That is how I end up with the acoustic slopes - I tried REW and Open Sound Meter getting very similar results. A combination of PEQ and various textbook crossovers to match a target curve. Maybe it is correct, since it sounds pretty good, but I keep getting weird impulse and step responses, so I think I am still missing something - I need to think about it more. Eventually, one day it will snap in
That is how I end up with the acoustic slopes - I tried REW and Open Sound Meter getting very similar results. A combination of PEQ and various textbook crossovers to match a target curve. Maybe it is correct, since it sounds pretty good, but I keep getting weird impulse and step responses, so I think I am still missing something - I need to think about it more. Eventually, one day it will snap in
Hi pelanj, i think final realized acoustic slopes are all the matters, with regard to xovers.
And they need to be fully complementary...ie both mag and phase..
It doesn't make a flying whatever, how they are electrically realized.... in the end whatever was done electrically, with whatever dsp functions, be they PEQs, shelving filters, conventional name crossovers ... to get the desired commentary acoustic xover..
.....all distinctions on what was used to get to fully complementary acoustic xovers.... end up being simply working with our processor capabilities, and how the processor names them.
That was the "snap in" realization for me.... that final realized acoustic slope is all that matters...and it MUST be fully complementary
Hi Speaker Bob, I don't think the 'synergy magic comes from first order electical filters.... at all.
They help, no doubt. But look on Danleys spec sheets and you'll see 2nd order acoustical is what's achieved.. at best.
The synergy magic ime/ime, is the tight 1/4 wave integration of driver sections. Acoustic close-coupling is truly magic...bless Tom D.
If one is willing to go to the work/expense of utilizing fully complimentary linear phase xovers, hell yeah !
But the acoustic design, with 1/4 wl spacing is so elegant, if one wants to stick with IIR xovers, it works wonderfully too..
No way to loose.
And I'll repeat my vote to follow Chris if you want to stick with pure IIR, low order. He knows that path well i think...
Lord knows I don't know the low order IIR path....
Last edited:
Hi Mark,
Reading Tom D's posts it seems the synergy speakers work by pushing the drivers together using acoustic 4th order slopes. These slopes are achieved without using 'conventional named' slopes, ie slopes that introduce phase shift through the xo region. This agrees with Chris's posts regarding 1st order slopes.
Rob.
Reading Tom D's posts it seems the synergy speakers work by pushing the drivers together using acoustic 4th order slopes. These slopes are achieved without using 'conventional named' slopes, ie slopes that introduce phase shift through the xo region. This agrees with Chris's posts regarding 1st order slopes.
Rob.
Over the range of 50Hz to 18kHz, the DSL SH-50 has smooth phase response, but over 600 degrees of phase shift.
As Mark mentioned, definitely not 1st order slopes, acoustically or electrically.
Art
As Mark mentioned, definitely not 1st order slopes, acoustically or electrically.
Not that I'd want to use first order IIR crossovers, but did you try switching the mid polarity when changing the crossover slope?
Art
I may be mistaken, BUT a phase-shift is inherent in a bas-reflex-design as you use in your SH50 clone. I don´t believe you can not flatten that out!
That didn't get tried yesterday
Just woken up here, so that will be on the days order, somewhere after morning coffee and before building the 2nd pyramid and mouth flares cutting.
I'm beginning to get the elegance of the design, in that with no X/O's at all, apart from a protect the HF hi-pass, there is beautiful synergy in the way the drivers handover to each other.
Really interesting discussions above - I'm learning a lot.
Will explore my real world X/O's more. Maybe try PEQs to assist instead of the XO's offered in Najda SW?
BW 6dB is the only 1st order offering.
FIR is for Further on up the road (Bobby "Blue" Bland)🎶
Interesting that Najda perhaps has some FIR legs - I knew there was some capability - just not Tapped Horn freqs, so never bothered.
I have not been able to acoustically dampen around the horn to mic when measuring.
Not going to take the horn that's up in my loft room outside to try measuring.
The 2nd one, when made, perhaps.
I'm going to put black flight case handles on them
While the pyramid was glueing, played around some more.
Tried X/O free with PEQs to take care of too much over play where present.
Time aligning the MF to HF brought the phase hump in that region down a bit.
Inverting the MF driver phase messed it up completely🙂
Current state of play - wife calling me to make lunch so have to leave it at that and think.
Got a drop between LF and MF though 🙂
I will try a higher X/O freq on the HF next.
Progress though! It's soundingc better end better.
Tried X/O free with PEQs to take care of too much over play where present.
Time aligning the MF to HF brought the phase hump in that region down a bit.
Inverting the MF driver phase messed it up completely🙂
Current state of play - wife calling me to make lunch so have to leave it at that and think.
Got a drop between LF and MF though 🙂
I will try a higher X/O freq on the HF next.
Progress though! It's soundingc better end better.
Could you post a link to this? I've not read that.
My experience with my SH-50 indicates only first order electrical slopes (i.e., not fourth order) as evidenced by the induced phase shifts of the electrical filters through crossovers, combined with the natural acoustic rolloffs combine to achieve something more like ~10 dB/octave total slope--which is nothing like 24 dB/octave that you would get with a 4th order crossing. There are considerable frequency overlaps between drivers in their transition bands.
The crossings even include the rise in response of the woofers and midrange between their first and second notch frequencies--which was a big surprise to me when I measured it. The crossover design actually looks more like a phase link design with a very narrow midrange passband of just over an octave, but with considerable attention paid to the phase responses across the transition bands.
I think if others measured an SH-50 directly (like me) to see the effects of the separate acoustic responses without cross electric filters and then with the electrical filters in place, they would be talking about other subjects than higher order electrical filters. The reason why one uses higher order filters in multiway loudspeakers is basically not present, so worrying about minimizing the width of the crossover interference bands is not relevant in an MEH which experiences "unity summation aperture" of its multiple drivers. This is the point the Danley wrote at length about in both relevant patents: US6411718 Unity patent and especially in US 8284976 Synergy patent.
MEHs aren't multiple-aperture horn loudspeakers and therefore do not behave in the same fashion. To treat them like multiple-aperture horn loudspeakers using higher order filters is, I believe, a real mistake. Smooth phase and consistent polar response is what they bring to the table over conventional multiple aperture horn loudspeakers.
Chris
Try removing the low-pass crossover filters while retaining the first order high-pass filters you have present, then trim off the low-frequency response of the higher frequency drivers in each case via PEQs until flat amplitude response is achieved across the crossover transition bands. This will ensure that the drivers are within 45 degrees of time alignment in their crossover bands.
Also, if you have enough remaining output channel PEQs (biquads), add attenuating PEQs above the first notch frequencies of the woofers and midranges to attenuate their rise in response above the first notch frequencies. I found that I had to add attenuating PEQs above both the first notch and second notch frequencies.
Chris
Last edited:
We had a discussion about this in 2018 . When I said 4th order acoustic slopes in the post above I did not mean 4th order electrical.
I'll have a look at my mids measurements later, I'm sure they have a natural 4th order slope, or close to it.
If you look at the image I posted in post #21 regarding crossing my 4550 CD you will see a 1st order highpass at 6k , with some PEQ's to get a 4th order acoustic roll off at xo. This kind of agrees with your 'use a 1st order highpass and some PEQ' suggestion in post 32.
Rob.
edit : my mids have approx 6th order acoustic slope at their low end (250hz and below), with approx 3rd order slope above 1k. Add a 1st order electrical low pass at 1k to get a 4th order acoustic slope. I use PEQ on the low end to turn the 6th order slope into a 4th order slope, crossing closer to 350Hz. So my mid to tweeter slopes are both 4th order acoustic slopes, using 1st order electrical xo's to get them.
Last edited:
Thanks for the explanations and clarifications, very much appreciated.
All low pass off.
Worked away on PEQs (not sure Najda has a limit, I'm only at 45% processor load set on 96KHz).
Some small changes had a quite big effect on that line esp the steep rise at the very high frequencies.
No matter what I do, the dip bottoming out at 3KHz is stubbornly persistent.
Didn't have time for a go at the LF rise this evening.
Seeing changed PEQ settings and the subtle effect they have on phase is new to me.
The overall sound esp, HF is now much more refined. Giving the other side a run for its money!
That's my take too on the SH-50 too, without having one to dissect.
Looks like something a bit less than 2nd order acoustic is achieved with each of the two xovers in play. Nice.
Yep, from the unity/synergies I've build that use bridging mids, I've seen that rising low (woofer) response too.
I knock that rising response down with a steep xover, because ime it's just an unwanted anomaly.
Ok, here's where we part ways. I think you are over-analyzing the patent, and overextending the SH-50 design to all unity/synergys.
For a number of reasons.
*Nobody's sure why the later 'synergy' patent (vs the 'unity' patent) was granted, or what the heck the difference is with the 'unity' patent.
*'Unity summation aperture' effectiveness is strongly a function of the horns pattern.
It's pretty easy to get 1/4 WL summations with narrow patterns horns like the SH-50, or say a 60x40. Get's a lot more difficult as pattern widens.
* What the 1/4 summation difficulty for wider patterns means, is that the easy-peasy low-order acoustic xover slopes that narrow horns naturally allow, get tough to achieve.
* The 1/4 WL summations occur only within the physical geometries dictated ....... low order acoustic xovers can easily exceed the bandwidth of 1/4WL summations.
And I think it's a mistake not to see that multiple-aperture horn speakers are still just speakers......
Nothing magic/sacred about them. We are free to optimize their acoustic advantages as best we can.
Frankly, I think that not realizing higher order linear-phase xovers can help, is a mistake....and leaving some acoustic goodness on the table.
I doubt you will give that statement any creedence...but let me say this.
Take your SH-50 or your K402 build outdoors, or anywhere you can make acoustic far-field measurements relatively free of reflections.
And ditch Psych smoothing.
Try out both low order and higher order xovers that all achieve acoustic complementary response. See which sets of polars you like best.
I most honestly would love to hear your findings.
Last edited:
With at least 10 inductors and 9 capacitors in the SH-50 passive crossover, there is a lot to dissect.
Hefty.
YMMV, but as I see it, the 'trick' to the Synergy Horns is that the low pass on the midrange and the woofer adds a delay, and this delay allows you to get the midrange in phase with the tweeter even though the midrange is physically ahead of the tweeter.
For instance, the midrange taps in a Danley SH50 are 89mm ahead of the tweeter. 1910Hz is 178mm long. Therefore, if you put a midrange 89mm AHEAD of a tweeter, it will be out of phase with the tweeter. (89mm is one half wavelength of 1910Hz.)
You could flip the polarity on the midrange, and get them in phase. But a better option is to tweak the pathlength and the crossover slope between the midrange and the tweeter, until the two of them are in-phase.
The key to all of this is the location of the midrange and the low pass slope on the midrange. As you increase the slope of the low pass crossover, you get more and more delay at the xover point. IE:
1) if you use a first order high pass and a first order low pass, the mid and the tweet basically need to be right on top of each other (coaxial), similar to the Thiel and Vandersteen and Dunlavy speakers from the 80s and 90s
2) if you use a second order low pass, you can bring the midrange forward a little bit, but not a lot
3) the 'sweet spot' is arguably around a 3rd order low pass, IMHO
Note that it's the ACOUSTIC delay that's important here. It is possible to make all of this work using plain ol' 1st order (electrical) filters because the midrange and the tweeter are rolling off already.
Anyways, IMHO, this is the "Synergy Magic" and I personally think that a lot of MEH horns that LOOK like Synergy Horns don't SOUND like Synergy Horns because they don't have that Synergy Crossover Magic.
I found the 2012 thread link to that Danley quote: https://www.diyaudio.com/community/...-drivers-for-music.114481/page-6#post-2964213
Tom has a different way of looking at the problem than I have used. I think he approaches from the standpoint of passive electrical filters and having to keep the input impedance of the passive networks electrically above a certain threshold (nominally 2-4 ohms) in order to keep the amplifier driving the whole network in mono-amping mode from self-destructing (i.e., the amp effectively seeing a short circuit at some frequencies). So his explanations always includes that extra baggage of monoamping passive multiway loudspeakers and having to simultaneously pay attention to minimum electrical impedances. This makes the problem space more than twice as complicated conceptually than DSP multi-amping.
Using a multi-amped DSP active system simplifies the way to look at the problem--because one no longer has to worry about the stack-up of electrical input impedances falling too low at any given frequency. Each driver's electrical circuit is separate (or identical drivers for each "way" of the loudspeaker), and no passive electrical components are used between the DSP output terminals and the driver's input terminals. So the problems of electrical reactances in the electrical circuits are eliminated (except the internal electrical/acoustic reactance properties of the acoustic drivers themselves). This makes the job much easier for each amplifier channel. Stability is effectively assured and potential electrical/acoustic oscillation problems are avoided altogether.
Tom also uses the phrase "mass corner" quite a lot to signify either the upper roll-off frequency of a driver or drivers or the breakpoint of the lower roll-off frequency. I always have to translate these terms into something that makes sense for me to intuitively feel. Using these terms always manage to throw me off. For me, these are nonstandard terms. (YMMV.)
I believe what Tom D. is doing is the equivalent of setting the crossover basic digital filter representation parameters (like Butterworth, Linkwitz-Riley, Bessel, Gaussian, etc.) to filter design values that do not match the equivalent bi-quad mathematical weights for any of the resulting filters used. Tom is doing this because the acoustic drivers are not ideal devices with ideal properties and benign acoustic loadings. He's compensating for the drivers' non-ideal electrical-acoustic transfer function properties (i.e., amplitude and phase) on a case-by-case basis. Hence custom "fractional order" analog electrical filters are apparently used in his crossovers.
Finally, Tom seems to indicate that first order electrical filters have no induced phase lags (high frequency to low frequency). This is actually incorrect: they induce 90 degrees of phase lag (asymptotically with frequency), but acoustic drivers in a typical MEH arrangement are spaced 90 degrees apart (acoustically) at their first notch frequencies, so Tom clearly views first order as time-aligned, and he ignores this first order lag in an MEH context. But this can be very confusing since the first order filters are actually inducing electrical lag on the lower frequency drivers. It's just that the first order filters induce just the right amount of phase lag for acoustic time/phase alignment if one electrically crosses at the acoustic first notch frequency.
First order filters still induce 90 degrees of phase lag...
Chris
Last edited: