The Argenta had a very flat response, also off axis.
The drivers where really in phase over a wide range.
Some people said they where a bit sharp and thin so i made the Argenta Edition.
I think that had the SEAS ER Reedcone.
I brought it to Audio for another test.
They did not like it as much as the original one.
Not as crispy etc. and so it goes ....
We professionals often face a firewall of revievers so it can happen that a model that you really like does not get the attention it deserves and sometimes it is the other way around.
They freak out over a model that is not so important for the designer.
The drivers where really in phase over a wide range.
Some people said they where a bit sharp and thin so i made the Argenta Edition.
I think that had the SEAS ER Reedcone.
I brought it to Audio for another test.
They did not like it as much as the original one.
Not as crispy etc. and so it goes ....
We professionals often face a firewall of revievers so it can happen that a model that you really like does not get the attention it deserves and sometimes it is the other way around.
They freak out over a model that is not so important for the designer.
Attachments
I find phase aligned LR4 a bit harsh even if it's revealing. BW3 seems to just fill the room better, maybe the drivers play nicely together in BW3 to cancel out distortions?
But, amazing Joachim! You got it straight away. The Mexican hat in a series filter! I think it's because a series filter doesn't actually perform an experiment in splitting the signal measureably and seeing which path the electron signal goes down, so is a different measurement from a parallel filter.
Of course Fourier Transform isn't the only game in town, just an easy one to measure for frequency response and amplitude. Laplace transform was never very useful to me. But those wavelets look useful. More fundamental albeit abstract looking. They are all just orthoganal functions in the end. I shall read that. Thanks.
I did some fourth order Steen Duelund theory here: http://www.diyaudio.com/forums/multi-way/258017-troels-gravesen-time-aligned-3-way-published.html
Phase aligned at all frequencies. His papers are hard to find now, but still true IMO. I'm not sure the B&O filler driver was quite the same thing, Lynn Olson said it was a quadrature filter. Which might be more of a BW3 thing. Steen's theory also extends well to second order LR2 filters in three ways.
SEAS Kit 503
But, amazing Joachim! You got it straight away. The Mexican hat in a series filter! I think it's because a series filter doesn't actually perform an experiment in splitting the signal measureably and seeing which path the electron signal goes down, so is a different measurement from a parallel filter.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Of course Fourier Transform isn't the only game in town, just an easy one to measure for frequency response and amplitude. Laplace transform was never very useful to me. But those wavelets look useful. More fundamental albeit abstract looking. They are all just orthoganal functions in the end. I shall read that. Thanks.
I did some fourth order Steen Duelund theory here: http://www.diyaudio.com/forums/multi-way/258017-troels-gravesen-time-aligned-3-way-published.html
Phase aligned at all frequencies. His papers are hard to find now, but still true IMO. I'm not sure the B&O filler driver was quite the same thing, Lynn Olson said it was a quadrature filter. Which might be more of a BW3 thing. Steen's theory also extends well to second order LR2 filters in three ways.
SEAS Kit 503
I do not think that the Mexican Hat you see is a fuction of the series filter.
The dimples left and right from the peak at the crossover are already present in the response of the woofer and tweeter. The peak corresponds with an impedance drop.
That is nothing less then an impedance transform so the speaker draws more current here, hence the peak.
When i talked about my experiment, Argenta series, Argenta parallel the TRANSFER FUNCTIONS WHERE THE SAME:
That required more passive parts in the series filter.
Even small changes in amplitude are audible so the two candidates should measure as close as posible.
The dimples left and right from the peak at the crossover are already present in the response of the woofer and tweeter. The peak corresponds with an impedance drop.
That is nothing less then an impedance transform so the speaker draws more current here, hence the peak.
When i talked about my experiment, Argenta series, Argenta parallel the TRANSFER FUNCTIONS WHERE THE SAME:
That required more passive parts in the series filter.
Even small changes in amplitude are audible so the two candidates should measure as close as posible.
I know this of cause but where is the deal ? :Crossovers
The Duelund simply has a somewhat more linear group delay.
The Duelund simply has a somewhat more linear group delay.
Linkwitx filters, like John Kreskovsky's stuff just pays homage to Steen Duelund. But you have an infinite variation of parameters, as Steen explains. More or less group delay. The flattest one is a wide bandwidth fullrange driver with a bass and a tweeter doing token duty at either end of the band. The worst one is where the mid almost disappears leaving you with near LR4 two way.
I've done a lot of target response modelling with series and parallel filters too. And I can get them much the same.
But if you look for a shallow spread gaussian response from the transfer function, I think you'll find that series filters do it more simply. Parallel filters work more simply with that sort of quadratic LR4 slope.
Series filters can do notches too, without effort. I don't know where the notches hide quite, but they are rather smooth and elegant. John Kreskovsky now reckons series filters are a waste of time, but I think they are just more challenging on drivers.
I now look for a bell shaped frequency response for a driver that will work well with a series filter. Like this:
H1499-06 27TBCD/GB-DXT
H1471-08 CA22RNY
Now, THAT isn't what most folks look for, is it?
I've done a lot of target response modelling with series and parallel filters too. And I can get them much the same.
But if you look for a shallow spread gaussian response from the transfer function, I think you'll find that series filters do it more simply. Parallel filters work more simply with that sort of quadratic LR4 slope.
Series filters can do notches too, without effort. I don't know where the notches hide quite, but they are rather smooth and elegant. John Kreskovsky now reckons series filters are a waste of time, but I think they are just more challenging on drivers.
I now look for a bell shaped frequency response for a driver that will work well with a series filter. Like this:
H1499-06 27TBCD/GB-DXT
H1471-08 CA22RNY
Now, THAT isn't what most folks look for, is it?
Last edited:
We build a transient perfect filter in 1993 at the university of Essex.
It is an L-R 4 implemented with IIR filters and the phase distortion removed with a FIIR filter.
The same is used in the Grimm LS1.
Bruno told me some years ago that he invented it.
Better he would have read our papers from 20 years ago.
That works of cause only in DSP.
It is an L-R 4 implemented with IIR filters and the phase distortion removed with a FIIR filter.
The same is used in the Grimm LS1.
Bruno told me some years ago that he invented it.
Better he would have read our papers from 20 years ago.
That works of cause only in DSP.
I build 120 DSPs in 1993.
Sold only 12 world wide.
I have left only a few, the rest landed in the waistbin.
Audio is a conservative busyness.http://www.audiophysic.de/download/virgo/image_hifi_1996_02.pdf
Sold only 12 world wide.
I have left only a few, the rest landed in the waistbin.
Audio is a conservative busyness.http://www.audiophysic.de/download/virgo/image_hifi_1996_02.pdf
That's quite off the wall for you, Joachim. Johnny Cash and June Carter, eh? I prefer the heartbreak and ultimate resolution of Kris Kristoferson's "Me and Bobby McGee" myself: The Highwaymen - Me And Bobby Mcgee - YouTube
I'll tell you why 6" bass and tweeter is my worst nightmare come true! See, do it on 3.5kHz LR4 and the tweeter filter is tuned to 5kHz. Which just happens to be the breakup frequency of a 6" bass. So it all interacts horribly and sounds like Scheiße.
Well, that is my theory. I can't get it to work in BW3 on flat baffle either, which is frustrating. But at least we have a clue why it's just an awful combination of drivers.
I did FIR and IIR filters back in the 80's at college when it was quite cutting edge. We figured it all out under Dr. Tony Constantinides who was rather clever. He's now deservedly a Prof.
We also did Shannon's Noiseless encoding and Information theory. Prof. Turner was explaining that even with a tiny 1W walkie-talkie transmitter on Earth, if you repeated the message often enough, people 4 light years away on Alpha Centauri would eventually be able to decode the faint message over the background noise.
At which point my brilliant student mate Mala Wanderagula from Sri Lanka nudges me. He whispers: "Do you know what this means?"
"What does it mean, Mal?"
"Woices from Space!"
We both collapsed in laughter. Leaving Prof. Turner wondering what he had said that was so funny.
This is the thing, Joachim. There has to be a point to all this deep abstraction. Otherwise you just reach the big "So, What?" with ideas. Happily this is a very interesting thread so far. I think we're going to come up with something INTERESTING here.
I'm thinking the 1 -2s +ss / 1 +2s +ss allpass transfer function (where s is a complex variable) lurking in a second order BW2 filter might mean TWO midranges should be used. But I haven't quite fleshed it out yet. But you might be onto something.
I'll tell you why 6" bass and tweeter is my worst nightmare come true! See, do it on 3.5kHz LR4 and the tweeter filter is tuned to 5kHz. Which just happens to be the breakup frequency of a 6" bass. So it all interacts horribly and sounds like Scheiße.
Well, that is my theory. I can't get it to work in BW3 on flat baffle either, which is frustrating. But at least we have a clue why it's just an awful combination of drivers.
An externally hosted image should be here but it was not working when we last tested it.
I did FIR and IIR filters back in the 80's at college when it was quite cutting edge. We figured it all out under Dr. Tony Constantinides who was rather clever. He's now deservedly a Prof.
We also did Shannon's Noiseless encoding and Information theory. Prof. Turner was explaining that even with a tiny 1W walkie-talkie transmitter on Earth, if you repeated the message often enough, people 4 light years away on Alpha Centauri would eventually be able to decode the faint message over the background noise.
At which point my brilliant student mate Mala Wanderagula from Sri Lanka nudges me. He whispers: "Do you know what this means?"
"What does it mean, Mal?"
"Woices from Space!"
We both collapsed in laughter. Leaving Prof. Turner wondering what he had said that was so funny.
An externally hosted image should be here but it was not working when we last tested it.
This is the thing, Joachim. There has to be a point to all this deep abstraction. Otherwise you just reach the big "So, What?" with ideas. Happily this is a very interesting thread so far. I think we're going to come up with something INTERESTING here.
I'm thinking the 1 -2s +ss / 1 +2s +ss allpass transfer function (where s is a complex variable) lurking in a second order BW2 filter might mean TWO midranges should be used. But I haven't quite fleshed it out yet. But you might be onto something.
Certainly to find a good crossover is very important.
I think there has being a lot of progress with drive units but the art of crossovers has not much advanced the last 20 years.
Yes, the advantage of certain all pass crossovers is that the drivers are in phase over a wide range.
One disadvantage is that there is a hole in the power response around crossover because with both drivers at -6dB we do not get constant voltage.
A third order Butterworth does not have that problem but has vertical lobing.
The out of phase version that has better phase response then the in phase version ( both sum to flat ) has that problem even more.
I think there has being a lot of progress with drive units but the art of crossovers has not much advanced the last 20 years.
Yes, the advantage of certain all pass crossovers is that the drivers are in phase over a wide range.
One disadvantage is that there is a hole in the power response around crossover because with both drivers at -6dB we do not get constant voltage.
A third order Butterworth does not have that problem but has vertical lobing.
The out of phase version that has better phase response then the in phase version ( both sum to flat ) has that problem even more.
Very nice sound, BW3! I like it.
Has none of the harshness or wrongness of LR4. I think I know why. Haven't decided which polarity I prefer though.
The pole zero representation of the 1 -2s + 2ss -sss / 1 +2s +2ss + sss third order allpass which lurks in it is rather lovely:
I think lobing is a FEATURE, not a problem. Why fight it? Because those s or jw terms can represent a 90 degree phase difference, OR a 90 degree difference in spatial orientation of the drivers. Hence the Allison Model Four speaker looks rather interesting. It breaks all the the rules, doesn't it! Two tweeters there. Quarter section filters.
Has none of the harshness or wrongness of LR4. I think I know why. Haven't decided which polarity I prefer though.
The pole zero representation of the 1 -2s + 2ss -sss / 1 +2s +2ss + sss third order allpass which lurks in it is rather lovely:
An externally hosted image should be here but it was not working when we last tested it.
I think lobing is a FEATURE, not a problem. Why fight it? Because those s or jw terms can represent a 90 degree phase difference, OR a 90 degree difference in spatial orientation of the drivers. Hence the Allison Model Four speaker looks rather interesting. It breaks all the the rules, doesn't it! Two tweeters there. Quarter section filters.
Attachments
Well, if that ain't the Sh***iest thing you ever posted....I'll tell you why 6" bass and tweeter is my worst nightmare come true! See, do it on 3.5kHz LR4 and the tweeter filter is tuned to 5kHz. Which just happens to be the breakup frequency of a 6" bass. So it all interacts horribly and sounds like Scheiße.
Just reminded me of being in Deutschland when I was younger. I would be speaking english to Deutsch friends and I wood occasionally use the word "Sh*t", like any american, but, I would get this real funny look. After about a week someone said something about how they use that word and what do I mean? Holly Sh*t!
However, for the uniformed: **** - Wikipedia, the free encyclopedia
Joachim, I can't believe you have allowed yourself to be so distracted by a troll.
Who cares about the local variations in language? Dreck, merde, skitt, scheiße, they all just describe failure in design.
The laws of physics don't favour any particular race or nation. There is no quintessentially German loudspeaker. Only good one and bad ones.
What it's all about, is that the laws of physics apply just as much to loudspeakers as they do to quantum mechanics.
There are some elegant solutions, and a whole world of speculative rubbish like Quantum entanglement. Why people spend time on that stuff beats me. They just don't get it. Schroedinger's Cat was dead all along in one box or the other.
Understand this filter, as Doctor Tony Constantinides taught us to do, and there is very little work left to do:
Even the Hydrogen atom uses that one in 2p orbitals. How fundamental is that?
Who cares about the local variations in language? Dreck, merde, skitt, scheiße, they all just describe failure in design.
The laws of physics don't favour any particular race or nation. There is no quintessentially German loudspeaker. Only good one and bad ones.
What it's all about, is that the laws of physics apply just as much to loudspeakers as they do to quantum mechanics.
There are some elegant solutions, and a whole world of speculative rubbish like Quantum entanglement. Why people spend time on that stuff beats me. They just don't get it. Schroedinger's Cat was dead all along in one box or the other.
Understand this filter, as Doctor Tony Constantinides taught us to do, and there is very little work left to do:
An externally hosted image should be here but it was not working when we last tested it.
Even the Hydrogen atom uses that one in 2p orbitals. How fundamental is that?
Please let me point out that this may be a small misunderstanding.
I am sure JG wanted to say that this is the way many german speakers
of the time were made. And it is true, in the seventies setups like this
one were quite common and today we wonder why they did it that way.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.