Absolutely agree. I did not want to compromise on 2nd order filters (ish).
But the parts count.... The all pass filter vs a higher order filter. Not too big of a parts count difference, really.
In hindsight, I wish I would have built in a physical distance difference for the tweeter, or used a waveguide, but I think the all-pass with some decent but not crazy expensive caps and coils does an Exceptional job here. I used the clarity caps and madisound house coils for my whole build.
Rats nest of a external crossover below. I will clean it up and lay it out better now that I am happy with it.
Attachments
Not really what I meant. It's more of, in some circumstances if you want to get 4th order LR, you'll need to ADD more parts for the all pass. That is, the insistence on a higher order, or specific order of filters may force the all pass.
Often with 2-way designs a combination of 3rd order HP and 2nd order LP (electrical) can be tweaked for ideal phase matching. If however you insisted on a particular order and would not use a stepped baffle, then the all pass is your answer.
If using a symmetrical filter in a case where one driver is delayed causes the lobes to be somewhere they shouldn't, wouldn't using an asymmetrical filter which brings phase to where it should be, fix this?
@AllenB - I was thinking more in terms of how higher order filters help reduce destructive sums off axis, and comparing that to well executed lower order filters, especially above the tweeter axis.
Clearly, better to have well constructed 2nd, 3rd order filters than any order if done badly. However, if I decided that 4th order responses were "da bomb" for my design, then I have to deal with the fallout of that decision.
For a passive speaker though, I normally pick the filter order at least in part, based on the phase alignment, and thus dread attempting LR4 in a 2-way. 🙂
Clearly, better to have well constructed 2nd, 3rd order filters than any order if done badly. However, if I decided that 4th order responses were "da bomb" for my design, then I have to deal with the fallout of that decision.
For a passive speaker though, I normally pick the filter order at least in part, based on the phase alignment, and thus dread attempting LR4 in a 2-way. 🙂
So I have a question: each seccesive "order" increases the complexity of the phase curve. So why would a 3rd order be preferred over a 2nd order with a phase shift?
So consider a simple case where you have 2nd order butterworth filters and drivers on a flat baffle. Electrically this would result in each driver turning 180 degrees at the crossover point. So, you reverse tweeter or woofer and voila, instant alignment. The problem is the acoustical distance of the tweeter and woofer to the listener are unequal. The tweeter is often closer to the listener, on the order of 1" or more.
And here is where you are left with several options:
- Put the woofer in front of the tweeter with a stepped baffle.
- Use an all pass filter to delay the tweeter
- Change your electrical orders and tweak....
By the way, all of this is easy to see if you use XSim or another crossover simulator. Use the mod delay setting on a simulated woofer and set it to 1.2" or so, and then attempt 2nd order filters. 🙂
I think we are all agreeing with you here.
I just chose to keep a flat baffle, and chose #2.
And i wondered why more enthusiasts dont consider a similar path vs #3.
Both are great options, just different ways to achieve a nice phase alignment.
I used the impedance shaping circuit in the bottom right to get that nice flat impedance. Without it, it is more variable, but still only peaks at 12 ohms.
You can add perhaps the tweeter off setting : mid rigth in front of ears and tweeter off setted above or below that axis to add distance from the sweet spot (=delay). Certainly works fine in the limit of 1.2 WL CtC btween the two drivers at cut off ?
if used to compensate for delay, this is one of the properties of an asymmetrical filter which needs to be verified, its lobing property.
Lobing can be determined, that is it can be expected to be a certain way if you measure, according to 3 primary factors. Source spacing, wavelength and source phase.. two of which don't change.
@ctrlx Slanted baffles are a compromise which keep drivers from performing to their best potential on and off axis. When you're on the time aligned axis where the driver's integrate in correct phase, you're listening off axis. That can be a big compromise on larger 3 way systems with slanted baffles. Its just as (in)effective to tilt the cabs.
Stepped baffles are a much better approach which fix the problem the correct way for best phase linearity and smooest possible FR.
Stepped baffles are a much better approach which fix the problem the correct way for best phase linearity and smooest possible FR.
Why not straight vertical tweeter and angled back woofer for alignment? This seems like it would help alignment, keep tweeter on axis, and be closer to on axis with the woofer tilted closer to facing the listener.
I've not done one like this myself yet, but have that on my list to try.
I've not done one like this myself yet, but have that on my list to try.
@wolf_teeth That baffle configuration would dictate a specific listening distance, which may be a problem if its not designed for the application. Otherwise its a better option than tilting the whole baffle IMO.
LX, if carried to its logical conclusion. (I've posted many minimalist LX to the Fullrange Photo Gallery.)
BTW, to find a driver's acoustic center at any frequency, just LX a pair and mark the top driver's offset over the bottom driver's physical center, where the test-tone on-axis combined loudness is max (or null if polarity reversed). Of course this works too for time-aligning different drivers etc. XO phase-alignment still needs to be done but may be greatly simplified.