Here are some measurements from yesterday. Not great, not terrible🙂 The first file has the raw responses of the AMT and dual mids, the second one is the second MEH with DE500 and 4 mids. It also contains comparison of the results at the listening place (with added low mid and sub - these were not really optimized, just added with LR24 crossover) I am still not sure about the polarity of the mids in the AMT MEH. When I inverted them, the FR was flatter, but it does not seem correct to me, I will need to double check the wiring.
So I can do whatever cutting with EQs out of the pass band. For the mids to get rid of the unwanted HF stuff, and for the HF to reduce the excursion/distortion.
So I can do whatever cutting with EQs out of the pass band. For the mids to get rid of the unwanted HF stuff, and for the HF to reduce the excursion/distortion.
Pelan J, I can't open them in REW 5.1, no screencap? (jpg)
Measurment at sweet spot are very messy, it's easyer to find driver breakups and standing wave peaks with this https://www.szynalski.com/tone-generator/
Measurment at sweet spot are very messy, it's easyer to find driver breakups and standing wave peaks with this https://www.szynalski.com/tone-generator/
It was saved in the latest REW, that is why it cannot be opened in 5.1. The measured data were EQd and first order crossover added, so screenshots would provide only limited information. I still seem to miss the big picture, but I feel like I am getting there.
Then you can keep asking questions if this is the case...even "big picture" questions...😉
There are a lot of strong opinions of "how to do it", as you can see. One of the things that seems to create an artificial divergence in opinions is that people are really talking past each other, making different assumptions as to the application domain.
One of the big assumptions that many seem to make is that you're trying to dial-in a direct-radiating loudspeaker (i.e., in the case of this thread). So they come forward with their "baffle step", "driver distortion", and VituixCAD simulation concerns and process steps, when you are apparently focused on a multiple entry horn (MEH) based off an AMT-1 for the high frequency driver. It can make a big difference in how someone approaches DSP crossover dial-ins and even using REW measurement data if these little bits of information are not made clear to the contributors.
Since I don't deal with direct radiating drivers/loudspeakers, I don't have much use for those bits of information that are only useful for direct radiating drivers.
Chris
Yes, this is mostly about how to make a DSP based crossover for 2 and (future) 3 way MEHs, based on AMT and also standard compression drivers. I am playing around with the measurements I made and will produce some questions soon. I find the EQ tool in REW pretty handy and can get a flattish FR pretty quick. I still need to figure out the proper level adjustment and crossover slope adjustment. I think I can create target responses in Rephase and then I will need some help on "how and why". I think the best would be if I try to describe the process as I understand it and get corrected where I make mistakes. The AMTSynHFandMF.mdat file will serve as the input data for my test process.
OK, so this is what I get - matched (not perfectly I see) by PEQ to target curves:
REW could pretty well do the HF part automatically, but the MF part I had to do by hand (and the match is not that good).
I think I should add some HP filters to the HF part, so that it is protected from low frequencies (this is something I do not get really - I think I need a HP filter there)
Adding these together plus EQ at the crossover point to compensate for the LR dip:
If I just convolute the target curves with the responses EQd flat, this is what I get (and that is what I did usually get before I got into this):
The FR does not look that bad, but it seems that the IR is a bit worse here.
The MDAT file with all the manipulations is here: https://drive.google.com/file/d/19ypXVqEhqKrf4tJo-USs0PYi3IWwQ17M/view?usp=sharing
Both results look comparable to me, phase flat, but I am sure this can be done better...
Also, I am not sure how the trace arithmetic works and how does it take delays into account. If I look at the measurements correctly, I see the delay between HF and MF is somewhere around 0.02-0.04 ms (two lowest settings in my DSP).
The predicted and measured responses in REW are actually pretty close in my experience. What am I really struggling with is how to match the FR to the target (what filters to use) and which target curves to choose and why.
REW could pretty well do the HF part automatically, but the MF part I had to do by hand (and the match is not that good).
I think I should add some HP filters to the HF part, so that it is protected from low frequencies (this is something I do not get really - I think I need a HP filter there)
Adding these together plus EQ at the crossover point to compensate for the LR dip:
If I just convolute the target curves with the responses EQd flat, this is what I get (and that is what I did usually get before I got into this):
The FR does not look that bad, but it seems that the IR is a bit worse here.
The MDAT file with all the manipulations is here: https://drive.google.com/file/d/19ypXVqEhqKrf4tJo-USs0PYi3IWwQ17M/view?usp=sharing
Both results look comparable to me, phase flat, but I am sure this can be done better...
Also, I am not sure how the trace arithmetic works and how does it take delays into account. If I look at the measurements correctly, I see the delay between HF and MF is somewhere around 0.02-0.04 ms (two lowest settings in my DSP).
The predicted and measured responses in REW are actually pretty close in my experience. What am I really struggling with is how to match the FR to the target (what filters to use) and which target curves to choose and why.
Hi Pelanj, hi All
The following article by late Steen Duelund maybe gives a hint of what you struggle with? It speaks of the relationship af phase to the target-curve... Just read it (some times) to grab the idea that all drivers need to be "in phase", i.e. play synchronious, to produce a coherent impulse. Fundamental and harmonics at the right time!
Steen Duelund´s idea/filter is derived from B&O´s fillerdrive-paper. It is the acoustic slope of the FR that determines the phase-response, due to minimum-phase relationship. If I did understand him correct, the individual crossoverpoints are NOT independent of each other, they "interact" and need to be seen as a whole! That is especially the case for the mid-driver (3 way system), where the 12 dB HP and 12 dB LP makes for a total phase-rotation for the mid-driver of 360 degrees (24dB/oct filter). Hence the LP on the woofer section needs to be a 24dB/oct filter with the same phase-rotation as for the mid-section to play "In phase/syncronious". The same goes for the HP for the tweeter (24 dB/oct). The phase-tracks for the three drivers must be exactly the same! If you overlay them you will only see one trace! Note, these 24 dB/oct filters are NOT Linkwitz/Riley or Butterworth but specially derived filters/slopes/target-curves.
https://duelundaudio.com/wp-content/uploads/sites/3621/2013/12/duelund-filter.pdf
A friend of mine who was involved with the development the Duelund-syncroniuos-filter says it (the loudspeaker) can reproduce a square-wave-signal when done right! As can a SH50! So I think there is something here!?
One of the things, that is VERY DIFFERENT in a MEH from a conventional loudspeaker, WMT or WMTMW is the placement of the drivers!
In a WMT or WMTMW the drivers are stagged on top of each other vertically, maybe offset a bit for time-alignment, i.e. all the acoustic centers on a vertical line or arc in respect to the listening position.
BUT that is NOT so in a MEH, where all the drivers are positioned around a horizontal axial line, with the Tweeter (T) furthest away and then the Mid´s (M) contributing through their portholes (Bandpass!) further out and lastly eventually a couple af Woofers contributing through another set of portholes even nearer to the listener/horn-mouth. This "horizontal/axial driver-offset" combined with the individual phase-tracks resulting from their individual acoustical slopes is a balancing act, that Mr. Tom Danley has mastered to perfection with his "no named" passive filters.
It is a completely different setup, as I see it, and I think that is also what Chriss is addressing in his last post. The summation of the drivers in a MEH is something quite different from the summation in a "normal" direct radiating Loudspeaker!
As said, I have still not understood what happens, I am not an engineer, but hopefully one day will understand. There is something right with Chriss´s approach to MEH´s as long as you stay in the IIR-domain. I know that Chriss has taken a very close look at a Synergy-horn. There is a lot more to discover about that on the klipsch-forum.
I believe, that the approach of Mark100 to crossover´s in MEH´s gives "similar" results in respect to coherency, but that is in the FIR-domain, that takes DSP a step further, and introduces latency!
As of now I think I will try hard to do it in IIR, but if that fails I will go FIR with the help of an advanced program called Acourate by Audio Vero. I do have the hardware to go both routes.
I hope this is a helpful post, brings up some thoughts/ideas.
I have done my best to convey my thoughts, although may native tung is danish!
Best regards
Steffen
I too do STRUGGLE to understand how to do the DSP-crossover in a MEH too (IIR that is), and I am NOT an experienced expert (not yet anyways). But I try to give my two cents here:
The following article by late Steen Duelund maybe gives a hint of what you struggle with? It speaks of the relationship af phase to the target-curve... Just read it (some times) to grab the idea that all drivers need to be "in phase", i.e. play synchronious, to produce a coherent impulse. Fundamental and harmonics at the right time!
Steen Duelund´s idea/filter is derived from B&O´s fillerdrive-paper. It is the acoustic slope of the FR that determines the phase-response, due to minimum-phase relationship. If I did understand him correct, the individual crossoverpoints are NOT independent of each other, they "interact" and need to be seen as a whole! That is especially the case for the mid-driver (3 way system), where the 12 dB HP and 12 dB LP makes for a total phase-rotation for the mid-driver of 360 degrees (24dB/oct filter). Hence the LP on the woofer section needs to be a 24dB/oct filter with the same phase-rotation as for the mid-section to play "In phase/syncronious". The same goes for the HP for the tweeter (24 dB/oct). The phase-tracks for the three drivers must be exactly the same! If you overlay them you will only see one trace! Note, these 24 dB/oct filters are NOT Linkwitz/Riley or Butterworth but specially derived filters/slopes/target-curves.
https://duelundaudio.com/wp-content/uploads/sites/3621/2013/12/duelund-filter.pdf
A friend of mine who was involved with the development the Duelund-syncroniuos-filter says it (the loudspeaker) can reproduce a square-wave-signal when done right! As can a SH50! So I think there is something here!?
One of the things, that is VERY DIFFERENT in a MEH from a conventional loudspeaker, WMT or WMTMW is the placement of the drivers!
In a WMT or WMTMW the drivers are stagged on top of each other vertically, maybe offset a bit for time-alignment, i.e. all the acoustic centers on a vertical line or arc in respect to the listening position.
BUT that is NOT so in a MEH, where all the drivers are positioned around a horizontal axial line, with the Tweeter (T) furthest away and then the Mid´s (M) contributing through their portholes (Bandpass!) further out and lastly eventually a couple af Woofers contributing through another set of portholes even nearer to the listener/horn-mouth. This "horizontal/axial driver-offset" combined with the individual phase-tracks resulting from their individual acoustical slopes is a balancing act, that Mr. Tom Danley has mastered to perfection with his "no named" passive filters.
It is a completely different setup, as I see it, and I think that is also what Chriss is addressing in his last post. The summation of the drivers in a MEH is something quite different from the summation in a "normal" direct radiating Loudspeaker!
As said, I have still not understood what happens, I am not an engineer, but hopefully one day will understand. There is something right with Chriss´s approach to MEH´s as long as you stay in the IIR-domain. I know that Chriss has taken a very close look at a Synergy-horn. There is a lot more to discover about that on the klipsch-forum.
I believe, that the approach of Mark100 to crossover´s in MEH´s gives "similar" results in respect to coherency, but that is in the FIR-domain, that takes DSP a step further, and introduces latency!
As of now I think I will try hard to do it in IIR, but if that fails I will go FIR with the help of an advanced program called Acourate by Audio Vero. I do have the hardware to go both routes.
I hope this is a helpful post, brings up some thoughts/ideas.
I have done my best to convey my thoughts, although may native tung is danish!
Best regards
Steffen
Last edited:
Hi guys,
first a quick reply to Steffen's post.
I don't see the approach of in & out of band flattening, then applying complementary xovers as particular to either IIR or FIR processing. Really doesn't matter, same for both.
I also don't think summation in a MEH is much different than in any traditional design, other than reflections from driver sections bouncing back from the throat, and perhaps a little more driver to driver interaction.
MEH's do help with setting delays however, as they take advantage of front to back spacing of driver segments to help tie driver sections together. Typically, a CD (or AMT) needs delay vs the lower drivers in any type build. A MEH gives some physical spacing, head start, in providing those delays.
So for me, there's no difference in how i go about measuring and processing a MEH, vs a MT, MTM, etc, or even a line array or the dodecahedron i built.
I use FIR mainly because it's so much easier to get good mag, phase, and impulse, than with using IIR.
I find this ease doubly true using rather straightforward automation like FirDesigner software.
I've tried IIR on about everything, including some MEHs....it's just so much more work achieving the same clean traces and impulse responses
given the need to keep IIR xover orders low, when compared to using complementary linear phase xovers,
Anyway, back to mr pelanj....
Seems like you're making some good progress.
If i may suggest....after you have applied driver flattening EQs to the AMT and mids, measure both sections from same mic placement.
Assuming that's what you did, and the measurements you posted were from the same mic placement,..... the low driver was 3.1791ms relative to loopback, and the AMT was 2.6889ms.
That difference of 0.49ms (3.18-2.69), is the delay needed in the processor for the AMT. Put that in place.
Then, remeasure from same spot, and put the mid delay of 3.1791 you measured into the Make a measurement Timing Offset:
This will remove the myriad phase wraps from the mid measurement and you should read very close to zero delay relative to loopback,
And with the 0.49ms delay in the processor for the AMT, you will also see it's phase trace excess wraps removed, with a close to zero time relative to loop back
Doing that will give phase traces (after 15 cycle FDW and 1/6th smoothing for clarity) that look like this, traces that make sense and have visual time alignment capability.
Note how the traces overlap around 800-1100Hz. This gets back to what Charlie was saying....phase trace overlay thru xover region is the name of the game.
If you want to widen the range of overlap, play with a series of shelving filters out of band for each section.
Once you are happy with the range of overlap, put xovers in place.
You should need very little, if any timing adjustment between sections if using complementary xovers (either IIR or lin phase FIR).
Just get the phase traces to overlay...you can find any small timing adjustments needed to do so using the Offset t=0 slider that pops up.
Agg their effect to the delay you have in place in the processor for the AMT.
There no need to go to trace arithmetic imho, if anything it makes it harder ime.
Hope all that was clear......hope it helps 🙂
first a quick reply to Steffen's post.
I don't see the approach of in & out of band flattening, then applying complementary xovers as particular to either IIR or FIR processing. Really doesn't matter, same for both.
I also don't think summation in a MEH is much different than in any traditional design, other than reflections from driver sections bouncing back from the throat, and perhaps a little more driver to driver interaction.
MEH's do help with setting delays however, as they take advantage of front to back spacing of driver segments to help tie driver sections together. Typically, a CD (or AMT) needs delay vs the lower drivers in any type build. A MEH gives some physical spacing, head start, in providing those delays.
So for me, there's no difference in how i go about measuring and processing a MEH, vs a MT, MTM, etc, or even a line array or the dodecahedron i built.
I use FIR mainly because it's so much easier to get good mag, phase, and impulse, than with using IIR.
I find this ease doubly true using rather straightforward automation like FirDesigner software.
I've tried IIR on about everything, including some MEHs....it's just so much more work achieving the same clean traces and impulse responses
given the need to keep IIR xover orders low, when compared to using complementary linear phase xovers,
Anyway, back to mr pelanj....
Seems like you're making some good progress.
If i may suggest....after you have applied driver flattening EQs to the AMT and mids, measure both sections from same mic placement.
Assuming that's what you did, and the measurements you posted were from the same mic placement,..... the low driver was 3.1791ms relative to loopback, and the AMT was 2.6889ms.
That difference of 0.49ms (3.18-2.69), is the delay needed in the processor for the AMT. Put that in place.
Then, remeasure from same spot, and put the mid delay of 3.1791 you measured into the Make a measurement Timing Offset:
This will remove the myriad phase wraps from the mid measurement and you should read very close to zero delay relative to loopback,
And with the 0.49ms delay in the processor for the AMT, you will also see it's phase trace excess wraps removed, with a close to zero time relative to loop back
Doing that will give phase traces (after 15 cycle FDW and 1/6th smoothing for clarity) that look like this, traces that make sense and have visual time alignment capability.
Note how the traces overlap around 800-1100Hz. This gets back to what Charlie was saying....phase trace overlay thru xover region is the name of the game.
If you want to widen the range of overlap, play with a series of shelving filters out of band for each section.
Once you are happy with the range of overlap, put xovers in place.
You should need very little, if any timing adjustment between sections if using complementary xovers (either IIR or lin phase FIR).
Just get the phase traces to overlay...you can find any small timing adjustments needed to do so using the Offset t=0 slider that pops up.
Agg their effect to the delay you have in place in the processor for the AMT.
There no need to go to trace arithmetic imho, if anything it makes it harder ime.
Hope all that was clear......hope it helps 🙂
Steffen, thanks for the link, I will have a look at it. I have spent quite some time around Svendborg since I have been working for a company located there for almost 15 years. It was actually a Danish colleague who got me into HiFi and horns, I was blown away by his Jealvox clone system. And some of my better speakers were bought in Denmark. In CZ, there is not much of the vintage stuff I like.
Mark, this is so great, thanks a lot, these are the steps I have missed till now - removing the offset to get clean phase readings. I will look at this when I have again free time for measurements. I have one question though - on the timing. It seems really counter intuitive to me that delay on the HF is needed when it is actually further away from the mic. If you have a look at the IRs in my file - do you have a feeling that REW did get the times right? On the HF, it seems fine, but the reference time on the MF is quite away from the beginning on the pulse and is on the second peak and below zero. Do the polarities also look the same for HF and MF to you? When I align the beginning of the impulses, I actually get that there should be a delay of 0.02 ms on the MF. So what is actually correct? I guess the phase will tell? I am getting really close - all the things start clicking together for me now.
Mark, this is so great, thanks a lot, these are the steps I have missed till now - removing the offset to get clean phase readings. I will look at this when I have again free time for measurements. I have one question though - on the timing. It seems really counter intuitive to me that delay on the HF is needed when it is actually further away from the mic. If you have a look at the IRs in my file - do you have a feeling that REW did get the times right? On the HF, it seems fine, but the reference time on the MF is quite away from the beginning on the pulse and is on the second peak and below zero. Do the polarities also look the same for HF and MF to you? When I align the beginning of the impulses, I actually get that there should be a delay of 0.02 ms on the MF. So what is actually correct? I guess the phase will tell? I am getting really close - all the things start clicking together for me now.
Hi Mark
Well, now I am leaning really far out of the window, knowing I am on deep deep water. Maybe I make a complete idiot out of my self!?
I am not sure, but I don´t think, that Tom Danley uses crossover-filters with their inherent phase-rotation! At least not anything steep like 24 dB/oct or more.
This is something that I have kind of read between the lines in some of Tom Danley´s mystical posts years ago here on diyAudio. But I would not be capable of pointing them out.
It´s about shaping the response-curve of the individual pass-bands (CD, mids and woofers) and using the inherent phase-rotation/delay of the Bandpass-enclosure of the mids and woofers to "time-align" all three passbands to sum coherently. He does that passively, with no possibility to use delay!
I have this hint that there is something to that, as there is not much global phase-rotation if you look at the phase-trace of the SH50.
Maybe I am very wrong but so be it. I must say that I have reached the limits of, what I am capable of explaining/putting in words. My knowledge also reaches limits.
Steffen
Well, now I am leaning really far out of the window, knowing I am on deep deep water. Maybe I make a complete idiot out of my self!?
I am not sure, but I don´t think, that Tom Danley uses crossover-filters with their inherent phase-rotation! At least not anything steep like 24 dB/oct or more.
This is something that I have kind of read between the lines in some of Tom Danley´s mystical posts years ago here on diyAudio. But I would not be capable of pointing them out.
It´s about shaping the response-curve of the individual pass-bands (CD, mids and woofers) and using the inherent phase-rotation/delay of the Bandpass-enclosure of the mids and woofers to "time-align" all three passbands to sum coherently. He does that passively, with no possibility to use delay!
I have this hint that there is something to that, as there is not much global phase-rotation if you look at the phase-trace of the SH50.
Maybe I am very wrong but so be it. I must say that I have reached the limits of, what I am capable of explaining/putting in words. My knowledge also reaches limits.
Steffen
Great, glad it's help .
Yes, it is counter intuitive......took me a long time to learn/realize that anytime there is an acoustic low-pass in place, it measures as a longer TOF (delay vs loopback) than when one is not in place. The lower the frequency of the low pass, the longer the TOF will be.
As an aside, acoustic high-pass does not add TOF until above around 2kHz, and then it adds maybe 0.03ms, rising up to 0.1-2ms all the way up at 15kHz or so...iow, high-pass is no big deal, where low-pass is.
Whenever you see a flat phase trace in the meat of the pass-band, and there is zero delay vs loopback, timing is correct.
If the flat phase trace is at 0 degrees, polarity is correct. At 180 degrees, obviously inverted.
Yes, it is counter intuitive......took me a long time to learn/realize that anytime there is an acoustic low-pass in place, it measures as a longer TOF (delay vs loopback) than when one is not in place. The lower the frequency of the low pass, the longer the TOF will be.
As an aside, acoustic high-pass does not add TOF until above around 2kHz, and then it adds maybe 0.03ms, rising up to 0.1-2ms all the way up at 15kHz or so...iow, high-pass is no big deal, where low-pass is.
Whenever you see a flat phase trace in the meat of the pass-band, and there is zero delay vs loopback, timing is correct.
If the flat phase trace is at 0 degrees, polarity is correct. At 180 degrees, obviously inverted.
Impulse alignment is a bitch and often misleading.....my advice is to stick with phase overlay....which will never let you down.
This is what I found in the Duelund paper.
If this is the correct alignment, then the impulses should be aligned with their beginnings (if I understand things correctly). I still have a little doubt about the timing found by REW, but will ignore it and the phase will tell🙂
If this is the correct alignment, then the impulses should be aligned with their beginnings (if I understand things correctly). I still have a little doubt about the timing found by REW, but will ignore it and the phase will tell🙂
Hi Steffen,
his filters have rotation, he just keeps it as little as possible.
You can see the rotation in the SH-50's spec sheet.....look closely at the phase scale on the right.
Yes, the delay initiated by an acoustic low pass of the driver sections, mids, and lows, is offset by the mids and lows being physically closer to the listener than the CD is. That delay is also offset/adjusted by the "net filters" that perform the xover duties. (the acoustic low-pass delay is what i was just posting about to pelanj).
And yes, without possibility of delay. I remember TD saying one of their boxes, the SM60f i think, keeps the CD about an inch away from optimal depth.
It's worth nothing, the larger the pattern, H&V in degrees, the less the physical depth ports to CD, given the same distance on the horn walls.
It's brilliant passive xover and driver port placement design, designed to help keep phase rotation low.
That said, DSL's main market is install, which lends itself to passives better. Without that market constraint, there's no question in my mind that even less phase rotation( like zero), and even smoother magnitude is available with FIR processing.
What's your take on the global rotation as shown on the spec sheet?
The big thing in my mind with MEHs, or any other speaker.....is in the end all we are doing is stitching together passbands and geometries.
The passbands may have more complicated interactions from geometries as in the case of MEHs, but they all still all abide by the same stitching rules .
That's a good example of why i think impulse alignment is tricky.
Because it's dang hard to visually nail down exactly when those impulses start.. and gets progressively more uncertain the lower the passband.
Steffen,
The phase-rotation of the DSL SH-50 over it's rated -3dB pass band is 720 degrees, equivalent to 8 "poles" (6dB crossover filters) or two "trips around the globe" .
Some of that phase rotation is inherent in the bass-reflex ("phase inversion") alignment, Fb (box tuning) at around 90Hz.
Reproducing a reasonable facsimile of a square-wave-signal requires smooth, not flat phase response.
Not to dismiss the ultimate sonic superiority of FIR over IIR filters, but I can't hear any difference between a flat frequency response loudspeaker with 720 degrees of smooth phase rotation through it's pass band compared to one with with no phase rotation.
That said, properly aligned coaxial or MEH speakers have the potential advantage of maintaining phase and frequency response off-axis in both horizontal and vertical axis.
Art
Hi Mark
Well, as I said I am at my limits of knowledge. And when I wrote my comment about global phase-rotation of the SH50, I actually came to remember the scale-thing on the right side of the official plot!
But it is stil only a 360 degrees rotation from ~ 85 Hz to ~ 15 kHz, and a pretty "straight" line. Maybe that is part of the "good story" and capability to perform square- waves through the crossover-regions?
A three-way-Duelund would have a 360 degrees "global" phase-rotation also. (He always opted for sealed bass enclosures, that dont have that big phase-growth in the sub octaves).
I still fight with late effects of a stress-breakdown last spring, like having constant hangovers and difficulty concentrating! So things progress veeeeery slowly her. I have collected all drivers now. Last week it was my turn to hit Corona, just as I felt ready to progress with my project! Well, well, well. I just have to accept that it is that way and be patient. I can´t find a way or good reason to abandon or give up the project so I have to fight my way through. It´s really a passion/obsession!
Steffen
Well, as I said I am at my limits of knowledge. And when I wrote my comment about global phase-rotation of the SH50, I actually came to remember the scale-thing on the right side of the official plot!
But it is stil only a 360 degrees rotation from ~ 85 Hz to ~ 15 kHz, and a pretty "straight" line. Maybe that is part of the "good story" and capability to perform square- waves through the crossover-regions?
A three-way-Duelund would have a 360 degrees "global" phase-rotation also. (He always opted for sealed bass enclosures, that dont have that big phase-growth in the sub octaves).
I still fight with late effects of a stress-breakdown last spring, like having constant hangovers and difficulty concentrating! So things progress veeeeery slowly her. I have collected all drivers now. Last week it was my turn to hit Corona, just as I felt ready to progress with my project! Well, well, well. I just have to accept that it is that way and be patient. I can´t find a way or good reason to abandon or give up the project so I have to fight my way through. It´s really a passion/obsession!
Steffen
Thanks for your comment Art. And actually thank you all for answering questions and ideas, trying to wrap my head around these things. 🙂
Steffen
Steffen
Hi Steffen, you give me good motivation to check stuff sometimes 🙂
I just ran a 500Hz square wave through a LR 24dB/oct xover at 500Hz, both sides summed back together. So 360 degrees of rotation.
Here's one half the wave with the xover bypassed...iow straight wire throughput....to check for test setup integrity.
And here is with the xover in play, with it's 360 degrees of phase rotation.
Not so dang square, huh? !!!
I also checked to see if it's better or worse with xover freq and generator frequency apart. Like 1000Hz xover, 300Hz square wave. Doesn't matter, square waves fall apart with xover wherever.
Anybody can check this easily, if they have a processor and can sum the outputs back together, and use REW's signal generator and scope.
The underlying crossover is agnostic, so if someone mentions a passive crossover or a DSP crossover then they are no longer talking about the crossover, but about the filters. The goal is the same for all kinds.
Assigning any portion of an octave in itself is a line in the sand. This is why there is uncertainty here about what is required and the tail is wagging the dog. By not getting the expected results, some are opting for steeper filters without looking any further.
This is fine for practical purposes. The problem is the assumptions which are being made about what's necessary, based solely on what's available. There is no inherent reason this use of phase and symmetry forms the basis of what is needed once you move beyond filter theory and on to a real speaker.
Hi Mark
I am not sure that a 24 dB/oct Linkwitz/Riley complementary filter can do a square-wave! The slope af a Duelund 24dB LP on the woofer in a three-way system is different from a 24 dB/oct Linkwitz/Riley-slope!? But again my filter-knowledge is limited and I dont won´t to be argumentativ without being sure I am right. 🙂
Steffen
I am not sure that a 24 dB/oct Linkwitz/Riley complementary filter can do a square-wave! The slope af a Duelund 24dB LP on the woofer in a three-way system is different from a 24 dB/oct Linkwitz/Riley-slope!? But again my filter-knowledge is limited and I dont won´t to be argumentativ without being sure I am right. 🙂
Steffen