Most of them (full rangers) do not have that good of a STEP. Breakup of the cone being a good reason to start messing it up among other things.
It isn't easy to produce a good step, certainly not with the dynamics and SPL you'd want too. That triangle in Stereophile isn't even that good. Look at a measurement from a Dunlavy speaker at the listening position 3 or 4 meter away from the speaker instead of the usual close up measurements. That's where it starts to being good.
That's quite a bit better than the one in the article.
Chasing it myself with my line arrays, not totally there yet
I can produce a cleaner one with a different EQ target (mine is sloped down to HF) so the start is more like the triangle in the Dunlavy STEP. But this is where I started after correction with DRC. It will hopefully get better. But in listening it makes quite a difference to a crossed over speaker. And not only on a few songs either. But you've got to really want it, because it won't come easy if you want the SPL etc.
Attachments
Not sure I understood what you said, or asked, but here goes.
I'm not trying to phase equalize or time align anything, unless I see a big suckout in the frequency response where I sit to listen, in which case I'll switch the polarity on one of the drivers and measure again. I'm one of those people who doubts if I can hear a time alignment issue if it hasn't caused the FR suckout, and it isn't between about 800HZ and 5kHZ, where the ear is most sensitive. I believe it to be a very small issue compared to many other issues in a speaker system.
The way I see time alignment issues is that they cause off-axis beaminess around the xover freq., which is undesirable, but depends on the acoustics of the listening room as to whether it's a big issue. Going 4 pole active at 7kHZ would obviously be way better on paper, but I'm not convinced it's such a big issue once you're above about 6kHZ.
To be honest, I just didn't want to have to have more poweramps. the 4 pole active xover at 500HZ I believe makes a big difference. Much more than at 7kHZ. I could be wrong.
Of course, break up. Engaging my mouth before brain again.
On that case, how does one even separate break up anomolies from time coherence issues?
Or to correct myself, breakups ARE time coherence issues, so how to separate them from filter induced ones?
I get that the mid polarity inversion in the article shows up rather obviously, but when we are talking sub 1ms discontinuity how audible is it?
Besides a plot, which can be interpreted differently and only.compared under identical conditions, how can it be quantified?
This is where I find difficulty in this (my step response experience is based in control theory, and here we are dealing with much more complex orders of transfer function wrt voltage)
BOB, my apologies I was likely just thinking aloud (heatwave today and my brain is fried)
I largely believe in what you said. I doubt the real audibility at the crossover point you use, and I would do the same. As I think Sreten has said, its not economical to use a power amp for each driver in a three way, and I agree, mostly. I also agree that if a problem is there in FR measurements or lobing polar then you have a clear indication of an issue, and if not then its debatable that it presents a real problem.
On that case, how does one even separate break up anomolies from time coherence issues?
Or to correct myself, breakups ARE time coherence issues, so how to separate them from filter induced ones?
I get that the mid polarity inversion in the article shows up rather obviously, but when we are talking sub 1ms discontinuity how audible is it?
Besides a plot, which can be interpreted differently and only.compared under identical conditions, how can it be quantified?
This is where I find difficulty in this (my step response experience is based in control theory, and here we are dealing with much more complex orders of transfer function wrt voltage)
BOB, my apologies I was likely just thinking aloud (heatwave today and my brain is fried)
I largely believe in what you said. I doubt the real audibility at the crossover point you use, and I would do the same. As I think Sreten has said, its not economical to use a power amp for each driver in a three way, and I agree, mostly. I also agree that if a problem is there in FR measurements or lobing polar then you have a clear indication of an issue, and if not then its debatable that it presents a real problem.
Last edited:
I wouldn't use a hard cone woofer with the usual substantial ringing peak in the upper mid freqs unless I was going to use a 4th order active crossover to push those upper mid peaks WAY down, at least 12dB.
You can do 2 things... deny it is audible and move on (like most do) or look into it. I've chosen the last path, not nearly finished, but based on past and present experiences it's definitely worth it to me. And frankly, I don't care if that makes me as strange as the highly ridiculed audiophile chasing perfection. I'm measuring my way there out of pure curiosity and it's working so far. More chills down my spine on familiar recordings than I've ever had.
So that would make me the one that says: yes, it's audible. But there sure are a lot of papers stating otherwise, you choose your journey.
One more thing: you don't need a perfect step for imaging etc. as long as both speakers have the same response imaging will still be good enough.
Last edited:
You gave me a good idea. Reminded me of my drivers laying around unused. Many woofer midrange that can go high, similar midrange like TC/TG (an Audax) and a tweeter capable of 1st order (a Dynaudio). But I want to use PLLXO
Here's what Duelund has to say related with your passive first order idea:
But you can use more drivers so bandwidth is very limited.
Some drivers might have 2nd order rolloffs. From what I've seen, the mechanical rolloff mechanisms are rarely perfectly spaced to provide a clean 2nd order rolloff curve. If you have a driver that is flat to 15kHZ (like the Peerless TG9 3 inch), and you roll it off with a one pole low pass (coil in series), at 7kHZ, you might be down by about 9dB at 15kHZ. The mechanical rolloff at 15kHZ will introduce it's own phase shift, which could cause addition at 7kHZ to be different enough that you'd get a flatter FR by reversing the polarity of the tweeter wires.
It isn't as perfect as it could be with a 4 pole active at 7kHZ, and yet another stereo poweramp, but it sounds real good to my ear. Above about 6kHZ, I don't think we are nearly as sensitive to time alignment issues. That's why I am thrilled to have found a midrange driver that allows me to keep the xover out of the 800HZ - 6kHZ range.
Wesayo,
I don't think you're crazy. But I understand less than some, more than others, and I'm a natural sceptic. My own active speakers are two ways and relative phase aligned, taking into account driver distance offset, but they are not absolute phase aligned due to use of analogue 4th order filters (approximately)
Is the cycle lag at 2khz audible? I cannot say, since I have not removed the cycle iffset in DSP.
Would it show in step response? I guess so, but how bad would it be? Id like to take my own impulse response, but I don't understand ALL the mechanisms, and I'm sceptical that one can without a very good measurement environment or methods.
But I am open to trying for sure
I don't think you're crazy. But I understand less than some, more than others, and I'm a natural sceptic. My own active speakers are two ways and relative phase aligned, taking into account driver distance offset, but they are not absolute phase aligned due to use of analogue 4th order filters (approximately)
Is the cycle lag at 2khz audible? I cannot say, since I have not removed the cycle iffset in DSP.
Would it show in step response? I guess so, but how bad would it be? Id like to take my own impulse response, but I don't understand ALL the mechanisms, and I'm sceptical that one can without a very good measurement environment or methods.
But I am open to trying for sure
I think that if you were in an anechoic chamber, you might hear a difference between time aligned and not, with a somehow perfect recording. In a real-world situation with room acoustics and program source wild cards, I think it's really just about whether or not the off axis beaming interacts with room acoustics in a way that's even worse than how the room screws it up anyway, even with perfect time alignment.
I suspect that the ear-brain mechanism (which I've studied a bit) analyzes the various frequency bands discreetly (every 1/6th of an octave I've heard), and then interprets them together, if that makes sense. As a survival based protection mechanism? For most of history, as our brains evolved, we always needed to instantly recognize the sound of a dangerous animal stalking us, regardless of what the acoustics of the environment did to the phase or time alignment response.
It would be interesting to create a time alignment error (with a DSP?) while listening through headphones. To switch that in and out. Especially with a binaural recording you made yourself, so you know what you've got. If you do this test with a recording you didn't make yourself, it may already be riddled with time alignment errors.
Last edited:
My measurements are taken in my living room that's far from ideal. But it still made it possible for me to get it better. I did chase first reflections with damping panels, as much as my spouse allowed me. The rest I do with FIR correction tools.
I'd love to have a better room but make do with what I got. A calibrated mic, a steady PC, REW and JRiver have gotten me quite far. I've learned much more over these past months than all the reading about it before that.
I could create much cleaner STEP responses with filtering, but it has to remail listenable, no better than that, convincing as well. So the more problems you solve without the help of the digital tools, the better the results get. Both speaker and room are of coarse big factors here. An impedance curve of your speaker in the enclosure is one of the first things to look at. It can tell you a lot to start with. A simple test, yet often ignored.
I've come to find out that every little detail seems to matter, what a surprise. Most commercial developers just seem to ignore the hard to get right parts, but that doesn't mean we have to.
I'd love to have a better room but make do with what I got. A calibrated mic, a steady PC, REW and JRiver have gotten me quite far. I've learned much more over these past months than all the reading about it before that.
I could create much cleaner STEP responses with filtering, but it has to remail listenable, no better than that, convincing as well. So the more problems you solve without the help of the digital tools, the better the results get. Both speaker and room are of coarse big factors here. An impedance curve of your speaker in the enclosure is one of the first things to look at. It can tell you a lot to start with. A simple test, yet often ignored.
I've come to find out that every little detail seems to matter, what a surprise
. Most commercial developers just seem to ignore the hard to get right parts, but that doesn't mean we have to.
A full range impedance curve is in itself difficult to take without a very very good DMM. Doable if you have a Fluke 28/29.
Ive done Z plots, generally up to 400 Hz, where the calibration accuracy of my DMM fails. Perhaps I'll borrow my work DMM, though I know ill find some trademark TL induced wiggles.
Ive done Z plots, generally up to 400 Hz, where the calibration accuracy of my DMM fails. Perhaps I'll borrow my work DMM, though I know ill find some trademark TL induced wiggles.
Ask yourself, would a "live" band in your room not sound live?
I think it would, which means that the loudspeaker should be able to mimic most of that. Think of stereo like this:
From: Moulton Laboratories :: The Brave New World: Loudspeakers to the Left of Us! Loudspeakers to the Right of Us!
Makes a lot of sense to me. Now what doesn't make sense to me is to let those first reflections be worse than they could be. We are already missing out due to not hearing the direct sound!
I'm doing it with REW and a resistor. Probably not ideal but it still revealed a lot to me. The free air plots do resemble the ones from Vifa (in my case) very closely though. It was merely one part on this long journey, but also very enlightening. It dictated the way I use damping in the (sealed) enclosure and how I build my (huge) baffle.
Last edited:
Ask yourself, would a "live" band in your room not sound live?
I think it would, which means that the loudspeaker should be able to mimic most of that. Think of stereo like this:
From: Moulton Laboratories :: The Brave New World: Loudspeakers to the Left of Us! Loudspeakers to the Right of Us!
Makes a lot of sense to me. Now what doesn't make sense to me is to let those first reflections be worse than they could be. We are already missing out due to not hearing the direct sound!
If a live band was in my living room,, one of the biggest differences would be that each instrument would create a unique room reflection "signature" in 3D, which doesn't seem possible with 2 speakers running stereo. So on our best day we will only be able to create an illusion that's close enough to be fun. Having said that, I very much admire your endurance and ambition. I'm always interested in what you have to say.
(way off topic now, perhaps I should sleep
Me to, it's late enough already and we're taking this way off topic. Enough ramblings for one night
.I guarantee you, fun it is! And as long as it keeps on getting better I'll work on it. The biggest problem will be knowing when to stop trying!
Last edited:
Waveguide... there, I am the first on this thread to say it. Waveguide..Waveguide..Waveguide..
A circular tweeter waveguide deep enough to time align the tweeter to the midrange should have more design value than a simple cut-down step. Even the 4" Midrange Illuminator 3kHz polar response is best mated to a waveguide tweeter.
The Kii all-digital 3-way speaker has caught my attention as the template for many China clones. Sharing Xover code instead of schematics. Easy time alignment.
http://kiiaudio.com/tech.html
A circular tweeter waveguide deep enough to time align the tweeter to the midrange should have more design value than a simple cut-down step. Even the 4" Midrange Illuminator 3kHz polar response is best mated to a waveguide tweeter.
The Kii all-digital 3-way speaker has caught my attention as the template for many China clones. Sharing Xover code instead of schematics. Easy time alignment.
http://kiiaudio.com/tech.html
Attachments
I my 2-way 10F/8424 FAST ref monitor I tried both a transient perfect BW1 xo and LR2 xo. You can hear the difference when it is transient perfect on some recordings - especially well recorded live-like mic'd records. I had too much woofer breakup leaking through though so went with LR2 in the end. But if I had a woofer with less breakup I would be at it again. The 10F or TC9FD is the ideal driver to do these wide band xo's. Wesayso is able to to do it because he has 50 of them to get enough Sd for real dynamics. It really is audible and you hear it as realism of a live band.
Time align
The difference between Time aligning & not, is a revelation. Until you've tried it you won't know how much more detail & depth & stereo information you've been missing.
The first time i had the opportunity to do this, many years ago, was with a nice Upper bass/mid 12" Xover to & quality horn/driver & then Xover to a HF unit. Bass was via 2 x 24" reflex cabs. I designed the Bass & Upper Bass cabs.
After hours of realigning the top 3 units to try & get the coils in line as much as possible, & playing many various assorted quality 12" records, it became more & more apparent what i was hearing. I had never heard those tunes sounding so good, or with the amount of info listed above.
6dB Xovers on the horn covering 1kHz - 7kHz.
That day was a long held dream come true The full stereo system was installed in a bar/disco. It was only a 2kW system, but more than loud enough for the venue without distorting. Nothing ever burnt out, & that's without a comp/limiter ! Even people who wern't into hifi/sound said how great it sounded
So if you get the chance, do it
The difference between Time aligning & not, is a revelation. Until you've tried it you won't know how much more detail & depth & stereo information you've been missing.
The first time i had the opportunity to do this, many years ago, was with a nice Upper bass/mid 12" Xover to & quality horn/driver & then Xover to a HF unit. Bass was via 2 x 24" reflex cabs. I designed the Bass & Upper Bass cabs.
After hours of realigning the top 3 units to try & get the coils in line as much as possible, & playing many various assorted quality 12" records, it became more & more apparent what i was hearing. I had never heard those tunes sounding so good, or with the amount of info listed above.
6dB Xovers on the horn covering 1kHz - 7kHz.
That day was a long held dream come true
The full stereo system was installed in a bar/disco. It was only a 2kW system, but more than loud enough for the venue without distorting. Nothing ever burnt out, & that's without a comp/limiter ! Even people who wern't into hifi/sound said how great it sounded So if you get the chance, do it
Why is impedance curve hard to measure if you have something like the Dayton DATS? I do it routinely and it takes 5 seconds. It doubles as an LCR Meyer to check your crossover components. They check out well when measuring known values. One of the most useful tools a DIY'er can own besides a calibrated microphone.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.