Is that true?
I think many manufacturers prefer good drivers capable of minimal crossovers than average drivers improved with complex crossover.
I was interested to find LR4 electrical or complex crossover in high end speakers from Stereophile measurement/website. As far as I can remember, I got only very few?
Jay,
It is in regard to ALL standard models of filter behaviours. These filter coefficients are just a mathematical exercise in definition of a filter.
As Juhazi rightly states, the acoustic end result is whats important. I found in my own experience that the right filter mostly does not fit into the mathematical standard definitions and is arbitrarily defined.
My own filters use a 5th order LP with broadly 4th order phase rotation, and a ripple response similar to a cheby 0.5 dB alignment, this crossed to a 4th order HP with protection capacitor which also doubles to align phase very very well (in simulation) to single figure degrees. Phase and amplitude are very good through the crossing point.
LR not required ��
There are many ways to skin the cat...
For the DSP guys, it's easy to dial in an alignment Bessel or whichever, and then add PEQ on top of PEQ to correct response errors.
I am a minimalist, if I can correct response errors within the filter transfer function, then I choose that way. Its also easier when working in analogue domain, besides....start with a good driver, EQ little, rather than vice versa. Occam's Razor.
It is in regard to ALL standard models of filter behaviours. These filter coefficients are just a mathematical exercise in definition of a filter.
As Juhazi rightly states, the acoustic end result is whats important. I found in my own experience that the right filter mostly does not fit into the mathematical standard definitions and is arbitrarily defined.
My own filters use a 5th order LP with broadly 4th order phase rotation, and a ripple response similar to a cheby 0.5 dB alignment, this crossed to a 4th order HP with protection capacitor which also doubles to align phase very very well (in simulation) to single figure degrees. Phase and amplitude are very good through the crossing point.
LR not required ��
There are many ways to skin the cat...
For the DSP guys, it's easy to dial in an alignment Bessel or whichever, and then add PEQ on top of PEQ to correct response errors.
I am a minimalist, if I can correct response errors within the filter transfer function, then I choose that way. Its also easier when working in analogue domain, besides....start with a good driver, EQ little, rather than vice versa. Occam's Razor.
I see. I think I agree with you. You said you're a minimalist but your crossover is 5th order? Hehe I think I understand that
I have noticed that LR4 can sound very smooth - like silk as you say. Seamless handover from one driver to the next as the sound passes through the XO region. Most music sounds great when LR4 is done right. But when the phase coherent BW1 XO is implemented properly, there is an almost magical ability for the speaker to transport you to the scene of the recording with the realism that comes from correct time alignment combined with correct step response. It doesn't happen for all music - just some that are recorded "live" style with single listener perspective stereo mics. Drum solos especially sound exceptionally real as well as small ensembles on a stage. But the BW1 is very hard to do in reality as it requires wide bandwidth drivers that are smooth. I have hear that the Dunlavy SC-IV uses this approach.
Even tho the Duelund's idea is not very applicable imho, it is very important to understand what has motivated him to arrive at such solution. I'm absolutely with him on that.
He thought that multi-way is perfect only in theory. In reality it often removes the speaker ability to transport you to the scene of the recording as you phrase it. Full range is best, 2-way still possible, 3-way getting harder. I agree with him that the issue is PHASE (but I have my own way to deal with it).
From his Mathematical model, it is obvious that his preferred 2-way is LR2 (Not LR4!) and by inverting the tweeter polarity. Troel Gravesen did it with staggered baffle. John Krutke used passive delay network in his ZD5 (a-la Dynaudio). Many others use sloped baffle. Almost all known designers have been fascinated by LR2
LR4 is "impressive", but most of the time it fails to bring the listener to the "joy", if you know what I'm talking about.
OOOPS, sorry, it seems that Duelund's formula will give LR4 with Q=0.707 ???
Last edited:
Jay you made me giggle �� Id love to be able to use first order filters all.round, but they really aren't practical for anything, except super tweeters.
The passive crossover I had before was 2nd order all round, and was acceptable, BUT going to 4th order meant better stop band rejection, without polarity inversion for the tweeter. The 5th order was just for BSC. The only sticking point is the full cycle delay between HP and LP, I could correct that but I am still undecided if it is a necessary complication.
So the decision to go to 5th order was due to limitations inherent to the drivers I was using, metal cone woofer and highish Fs of the tweeter. Both of these could be fixed by choosing different drivers, but what I gain in one area, I lose in another.
So perhaps I should rephrase... Occam's razor, with compromises ��
Of course, DSP would possibly solve issues for me, but there aren't enough to solve (for me) to justify its use.
The passive crossover I had before was 2nd order all round, and was acceptable, BUT going to 4th order meant better stop band rejection, without polarity inversion for the tweeter. The 5th order was just for BSC. The only sticking point is the full cycle delay between HP and LP, I could correct that but I am still undecided if it is a necessary complication.
So the decision to go to 5th order was due to limitations inherent to the drivers I was using, metal cone woofer and highish Fs of the tweeter. Both of these could be fixed by choosing different drivers, but what I gain in one area, I lose in another.
So perhaps I should rephrase... Occam's razor, with compromises ��
Of course, DSP would possibly solve issues for me, but there aren't enough to solve (for me) to justify its use.
But you said "start with a good driver, EQ little, rather than vice versa"
But, those are also my experience. I prefer rigid and lightweight cone for the details. But I will prefer the best treated paper cone if the price is acceptable
Yes, yes I did
The variability of paper cones was my motivation for choosing metal, and within its bandwidth, for a metal cone Visaton AL130 is one of the best. It has its breakup, but far less than many of the same "quality".
I would ideally have liked a good paper cone, so many formulations make it a difficult choice. Rigid paper, same issues with breakup, softer cone and meh quality.
I gave up trusting hearsay and opinion here a long time ago, too much vested interest. I'm aching to use the mission paper woofers I have, very cheap and surprisingly good. The Audax HDA are still my all time favourites, almost a best of both worlds, warts and all.
So, take what I and many others say with a pinch of salt. Data doesn't tell all, but it tells more than opinion
The variability of paper cones was my motivation for choosing metal, and within its bandwidth, for a metal cone Visaton AL130 is one of the best. It has its breakup, but far less than many of the same "quality".
I would ideally have liked a good paper cone, so many formulations make it a difficult choice. Rigid paper, same issues with breakup, softer cone and meh quality.
I gave up trusting hearsay and opinion here a long time ago, too much vested interest. I'm aching to use the mission paper woofers I have, very cheap and surprisingly good. The Audax HDA are still my all time favourites, almost a best of both worlds, warts and all.
So, take what I and many others say with a pinch of salt. Data doesn't tell all, but it tells more than opinion
It seems that one of the main advantages of having a midrange driver handle a wide range of frequencies is that you can use a woofer that has a hard cone and a tweeter that has particularly good high end response and dispersion. The hard cone woofers all seem to have severe resonance and/or "breakup" right at the frequencies where the ear is most sensitive (3-5kHZ), so they need to be crossed over low in freq. and/or with high order filters. Most tweeters have distortion issues when they are ran down below about 3kHZ. So I guess it's a tradeoff against being able to reproduce a clean squarewave with the approach presented in this thread.
With the Peerless TG or TC 3 inch drivers, I've been able to keep crossover points away from where the ear is most sensitive, which seems like a good thing. I use it from 500HZ to at least 7kHZ (in one system I built the upper crossover point is at 16kHZ, where the TG driver naturally rolls off). Although I use 4 pole active crossovers (for the 500HZ point only - One pole passive at 7kHZ or 15kHZ), I suspect that a simple 1 pole passive might work pretty well with the drivers I'm using, because of how wide a frequency band I can leave to the 3 inch mid driver. Does this time alignment technique give an advantage that outweighs these features? Perhaps with just the right drivers (?).
With the Peerless TG or TC 3 inch drivers, I've been able to keep crossover points away from where the ear is most sensitive, which seems like a good thing. I use it from 500HZ to at least 7kHZ (in one system I built the upper crossover point is at 16kHZ, where the TG driver naturally rolls off). Although I use 4 pole active crossovers (for the 500HZ point only - One pole passive at 7kHZ or 15kHZ), I suspect that a simple 1 pole passive might work pretty well with the drivers I'm using, because of how wide a frequency band I can leave to the 3 inch mid driver. Does this time alignment technique give an advantage that outweighs these features? Perhaps with just the right drivers (?).
Last edited:
Bob,
First order for which? HP or LP?
I can see it working for LP, but not a cat in hells chance for HP unless you combine with a undertuned reflex for the TG to give an acoustic 4th order response. For the LP you have to align phase which is at quadrature. All pass could do it I guess, with the crossover being very high (your 16k) and the delay being small(er) than attempting the same at say 4k.
A bit daunting task though
I am tempted.to assert that a 3rd order LP on the TG would align phase better to the first order filtered supertweeter. But then...is quadrature phase shift audible at >8k or so? That is something I dare not speculate over.
I would say that such a driver would probably suit the Delund filler midrange task quite well and its 2nd or 4th order crossover requirements.
First order for which? HP or LP?
I can see it working for LP, but not a cat in hells chance for HP unless you combine with a undertuned reflex for the TG to give an acoustic 4th order response. For the LP you have to align phase which is at quadrature. All pass could do it I guess, with the crossover being very high (your 16k) and the delay being small(er) than attempting the same at say 4k.
A bit daunting task though
I am tempted.to assert that a 3rd order LP on the TG would align phase better to the first order filtered supertweeter. But then...is quadrature phase shift audible at >8k or so? That is something I dare not speculate over.
I would say that such a driver would probably suit the Delund filler midrange task quite well and its 2nd or 4th order crossover requirements.
Last edited:
This is why I want to explore this with my 3-way test bed that will soon be built using a TG9 or TC9 for the mid and a silk dome DC28F-8 for the top. Still wondering what a good woofer to use is, I am considering a papercone Beta 12cx coax operated as just a woofer. Center XO freq I am thinking 1.5k which gives more than 2 octaves above and below for the Vifa.
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:
Duelund: said:
But you can use more drivers so bandwidth is very limited.
Last edited:
Those are very typical step responses of regular non transient perfect speakers. Look at what Stereophile says a good step response should look like and how rare it is (in multiway speakers anyhow, common in full range speakers).
Measuring Loudspeakers, Part Two Page 3 | Stereophile.com
Theoretically, ( I would think ) full range speakers should ALL have a good step response...unless something real ugly is happening.
The question that article spawns (in my mind) is what criteria determine a good step response, other than a coherent plot? Where do the bounds exist?
How is it quantifed from what we know is the ideal attack decay envelope?
Time coherence in a 3 way is trickier, as almost all things already are in a 3 way.
Just one reason I haven't bitten and tried one yet.
The question that article spawns (in my mind) is what criteria determine a good step response, other than a coherent plot? Where do the bounds exist?
How is it quantifed from what we know is the ideal attack decay envelope?
Time coherence in a 3 way is trickier, as almost all things already are in a 3 way.
Just one reason I haven't bitten and tried one yet.
Last edited:
X
X,
Since you are good at the blind comparisons...
Would a blind test of multiway system with typical step response, vs corrected aligned step response, be an interesting test?
Id imagine you could screw up time alignment and correctly align it in DSP well enough to test the audibility.
Or is it not that simple?
X,
Since you are good at the blind comparisons...
Would a blind test of multiway system with typical step response, vs corrected aligned step response, be an interesting test?
Id imagine you could screw up time alignment and correctly align it in DSP well enough to test the audibility.
Or is it not that simple?
Ha! Thanks. I remember the Revel. Didn't realize Magico also used steep slopes.
It will be interesting for us, but not for most unfortunately. It's a pity, people do only debate/disapprove but don't want to do an effort to do collective research regarding audibility of things in audio.
I have read that several times. The author John Atkinson does not consider phase/transient perfectness as a critical factor. There are other more important parameters. Have you read the autitory verdicts of these too, do they complain about poor resolution, imaging, etc?
One more Vivid Audio Giya G3 loudspeaker Measurements | Stereophile.com
"Yes, at $39,990/pair, it is expensive, but for that money you get a combination of transparency and, rich low bass aside, almost full-range neutrality, coupled with superbly precise imaging and an easy-on-the-ear yet revealing balance that proved addictive."
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.