the most underappreciated and undervalued feature in speaker design.
That's why coax is here For me coax actually should simplify the design rather than make it more challenging. Why do you think it's more tricky than separate drivers ?
However, initially I'll go as most people do for on-axis optimization and then listening trial & error. Anyway, the key principle for this xover is more impulse timing optimization than FR. I find timing the most underappreciated and undervalued feature in speaker design.
That's why coax is here
For me coax actually should simplify the design rather than make it more challenging. Why do you think it's more tricky than separate drivers ?
Last edited:
Also coaxial can have excess group delay due to crossover such as any other multi-way. Also time alignment (including excess GD) could be off. So there's not much difference - certainly not so much that you should ignore something else.
Just look the datasheet. Is on-axis, 30 deg or 60 deg the right one and representative for crossover design or not? Or is it electrical response based on assumption that tweeter of coax converts balanced electrical power to balanced acoustical power?
Single off-axis response could be representative, but the problem is to know which one.
>Also coaxial can have excess group delay due to crossover such as any other >multi-way.
Obviously, but this has little to do with coax.
>Also time alignment (including excess GD) could be off.
Do not get it - for me this sentence is basically the same as the previous one - no time alignment but solely due to the xover. So you "just" should be careful with the xover to preserve it. As far as it is doable with a pasiive xover of course... I shall also do a DSP xover version as a next step shortly after passive.
>So there's not much difference - certainly not so much that you should ignore >something else.
For me there's a whole lot of difference - without a coax you have a serious built-in compromise in time alignment and the xover is rather on top of that.
It is not about ignoring just setting the priorities for the case they cannot be optimized simultaneously
Obviously, but this has little to do with coax.
>Also time alignment (including excess GD) could be off.
Do not get it - for me this sentence is basically the same as the previous one - no time alignment but solely due to the xover. So you "just"
should be careful with the xover to preserve it. As far as it is doable with a pasiive xover of course... I shall also do a DSP xover version as a next step shortly after passive.>So there's not much difference - certainly not so much that you should ignore >something else.
For me there's a whole lot of difference - without a coax you have a serious built-in compromise in time alignment and the xover is rather on top of that.
It is not about ignoring just setting the priorities for the case they cannot be optimized simultaneously
Last edited:
For example TAD Compact Reference CR1:
Some real life 4-way with separate conventional drivers:
An externally hosted image should be here but it was not working when we last tested it.
Built-in compromise in time alignment is quite well handled isn't it?
Of course you are free to focus any individual feature and use as narrow view to reality as you like. Not my bad.Interesting discussion and example
My first impression of this IR was it is not my high fidelity.
But if understand this correctly on a fast glance actually the key timing problem here is the woofer being delayed and that's less of a problem.
The mid and twt are not fully in sync but how problematic it is would need some closer analysis and comparing how it compares to other commercial products.
What is your view on that ?
If you attached a well-behaved 4-way doesn't it mean you also pay quite a bit attention to it ?
Last edited:
Have you tried using the Xsim circuit blocks, for a series resonant circuit? Put it across the driver (or earlier in the network -- but NOT across the power amp) and a resistor or other series impedance ahead of it.
But XSim was designed to be mostly used iteratively, to avoid things like needing formulae and related tools. Put in the component arrangement for what you are after, then scroll through values of eac components to see what happens -- it doesn't take long to get a feel for what is going on or for the size of inductors, resistors or capacitors that affect the frequencies you desire. I think that way is more instructive than using some online calculator (which seldom gives the expected results when attached to real world impedances like those of drivers).
BTW, it might be better if you attached the '.dxo' file for your crossover to your post. Then other people can start with what you have, including the driver responses and impedances you are using (those are part of the .dxo), Not many members here will want to spend the time to enter the circuit by hand and guess what your drivers behave like. With a dxo, they can just click, open, and play with components.
edit: for example, download (click in most browsers) to get this example, play with C6 to see how it behaves, then with other components like L4 and R5.
Last edited:
If you attached a well-behaved 4-way doesn't it mean you also pay quite a bit attention to it ?
I design minimum phase speaker if that's possible with selected drivers and crossover technology. All combinations do not allow minimum phase design with practical methods. In those cases I try to keep excess group delay below ~2 ms at 100 Hz.
Effect of timing to dynamics is more perceptible by tactile sense than ears, and dynamics depends on some other factors such as dynamics/compression of drivers and acoustical design. Minimum phase design is far from priority #1 if the other factors are favorable for dynamics so acoustical 2nd order for the lower and 4th order for the higher crossover point in 3-way are usually good enough, and no need to make compromises elsewhere to force speaker to minimum phase.
I assume you meant physical time alignment? That is meaningless as a primary design parameter because it does not define timing. You can't read timing from physical distance difference(s). Excess phase and excess group delay responses show timing compared to minimum phase design which is valid goal (though not necessarily priority #1 as mentioned earlier). In the contrary, timing could be the only hope in case drivers and acoustical concept are weak and loose. In addition, speaker where all these are flawless may sound disturbingly dynamic. That is very rare.
Last edited:
Hi Kimmo, I am not sure what kind of phenomenon effect of timing to dynamics is, does it mean that the acoustic point changes depending on matters such as VC position in the magnet gap?
Could you please give me feedback if again my EE knowledge and passive xover assumptions are correct: as far as the dips correction (local dips below the target average FR level in the region covered by only one driver) is concerned the only way to do it is to lower the overall efficiency and then use a reverse notch (bandpass filter) that could locally lower the impedance in this region and effectively boost locally FR ?
If the above is correct then it seems to have little sense with a passive xover it is rather not done in practice ? So dip correction make more sense with DSP (although along similar principles but with much more flexibility) ?
And one more tool question - the fact the acoustic centres are displaced in space has impact on the overall speaker FR but it is not possible to model in XSim ? So XSIm model assumes all the drivers are coincident in space ?
If the above is correct then it seems to have little sense with a passive xover it is rather not done in practice ? So dip correction make more sense with DSP (although along similar principles but with much more flexibility) ?
And one more tool question - the fact the acoustic centres are displaced in space has impact on the overall speaker FR but it is not possible to model in XSim ? So XSIm model assumes all the drivers are coincident in space ?
It is not due to room, my Seas coaxial has a dip at 9k probably due to it being coaxial and mechanical consequences of tweeter location.
I have tried to boost it with an RLC but it did not work in XSim and I thought it does not work by principle - if the RLC should work by providing lower impedance at some resonance region then it is not gonna be effective since there is no higher impedance to work against - the tweeter xover is low impedance in this region as well ?
If I am lost EE-wise again please suggest what kind of circuit for 9k dip should I try ?
I have tried to boost it with an RLC but it did not work in XSim and I thought it does not work by principle - if the RLC should work by providing lower impedance at some resonance region then it is not gonna be effective since there is no higher impedance to work against - the tweeter xover is low impedance in this region as well ?
If I am lost EE-wise again please suggest what kind of circuit for 9k dip should I try ?
Does the dip do anything bad acoustically? I wouldn't think so, and probably even with active EQ it would be hard to flatten the dip and unpleasant to listen to the corrected driver.
Bumps you can flatten nicely with a notch and it sounds better. Dips normally go unnoticed.
Bumps you can flatten nicely with a notch and it sounds better. Dips normally go unnoticed.
