Partial summation, the nulls won't be complete when they are at different levels.
Knowing where each range is low enough to be out is something that should be done deliberately. Use as shallow a slope as you can to overcome all the acoustic issues, be they breakup or what you mention here or something else. The acoustic conditions set the needed slope.
Well, you're hitting on one of the two reasons i like steep linear phase xovers...that being it narrows the range of summation, and hence the range of poential ripple from geometric offsets.
I like to mentally picture the phase wheel to help remember that the 1/4 WL spacing is just a rule of thumb for getting a high probability of good results.
There's still quite a bit of variability in summation, even when within +/- 90 degrees.
But the good news is that summation is for 1 dimension electrical summation, and it's never quite as variable for drivers in 3D space radiating out to different points.
(Second reason I like steep xovers is that it is simply so much easier to get drivers response flattened over a narrower range than a wider range.)
Yes. Remember all the port hole size/location/shape experiments i posted on another forum a while back? Those were to see the disruption of ports in the horn walls, on the CD's output. Heck, I was even trying smiley and frowny faced ports, and airbake pizza pan ports (all the little holes) lol.
Sure, some made a difference in CD polars (frowny face arc was best)......
But really, they had to be compared to a horn with no ports at all, and I've concluded rightly or wrongly, the ports don't mess up a CD's response much worse that it is with just a straight horn
Mouth termination seems to matter more than ports ime.
And I've kinda decided even that isn't worth fixing given all the trouble it takes with secondary's curved flares, etc. Good ole audio compromises, huh 😕
Sure, that's a totally viable method...in fact, THE method if you want to use passives.
I'm going to try to offer my 2c re the high-pass for CD, in a reply to pelanj now
You just use some PEQs to knock down response below xover frequency instead of a high-pass filter. Kinda the same principle as boosting the bottom end of a subs response with PEQ's instead of shelving.
The only drawback to using just PEQs, is that usually folks use them for a comparatively narrow range below xover frequency, knocking down response that is deemed to create potential over excursion.
Which is all fine and good until the CD is run hard for a long time. The frequency region below the PEQ's, while not producing much if any excursion due to natural rolloff, is still being passed unattenuated to the CD, producing current and heat. Bur again, probably a problem only for prolonged hard running.
Named vs un-named filters is 99% semantics, imo.
Because unless ALL you use as filters are named filter high=pass, low-pass, or xover ....... no other PEQs, shelving, all-pass, notch, etc....
the moment you combine the named filter with any other filter, whatever it is...the combination of the named and the other filters, which is the NET filter, certainly has no name and is unique.
Personally, I know I want to achieve acoustically complementary output, or iow, achieve fully complementary acoustical xovers .
I see certain electrical xover topologies like Linkwitz-Riley are fully complementary.
I say to self, why not massage a driver's acoustic response to match the shape of LR response, at whatever order I desire?
Then I achieve fully acoustically complementary xovers.
It's so much easier ime/imo, to simply make a driver's target response an LR xover curve, than start from scratch with to drivers, juggling both their resposnes looking trying to get to acoustic complementary.
And the icing is, given the ease of just matching a a driver's response to LR (or any other fully complementary xover),
is that for any given realized acoustic order achieved, both methods necessarily give the exact same net filter....which again will be un-named 😉
Perhaps the only drawback to matching a named filter's response, is that you can't really get in between common orders, like say 9 dB octave. Or at least I don't know how to yet....and not that I'd want to anyway....🙂
Lol Kevmoso 🙂
My girlfriend thinks I'm simply too stupid to realize what i can't do anymore ...
Thanks, I think I get it now. So I really only need to attenuate around the knee, and if possible, attenuate also below.
Yep, although do both and it's just a high-pass, huh? 🙂
Personally, I'd never-ever use PEQs as a substitute for a CD high-pass...
I have found that MEH crossovers do not affect the loudspeaker output like multi-way multi-aperture loudspeakers--in ways that most people do not seem to see. Essentially, you're guaranteed no lobing in an MEH due to the crossover, so higher order filters are just not needed. Particular attention to phase response is ushered in as the most important factor to achieve in the crossover implementation.
In the case of MEHs, I've verified experimentally that a first order crossover achieves almost identical phase and SPL response results as a "zeroth order" crossover (i.e., no phase shifts are induced electrically by the crossover network) with electronic delay of the LOWER frequency drivers corresponding to 90 degrees of phase lag at crossover center frequency. This means that a first order crossover filter is sufficient for the crossover in an MEH. No higher order filters should be used (unless using FIR filtering).
Now...the (unfounded) fears of DIYers using this approach:
What about the compression driver "seeing" lower frequency electrical power from the crossover network below the crossover frequency? Won't that lead to compression driver damage?
If you're trying to use a 1" compression driver and trying to cross too low in your MEH (usually, this is below 1 kHz), you've already got a problem, regardless of the crossover network, because you're trying to use the driver too low in frequency. I generally don't recommend 1" compression drivers at all. Use a 1.4-1.5 inch exit driver instead.
So if we're now on the same page (and you've gotten over your issues with not using 1" compression drivers in your MEH), then the next fear seems to be that crossing a 1.4/1.5 inch driver as low as 500-600 Hz. If you make your choice of compression driver carefully, then they can easily cross as low as 500 Hz using first order filters without fear of any damage under acoustic load conditions. So now "what about the sound quality of the compression driver playing that low?". I've heard compression drivers that don't sound very good if playing below ~400 Hz (i.e., below the crossover frequency) due to gentler slope electrical crossover filters. The solution is simple if you're using DSP crossover: just add attenuating PEQs and perhaps an attenuating shelf filter (i.e., not a crossover filter but rather an EQ filter) to more quickly attenuate the compression driver's output. This is easy to do after you get flat phase and smooth SPL response through the crossover band. NOTE: This is only required if you are consistently playing the MEHs at levels above 100 dB (which is usually a small fraction of a watt electrical input power).
An example:
No "named" crossover filters are used in the above example. Two attenuating PEQs and one attenuating shelf filter are used to achieve a ~30 dB/octave high pass filter. (You can probably do essentially the same thing with only two EQ filters.)
Many other variations on this approach are possible. The above example is a very simple one. One should also realize that the LF response of the compression driver itself should be taken into account, At 400 Hz, a 2" compression driver's output is likely to be 10-20 dB down, and at 200 Hz, perhaps 30-40 dB down--by itself.
Please do not use the above simple example as "the solution" for your application. Instead, use the example to generate ideas on how to create a better zero-phase-shift high pass filter for your application. The low pass portion (the woofer) is even easier to implement.
The rule here is don't boost the compression driver's low frequency output down low (around the crossover frequency), and you won't have any problems. In compression drivers, if you are exceeding the diaphragm travel limit, you will hear it when the diaphragm bangs into the pole piece or the phase plug. For home hi-fi applications, this should never happen, especially at sub-1 watt input power (which is deafening in a home environment at 100-110+ dB).
For higher power applications like PA duty, you would likely use first order filters in your MEH with attenuating PEQs below the crossover point (like DSL uses) to achieve very high power performance.
As for the lower frequency driver low pass filter (in my MEHs this is the woofer, since I use two-way MEHs)--it will naturally roll off above the crossover (first notch frequency), so only the attenuation of the second and perhaps third response peaks above the crossover (first notch) frequency is needed. No worries there.
Chris
Last edited:
Note that if you are trying to use higher order filters with an MEH, you must use FIR filters to avoid the 90-degrees-per-order phase lag of the lower frequency drivers relative to the higher frequency drivers. This is what Mark is doing.
I don't ever recommend using anything higher than first order filters in an MEH using IIR filters, or you will induce a resulting loudspeaker phase response that will effectively create multiple point sources in the MEH aperture, and you will lose the single point source performance of a "Synergy" class MEH--which is a huge loss in performance in my experience.
Chris
I don't ever recommend using anything higher than first order filters in an MEH using IIR filters, or you will induce a resulting loudspeaker phase response that will effectively create multiple point sources in the MEH aperture, and you will lose the single point source performance of a "Synergy" class MEH--which is a huge loss in performance in my experience.
Chris
Thanks Mark and Chriss for sharing all that knowledge and experience. And all that in one thread!😀
Also thanks for addressing "common DIY fears", I certainly suffered from them! 😱 It´s good to know that it actually is that "simple"/safe/possible to achieve a good "synergy" class result.
Steffen
Also thanks for addressing "common DIY fears", I certainly suffered from them! 😱 It´s good to know that it actually is that "simple"/safe/possible to achieve a good "synergy" class result.
Steffen
Last edited:
I'd like to strongly second this advice.
If either going passive, or with DSP that is IIR, I personally believe anything beyond 4th order induces too much phase rotation....for any speaker design.
And as Chris points out, a MEH/unity/synergy has an acoustic design especially made to dovetail with first order....(and as explained in the synergy patent.)
So really, unless experimenting with my steep complementary linear phase xover FIR method.....first order is the way to go with these speakers.
You're referencing the electrical slope, right? Not the combined slope that's influenced by the driver's rolloff and the acoustic filtering that's introduced by the horn itself?
This is a SUPER clever way of enclosing the mids, but not making it permanent.
I'm totally copying this!
I have an entire shelf full of midranges that I sealed off with 3D printed enclosures that are kinda useless now, because the enclosure is glued to the basket.
As the frequency decreases to 400 Hz and below, the total SPL response attenuation will be greater than the example that I gave, above for almost all compression drivers (except perhaps the Celestion Axi2050--which has another octave of low frequency performance on a horn the size of Marks 11th prototype).
The example I presented above stands on its own for those that wanted to know how zeroth order crossover filters work. If you wish to have more discussions on this, I think a new thread is warranted so Mark can retain this thread's purpose for his full-range MEH test bed.
Chris
Aaah thx !
Glad the method might be of use.
I'm always, always. looking for the easiest, most flexible way to proto stuff way, possible. Fit's my lazy man, scrounging type style .
Btw, on syn10 which uses the same mids/same place (and I totally love the sound of)...i leave the mids open-baffle.
In syn9 version comparisons, this proto box was sealed as a comparison to open-baffle, and I couldn't hear or measure a difference even backed up against a reflective rear wall...so I built syn10 open-baffle like.
But for this big syn11, I needed of course needed the mids sealed up from the 18"s....glad i had the old proto, sealed version...
Hi Chris, Please don't have any concerns about this thread's purpose.
I personally don't feel that as thread starter, that I (or OP's in general) hold any rights on what a thread should stay about.
Sure, I want to show my build, lay out my design strategies, and the results.
But i think it's healthy when contrasting ideas, strategies, methods, alternatives, critiques , even challenges .....emerge in a good civil thread....
Otherwise is seems too controlling and classroom-like to me, like back in school, (yuck haha) ...instead of dudes sharing...
I am now dreaming of Synergy horns again. Amazing work. Can someone make a decent kit please...
Thx Vermouth, hope you can find some plans or a kit that works for you.
@Djkidmt,
Square waves are fun to play with.
They are soo dang sensitive; I think more so than our FFT measurement programs.
Biggest problem I have with them though, is how many dang measurements you have to make, and then how difficult it is to piece them together for what they mean across the entire freq spectrum.
For instance, if i use a square wave test signal whose frequency is that of the xover frequency, all the resultant measurement will show me is the relationship of that fundamental xover frequency with its 1/3 octave odd harmonics (that were used to generate the test signal square wave.)
So not a lot of info....plus, since square waves rely on a sizeable series harmonics, any high freq work is out of the question due to a 20Khz audio range ceiling that truncates needed harmonics.
(Maybe you already know all that.)
Anyway, good old mag and phase say it all for me, which is in post #2. (and those traces will give very nice square waves )
@Djkidmt,
Square waves are fun to play with.
They are soo dang sensitive; I think more so than our FFT measurement programs.
Biggest problem I have with them though, is how many dang measurements you have to make, and then how difficult it is to piece them together for what they mean across the entire freq spectrum.
For instance, if i use a square wave test signal whose frequency is that of the xover frequency, all the resultant measurement will show me is the relationship of that fundamental xover frequency with its 1/3 octave odd harmonics (that were used to generate the test signal square wave.)
So not a lot of info....plus, since square waves rely on a sizeable series harmonics, any high freq work is out of the question due to a 20Khz audio range ceiling that truncates needed harmonics.
(Maybe you already know all that.)
Anyway, good old mag and phase say it all for me, which is in post #2. (and those traces will give very nice square waves )