That hybrid part was born out of fear
. Fear of not being able to use my 'reference' amplifier anymore. If I had not made it a passive solution I would have needed lots of small amplifiers. And I 'believe' that amplifier to be something special .
I wish there were not the misperceptions about FIR; that it automatically provides too much correction, or is automatically linear phase.
Neither of those is remotely true... FIR is a blank sheet of paper, waiting for the author to place on it, what they want it to do.
FIR can have as few or as many EQs as desired.
With as low or high a Q, and with as much gain or attenuation, as desired.
FIR can have no EQ's at all, and be xovers only.
Or it can be xovers only, with no EQs.
FIR can act as 100% IIR, not only for EQs, but for IIR xovers too.
(And do so with higher IIR precision than non-FIR DSP's.)
FIR can implement precise timing delays down to one processing sample.
And (not automatically !
) FIR can be linear phase.And also be: minimum phase, maximum phase, and mixed phase.
FIR is a very powerful tool that demands good measurements to judiciously act on.
The choices are simply up to the user.
I guess the automatic room correction stuff, and some of the automatic FIR-generator software have created the "over-correction" misperception.
(And it often is over-correction imnsho).
I guess the outspoken zeal of FIR users who have experienced the benefit of using complementary linear-phase xovers, at whatever order desired with no phase rotation or group delay, has created the "automatic linear-phase" misperception.
(i'm guilty of this zeal, and maybe enhancing the misperception. I'd like to kill the misperception, but will continue to strongly advocate complementary linear phase xovers as one of the best things you can ever draw on the blank sheet of FIR paper.)
I know that I kicked the can, by addressing FIR
But I also think that it's a huge subject, since we have to make some kind of decision, like filter type, slope, speaker design/layout, amplifier, size, price, active, known kit or totally DIY from the ground up. And suddenly active vs passive in the middle of everything.
My honest opinion is totally biased, since I started DIY speakers with actives - right away. All my prior speakers were premade or bought finished. I saw it as absolutely brutal to try and calculate and simulate passive components to match and work with drivers in unity.
Active was going directly to the acoustic output and adjust until it worked. Seemed simpler.... it was not.. maybe.... just different. I just still think it can have an advantage, and I'm not keen on using a pc to render FIR, simply because I want something usable for daily listening. And my Pre-amp with build in IIR DSP, does just great as a finished product.
When I listen to speakers that employ different topologies, like low orders filters in the Dunlavy or higher orders in many other speakers. Well.... again it's the way it's done... rather than the technology or principle. If it gets the job done and I like to listen to, and use it - then I'm happy.
I must admit though.... that until now... I have not heard a speaker that made it worth the complexity, to involve FIR - for correcting phase that is. Furthermore, I can surely see the cleverness in using FIR as a much more precise adjustment than IIR. But again. If we use good (appropriate) drivers in good well-designed cabinets.... then the few linear problems that we can actually do something about... is often easily fixed with simple IIR/EQ.
But I also think that it's a huge subject, since we have to make some kind of decision, like filter type, slope, speaker design/layout, amplifier, size, price, active, known kit or totally DIY from the ground up. And suddenly active vs passive in the middle of everything.
My honest opinion is totally biased, since I started DIY speakers with actives - right away. All my prior speakers were premade or bought finished. I saw it as absolutely brutal to try and calculate and simulate passive components to match and work with drivers in unity.
Active was going directly to the acoustic output and adjust until it worked. Seemed simpler.... it was not.. maybe.... just different. I just still think it can have an advantage, and I'm not keen on using a pc to render FIR, simply because I want something usable for daily listening. And my Pre-amp with build in IIR DSP, does just great as a finished product.
When I listen to speakers that employ different topologies, like low orders filters in the Dunlavy or higher orders in many other speakers. Well.... again it's the way it's done... rather than the technology or principle. If it gets the job done and I like to listen to, and use it - then I'm happy.
I must admit though.... that until now... I have not heard a speaker that made it worth the complexity, to involve FIR - for correcting phase that is. Furthermore, I can surely see the cleverness in using FIR as a much more precise adjustment than IIR. But again. If we use good (appropriate) drivers in good well-designed cabinets.... then the few linear problems that we can actually do something about... is often easily fixed with simple IIR/EQ.
I agree with you on on these things.... some of the purde digital active setups just simply get too complicated with computer, interfaces, lots of cables, connections and so on....
I am a computer geek at work and.... I don´t want to continue down that road when I want to relax with music!
I would simply have an appliance that does the job and always works, I don´t want to find myself in the middle of a windows forced update at the moment that I want to listen to "Batman Begins".... I don´t know if there is anything that would annoy me more than that!
The Trinnov Amethyst that I got is really that thing, there is a Linux based system inside and it´s an appliance that simply works and always does the job, no interruptions, it´s insanely user friendly... The 16 and 32 channel versions of the Trinnov´s have also found their use in completely active stereo setups with DIY speakers or passive speakers converted to active.... I considered the more upscale Trinnov Altitude 16 or 32 for Stereo setup with fully active speakers but the very high price tag stopped me...
Maybe I would think about a computer based setup for active speakers in addition, but the complexity of setup has so far halted me....
I am a computer geek at work and.... I don´t want to continue down that road when I want to relax with music!
I would simply have an appliance that does the job and always works, I don´t want to find myself in the middle of a windows forced update at the moment that I want to listen to "Batman Begins".... I don´t know if there is anything that would annoy me more than that!
The Trinnov Amethyst that I got is really that thing, there is a Linux based system inside and it´s an appliance that simply works and always does the job, no interruptions, it´s insanely user friendly... The 16 and 32 channel versions of the Trinnov´s have also found their use in completely active stereo setups with DIY speakers or passive speakers converted to active.... I considered the more upscale Trinnov Altitude 16 or 32 for Stereo setup with fully active speakers but the very high price tag stopped me...
Maybe I would think about a computer based setup for active speakers in addition, but the complexity of setup has so far halted me....
Last edited:
But isn't an Amethist a pc mobo with dedicated converters cards, a simplistic user interface and a nice case?
Just take a look at backside of it there is obvious clues!
And it is in no way a recent approach into sound related industry as my 1995 Kurzweil K2000R sampler is exactly the same.
It looks complicated but isn't really. At least when one take a look at pro offer not consumer and from hardware side, the real need in skills is measurements related and interpretation of them.
To make/built FIR yes you needs tools: a typical measurement set up ( the same used for passive). So mic, stand, xlr, prosoundcard and a computer.
To perform it ( the filter(s)) we now have software solution. At least under Windows but i doubt Mac users don't have equivalent.
Something like Acourate Vst convolver run stand alone and if needed , bridged to the player of choice running on same computer as the one you had make measurements, built the filter profile... and if you choosed a 8x in/out soundcard you have everything for a 4 way or three way+sub or 2 passive + multisub, etc,etc,... and external analog input in case you still have tape, vinyl, wax cylinders, iron strings...
Iow if your source are digital and streaming there is less machine involved in the signal path that 'usual' analog chain ( source/pre/amp/loudspeakers).
Sure if you use analog source too it complicate a bit but nothing frightning imho.
That said i agree about ergonomy. But it is not dsp related rather multiamp architecture.
That said digital volume control may render moot all this as it prooved perfectly fine in most usage.
The real ambush i see is OS related ( and why Trinnov run open source i suspect): once the corporation decide they need more money ( 'evolute' the OS) there is not anymore support which may be an issue.
That said i runpro stuff from circa 2000/2010 and don't feel the need to upgrade either sonically or from a limitation in use ( for this particular role in the chain, for others there is no escape you have to follow...).
And yes following good design principle is the way to good sound imho. A bit of help from digital can be great or close to magic: now i've seen and understood principle there is some 'magic' behind being able to remove some bass first reflection. At least to me.
Whatever i would really like to listen to some Dunlavy with the Amethist. Curious about 'upgrade' it brings.
Last edited:
I don't think that's the point being made, I'd use an FIR processor without worries.
It's the use of it, and if someone has never achieved a correct result using IIR only and the acoustic information that process grants, then it's hard to see how they'd recognise the subtleties. In addition it is not easy to get this information out of measurements, assuming good ones of course. It all kind of moots the process. There are some real possibilities, though it's also a simple way to find an acceptable result.
I hear where you are coming from Allen, and i quickly agree that having a good grasp on acoustic design and making good measurements, are a prerequisite for achieving good results....no matter what the choice of processing or passive setup.
Where i strongly disagree is that one needs to achieve correct results with IIR only, before moving to FIR.
That is backwards in terms of assisting and accelerating learning about xovers, ime.
IIR xovers are simply much harder to implement than linear phase xovers. Let me repeat much harder.
A person can learn how to choose xover points and directivity matching dramatically easier with FIR's linear phase xover ability, which eliminates the added complexity of IIR's xovers' phase rotation and group delay.
Heck, once they get the speaker set up with lin-phase xovers, they can always take the harder next step, and add in the phase rotation & group delay by switching to IIR xovers ! Lol
Where i strongly disagree is that one needs to achieve correct results with IIR only, before moving to FIR.
That is backwards in terms of assisting and accelerating learning about xovers, ime.
IIR xovers are simply much harder to implement than linear phase xovers. Let me repeat much harder.
A person can learn how to choose xover points and directivity matching dramatically easier with FIR's linear phase xover ability, which eliminates the added complexity of IIR's xovers' phase rotation and group delay.
Heck, once they get the speaker set up with lin-phase xovers, they can always take the harder next step, and add in the phase rotation & group delay by switching to IIR xovers ! Lol
Last edited:
If this was simply a matter of response and phase then I'd agree, yes simpler, and if that's the way you wanted to learn it that's your choice, this isn't about being pedantic. No I'm talking about acoustic issues, and I'm not talking about basic acoustic response and phase, but actual acoustic problems and the way they hide themselves in the data. I'm talking about the kind of problem that is ideally fixed acoustically and which can be made worse by attempts to fix electrically. If there's a chance these issues will continue to stand out in their subtle ways and have a chance to be found and fixed, because I haven't carelessly white-washed the phase, then in cases where it matters to me I'll be doing that.
Allen this probably relate to the way Mark ( and i ) approach the issue by first treat each ways separetely then xover them, then whatever please you...
I agree about the issues you could face trying to linearise an already done passive design. And why a good one to boot is essential in thid case.
Hence why i would be curious to listen to a compensated Dunlavy design.
I agree about the issues you could face trying to linearise an already done passive design. And why a good one to boot is essential in thid case.
Hence why i would be curious to listen to a compensated Dunlavy design.
But isn't an Amethist a pc mobo with dedicated converters cards, a simplistic user interface and a nice case?
Just take a look at backside of it there is obvious clues!
And it is in no way a recent approach into sound related industry as my 1995 Kurzweil K2000R sampler is exactly the same.
It looks complicated but isn't really. At least when one take a look at pro offer not consumer and from hardware side, the real need in skills is measurements related and interpretation of them.
To make/built FIR yes you needs tools: a typical measurement set up ( the same used for passive). So mic, stand, xlr, prosoundcard and a computer.
To perform it ( the filter(s)) we now have software solution. At least under Windows but i doubt Mac users don't have equivalent.
Something like Acourate Vst convolver run stand alone and if needed , bridged to the player of choice running on same computer as the one you had make measurements, built the filter profile... and if you choosed a 8x in/out soundcard you have everything for a 4 way or three way+sub or 2 passive + multisub, etc,etc,... and external analog input in case you still have tape, vinyl, wax cylinders, iron strings...
Iow if your source are digital and streaming there is less machine involved in the signal path that 'usual' analog chain ( source/pre/amp/loudspeakers).
Sure if you use analog source too it complicate a bit but nothing frightning imho.
That said i agree about ergonomy. But it is not dsp related rather multiamp architecture.
That said digital volume control may render moot all this as it prooved perfectly fine in most usage.
The real ambush i see is OS related ( and why Trinnov run open source i suspect): once the corporation decide they need more money ( 'evolute' the OS) there is not anymore support which may be an issue.
That said i runpro stuff from circa 2000/2010 and don't feel the need to upgrade either sonically or from a limitation in use ( for this particular role in the chain, for others there is no escape you have to follow...).
And yes following good design principle is the way to good sound imho. A bit of help from digital can be great or close to magic: now i've seen and understood principle there is some 'magic' behind being able to remove some bass first reflection. At least to me.
Whatever i would really like to listen to some Dunlavy with the Amethist. Curious about 'upgrade' it brings.
Yes, all the trinnov's have a computer at the heart but to compare it to legacy pro products is a gross injustice.... Ray Kurzweil is a complete genious, still I believe a 1995 Kurzweil K2000R sampler is lightyears from Trinnov. IMHO there are no products out there in the pro or home audio arena that does what the Trinnovs do...
(I will not provide arguments for this, it's just to read their webpages)
If I were going to build a "Dunlavy like" speaker I would base it on a Trinnov multichannel pre.
Last edited:
I hear where you are coming from Allen, and i quickly agree that having a good grasp on acoustic design and making good measurements, are a prerequisite for achieving good results....no matter what the choice of processing or passive setup.
Where i strongly disagree is that one needs to achieve correct results with IIR only, before moving to FIR.
That is backwards in terms of assisting and accelerating learning about xovers, ime.
IIR xovers are simply much harder to implement than linear phase xovers. Let me repeat much harder.
A person can learn how to choose xover points and directivity matching dramatically easier with FIR's linear phase xover ability, which eliminates the added complexity of IIR's xovers' phase rotation and group delay.
Heck, once they get the speaker set up with lin-phase xovers, they can always take the harder next step, and add in the phase rotation & group delay by switching to IIR xovers ! Lol
Thanks for input, I would like to learn more .....
I received advice from Joachim Gerhard about this one:Accurate Sound Reproduction Using DSP
Any inputs anyone?
Rephase Room Correction
https://www.diyaudio.com/forums/roo...205-apply-correction-tools-2.html#post6716218
I just lost an answer to you about this, Trinnov and other things... will try to redo once work is over.
https://www.diyaudio.com/forums/roo...205-apply-correction-tools-2.html#post6716218
I just lost an answer to you about this, Trinnov and other things... will try to redo once work is over.
Let's get back to what this thread really is about, sorry for derailing:
if I can make any meaningful input to driver selection....
Is it the correct way to look for the drivers Dunlavy used or anything similar? That is so many years ago and advancement in driver design and execution since then has been immense... I read in another thread another place (stated by Joachim Gerhard) that just with the last decade or so advancement is so huge that with current technologies you will get 10 dB more output for a given driver/size and similar distortion level.
So I wonder, there must be recent drivers with similar capabilities and out-of-band performance, but way better, that will fit the bill for a first order design.
I can see that Troels Gravesen used some of the ScanSpeak Ellipticor 18WE/8542-T00 and ScanSpeak 21WE/8542-T00 drivers in first order design with extremely simple crossovers. That crossover has 7 components!
From what I understand, Ellipticor is the best ScanSpeak can currently do, the drivers seem to be extremely linear way outside the crossover points....
This is the Troels Gravesen design I was referring to: Troels Gravesen Ellipticor 3
What Gravesen did her is set the tweeter in reverse phase so it's not fully phase/time coherent, some work there, I guess
would the Ellipticor drivers be something to consider guys?
Could you consider en WMTMW design with all those above drivers?
If we add two of the ScanSpeak revelator 32W drivers at bottom and top (front facing) I imagine we could be Talking Dunlavy SC-VI performance and beyound
Could this be potential for a Dunlavy SC-VI killer?
Heck, I would like to consider such a design myself too
if I can make any meaningful input to driver selection....
Is it the correct way to look for the drivers Dunlavy used or anything similar? That is so many years ago and advancement in driver design and execution since then has been immense... I read in another thread another place (stated by Joachim Gerhard) that just with the last decade or so advancement is so huge that with current technologies you will get 10 dB more output for a given driver/size and similar distortion level.
So I wonder, there must be recent drivers with similar capabilities and out-of-band performance, but way better, that will fit the bill for a first order design.
I can see that Troels Gravesen used some of the ScanSpeak Ellipticor 18WE/8542-T00 and ScanSpeak 21WE/8542-T00 drivers in first order design with extremely simple crossovers. That crossover has 7 components!
From what I understand, Ellipticor is the best ScanSpeak can currently do, the drivers seem to be extremely linear way outside the crossover points....
This is the Troels Gravesen design I was referring to: Troels Gravesen Ellipticor 3
What Gravesen did her is set the tweeter in reverse phase so it's not fully phase/time coherent, some work there, I guess
would the Ellipticor drivers be something to consider guys?
Could you consider en WMTMW design with all those above drivers?
If we add two of the ScanSpeak revelator 32W drivers at bottom and top (front facing) I imagine we could be Talking Dunlavy SC-VI performance and beyound
Could this be potential for a Dunlavy SC-VI killer?
Heck, I would like to consider such a design myself too
Last edited:
^ this is good question: is it a correct way to look for drivers used by Dunlavy?
Well if you stick to first order design then yes you need something similar.
But given the Magnus existed....?
I mean, he made choice about them because he had access to dsp ( they had a flat front: dsp used to time align the drivers) and so to answers in digital domain to acoustic domain.
Why did he used first order filters? Because of step response, not smeared transient.
Then it seems logical he would have used FIR complementary crossover filters too. And from there drivers selection is much less an issue as you don't have to worry about out of band behavior this much ( cause steeper filters doesn't introduce transient smearing).
This lead us to Mark's previous comment about FIR and directivity and the easier path they allow.
What i would keep from his design is layout, sealed and the way to design the loudspeakers ( by measurements).
The layout for the vertical aligned models is most important as it locks a number of variable ( listening height and distance, etc,etc,...). In fact the only parameter he could not managed was relative to ceiling ( height, angled or not,...). I've heard some guys used some sheets of plywood from cabinet to ceiling to further enhance coupling...
So i would venture an infinite bafle vertical aligned loudspeaker with sealed load, wide horizontal directivity and controled vertical ( thanks to dsp and waveguided tweet a la 'horbach Keele') could well be how i see his next loudspeaker.
Sound not very different than what SpeakerDave ( Dave Smith) designed for Snell...
Not how mine are gonna be despite a high degree of similitude.
Well if you stick to first order design then yes you need something similar.
But given the Magnus existed....?
I mean, he made choice about them because he had access to dsp ( they had a flat front: dsp used to time align the drivers) and so to answers in digital domain to acoustic domain.
Why did he used first order filters? Because of step response, not smeared transient.
Then it seems logical he would have used FIR complementary crossover filters too. And from there drivers selection is much less an issue as you don't have to worry about out of band behavior this much ( cause steeper filters doesn't introduce transient smearing).
This lead us to Mark's previous comment about FIR and directivity and the easier path they allow.
What i would keep from his design is layout, sealed and the way to design the loudspeakers ( by measurements).
The layout for the vertical aligned models is most important as it locks a number of variable ( listening height and distance, etc,etc,...). In fact the only parameter he could not managed was relative to ceiling ( height, angled or not,...). I've heard some guys used some sheets of plywood from cabinet to ceiling to further enhance coupling...
So i would venture an infinite bafle vertical aligned loudspeaker with sealed load, wide horizontal directivity and controled vertical ( thanks to dsp and waveguided tweet a la 'horbach Keele') could well be how i see his next loudspeaker.
Sound not very different than what SpeakerDave ( Dave Smith) designed for Snell...
Not how mine are gonna be despite a high degree of similitude.
Last edited:
Mitch is a member here, he posts as: mitchba
I would recommend his book! If you want to see his writing style and a few interesting tutorials, just visit: Audiophile Style and see his 'Audiolense walktrough' as an example of one of his in depth speaker reviews. (You'll see a familiar picture in the review I linked
)Long ago when I started my DSP learning process, the conversation Mitch had online with Bob Katz and the maker of Audiolense served as a guideline for my own experiments (see links here): https://www.diyaudio.com/forums/full-range/242171-towers-25-driver-range-line-array-72.html#post4168817
If you're serious about looking into a Dunlavy style speaker, I'd highly recommend simming it with VituixCAD first. Don't gamble with something like this, a tool like Vituix is a great way to get you close enough to be confident of making the right decisions.
I don't think the drivers Dunlavy picked are that easy to replace. Due to the use of first order crossovers they all had to have a well behaved wide bandwidth.
With DSP tools one could afford a little more slack, but ultimately it would be better to solve as many things as one can with proper selection and using acoustic solutions. Leaving less to be depending on DSP.
Thanks @wesayso, that's cool
By the way, the maker of Audiolense is pretty active in a Norwegian forum and there are quite a few guys around here using Audiolense in fully active systems, streching the rubber band beyond the insane, some pretty awesome systems built around Audiolense. Some of the DYI build threads around here are insane
(www.hifisentralen.no sorry, all in Norwegian language)
By the way, the maker of Audiolense is pretty active in a Norwegian forum and there are quite a few guys around here using Audiolense in fully active systems, streching the rubber band beyond the insane, some pretty awesome systems built around Audiolense. Some of the DYI build threads around here are insane
(www.hifisentralen.no sorry, all in Norwegian language)
Last edited: