Computer-Aided Design of Loudspeaker Crossover Networks

[Cosmik]

An older discontinued design (again click measurements on the RHS) used to use linear phase filters but this was dropped. What might have prompted that?

Latency?

 

[dallasjustice]

I agree but there's a downside to using digital crossovers with very steep slopes.

Think about what a traditional passive crossover consists of , coils,inductors, capacitors etc etc the signal has to go through all of them, electronic crossover is better and a digital implementation allows you to use steeper slopes , use dSP to correct any driver anomalies , correct phase ,delay etc etc.
Active means one amplifier dedicated to one driver, how can that not be advantageous?
Of course there is no oppurtunity to 'upgrade' amps, cables etc.
Keith

 

[Sal1950]

Active loudspeakers are technically the way to go, everyone accepts that, digitally implemented crossovers such as Grimm and KII offer further advantages over traditional 'electronic' crossovers.
I will leave you to come up with reasons why active speakers are not more widely promoted by magazines and amplifier manufacturers.
Keith.

Bottom line Keith is it will be a very hard row to hoe trying to get audiophiles to give up on the building box method of putting together their systems. Every enthusiast wants to show how knowledgeable they are buy picking all the right pieces. Also active systems come up a bit short on making an impressive looking audio shrine.
On top, you know that, when greedy manufacturers become involved there's much more profit in separate components no matter how overpriced the active systems can be spin marketed.
If integrity and honesty were ever to become a factor again in audioland, the economies of a active system should be able to produce incredible sounding gear at very reasonable costs. Nobody wants that including the High End consumer, got to retain those luxury goods bragging rights. LOL

 

[dallasjustice]

Whether a speaker is passive or digitally active isn't really the thing which separates real high performance speakers from the manufacturers who slap a new paint job on thirty year old designs.

 

[fas42]

I agree but there's a downside to using digital crossovers with very steep slopes.

That's interesting - what have you found the problem to be?

 

[dallasjustice]

That's interesting - what have you found the problem to be?

With FIR filters, there will be pre-ringing. Every case is different tho, so one needs to study the group delay plot to make sure there's no problematic pre-ringing. All FIR filters have pre-ringing. The question is whether it's audible. To the paranoids, it's always audible, even if it's 140db down.

This is the common objection the anti-digital crossver folks claim. Like every other audiophile boogie man (eg. Jitter), there's usually a grain of truth to the fear. But the actual practice is usually different from the boogie man fears.

 

[Purité Audio]

I must ask about the type/s of filters used , but in my experience,and I am sitting in front of a more 'conventional'
active (PSI) the Kiis which are digital everything and a pair of passive horns, I actually hear more resolution through the Kiis, now whether that is other aspects of the Kiis design ,the cardiod response perhaps I can't say.
But I hear no downsides with digital implementation whatsoever.
The only reason manufacturers don't do it, is because they can't.
Keith.

 

[Cosmik]

With FIR filters, there will be pre-ringing. Every case is different tho, so one needs to study the group delay plot to make sure there's no problematic pre-ringing. All FIR filters have pre-ringing. The question is whether it's audible. To the paranoids, it's always audible, even if it's 140db down.

This is the common objection the anti-digital crossver folks claim. Like every other audiophile boogie man (eg. Jitter), there's usually a grain of truth to the fear. But the actual practice is usually different from the boogie man fears.

Yes, there will be pre-ringing for each individual driver, but the theoretical sum of all the drivers adds to zero ringing. Presumably, the off axis overlap will not be perfect so the dreaded pre-ringing may be heard off axis. However, for the appropriate smooth blend between drivers, I only use 4th order filters or less (with FIR you can have any slope you like) so the pre-ringing will be so low in level that it will never be audible, even if you listen for it on an individual driver. I've been listening to speakers with such filters for over three years, and never thought them to be anything but 'perfect'.

At the end of the day, if someone wants to believe it is a problem, there will be no persuading them. For me, curiosity about what a transient-perfect (not just phase perfect) system sounds like trumped any other issues.

Someone here does some theoretical research on the basis of the pre- and post-masking effect:
http://www.bodziosoftware.com.au/pre_post_ringing_ir_and_pulses.pdf

 

[Cosmik]

Reading a very interesting discussion about crossover design and phase:
http://techtalk.parts-express.com/forum/tech-talk-forum/68768-why-does-phase-matter/page3

Someone says

One should set targets, but that's all they are. You may have read Jeff or me say that you have to let the drivers talk to each other, this means that the phase may not line up where you set your targets. So you have to go lower or higher on both drivers until things start to line up. This is where other factors start to play a part like where the tweeter starts to distort, where the woofer's breakup is. These factors may call for a notch of some type or a higher order slope.

Clearly a computer can be set the task of searching for the ideal response, too. But the impression given by the quote above is that the result is a patchwork, and that although a certain sum between the drivers is targeted, the individual drivers are all over the place, with the off axis response presumably affected and/or the phase arbitrarily shifting about. The start-to-finish DSP active approach is *so* much simpler, and automatically results in the drivers tracking perfectly throughout the crossovers as well as delivering perfect time alignment (at one region in space), phase response etc. The result is 'solid' with better inter-channel matching too. How many people have heard such a speaker? Not many, I reckon.

 

[Phelonious Ponk]

After chatting to a speaker designer a while back who was a believer of active methods but of late feels the gap has been closed by the use of CAD programs to design x over networks I though I would ask the opinions of our members..

I don't doubt this works in theory, but I wonder about practice. Even if you can design a passive crossover that measures like an active one, that doesn't address what I think may be the active design's major advantage - the ability to match amplification to individual drivers. That ability allows designers to take the load of the low frequencies off of the critical midrange and the highs entirely, giving the amp a potentially very simple and easy load to drive. Simplify that load enough and, as I understand it, the "all amps sound alike" theory gets very close to reality, allowing for much more efficient and less expensive amplification to perform like a short stack of grunting A/B iron.

But I'm no engineer; just a guy who reads too much, so it's best to assume that everything I say has a "?" on it. I'm open to correction or clarification.

Tim

 

[Phelonious Ponk]

While I'm thinking about it, it seems curious to me that 3-way actives are relatively rare, and on the Pro side, even many large and expensive active monitors are only two way. Do any of you have any idea why this would be? Is there something about active implementation that makes a two-way a more viable design?

Tim

 

[Phelonious Ponk]

Yea, here's a quote from the PCM web site on the issue..,

While it's hard to better a well-engineered set of active speakers, passive monitors, when designed with care and respect like PMC's range, will always outperform inferior active designs. Aimed at customers who prefer to define their own speaker/amplifier combination, our passive speakers are just as uncompromising when it comes to quality.

As a retired marketing guy, I find this kind of language interesting...

"passive monitors, when designed with care and respect like PMC's range, will always outperform inferior active designs"

Well, yeah. and active monitors designed with care and respect will always outperform inferior active designs as well. I'm sure this is what they mean, not that active design is the inferior choice, because I know they're not stupid, and I know it because they immediately got to the real point in the next sentence...

"Aimed at customers who prefer to define their own speaker/amplifier combination, our passive speakers are just as uncompromising when it comes to quality."

Tim

 

[March Audio]

As you can see from my avatar I am experimenting with my own 3 way speaker design. I have spent a lot of time optimising a passive xo and then recently moved on to a digital one.

A minimal amount of time playing with a modest minidsp 2x4hd provided a superior sound, more neutral and dynamic. To me active with dsp xo is a no brainer.

Now experimenting with a minidsp minisharc for dsp and different dacs for output to ncore amps.

 

[Cosmik]

While I'm thinking about it, it seems curious to me that 3-way actives are relatively rare, and on the Pro side, even many large and expensive active monitors are only two way. Do any of you have any idea why this would be? Is there something about active implementation that makes a two-way a more viable design?

In post 19 I quote someone on the pitfalls of designing three-way passive crossovers - which explains why so many passive speakers are two-way.

From earlier comments, it seems that many people think of active and passive crossovers as equivalent, merely differing in the method of implementation. If active is just a different way of achieving the same thing, then speaker designers brought up on passive crossovers will still shy away from three-way designs even given the chance of building them active.

There are different levels of 'active'. Using analogue filters will still provide ample opportunities for a three-way design to mess up, but the DSP variety with driver correction and time alignment makes a crossover pretty much transparent - there's no reason to believe me in particular, but if you just think about what the crossover is trying to achieve, then it is obvious that giving the designer the ability to completely define how the cones of adjacent speakers move is going to help avoid much of 'the black art' of speaker design.

Come to think of it, why is speaker design so mysterious? The entirely practical business of making the cones move the way the designer wants them to move (if they use DSP) seems like a good place to start - but they don't. Instead the whole process starts from the position of designing filters in the vague hope that the cones will waft about in a way that just about allows adjacent drivers to blend to a sum that is about right in terms of frequency response at the listener's position - ish. Off-axis?... a nice to have. Accurate phase, timing?... an unaffordable luxury. Damping?... what does that matter? If we use the very finest drivers they can be coaxed into stretching over a wide frequency range, thus avoiding an extra-complicated three way passive crossover and all its unwanted side effects. This is still the reality for most audiophile speakers.

 

[Blumlein 88]

As a retired marketing guy, I find this kind of language interesting...

"passive monitors, when designed with care and respect like PMC's range, will always outperform inferior active designs"

Well, yeah. and active monitors designed with care and respect will always outperform inferior active designs as well. I'm sure this is what they mean, not that active design is the inferior choice, because I know they're not stupid, and I know it because they immediately got to the real point in the next sentence...

"Aimed at customers who prefer to define their own speaker/amplifier combination, our passive speakers are just as uncompromising when it comes to quality."

Tim

Don't fret too much Tim. I feel there is a shift, finally, toward wider acceptance of active speakers. One I thought would have happened long ago.

Now when it gets a little further along I have some ideas that perhaps as a retired marketing guy you could evaluate. Firstly I am thinking of a quality active speaker using separate vacuum tube amplifiers with each sized and tubed for the particular driver in use. You know 6550/KT88 amps for the woofer, EL34's for the mid-range, and perhaps EL84 amps for the tweeter. Active xovers prior to the amps of course. And as an option we can offer active vacuum tube crossovers. Another model would feature an Ultralinear KT88 amp for the woofer, a triode connected EL34 for the mid, and a SET for the tweeters. Allowing people to change out the tubes to taste of course.

 

[leobloom]

Unless you believe all amps sound the same there is merit in being able to chose your own.

Do you have any blind testing to show amps can sound different?

 

[Phelonious Ponk]

Don't fret too much Tim. I feel there is a shift, finally, toward wider acceptance of active speakers. One I thought would have happened long ago.

Now when it gets a little further along I have some ideas that perhaps as a retired marketing guy you could evaluate. Firstly I am thinking of a quality active speaker using separate vacuum tube amplifiers with each sized and tubed for the particular driver in use. You know 6550/KT88 amps for the woofer, EL34's for the mid-range, and perhaps EL84 amps for the tweeter. Active xovers prior to the amps of course. And as an option we can offer active vacuum tube crossovers. Another model would feature an Ultralinear KT88 amp for the woofer, a triode connected EL34 for the mid, and a SET for the tweeters. Allowing people to change out the tubes to taste of course.

 

[Sal1950]

Don't fret too much Tim. I feel there is a shift, finally, toward wider acceptance of active speakers. One I thought would have happened long ago.

Now when it gets a little further along I have some ideas that perhaps as a retired marketing guy you could evaluate. Firstly I am thinking of a quality active speaker using separate vacuum tube amplifiers with each sized and tubed for the particular driver in use. You know 6550/KT88 amps for the woofer, EL34's for the mid-range, and perhaps EL84 amps for the tweeter. Active xovers prior to the amps of course. And as an option we can offer active vacuum tube crossovers. Another model would feature an Ultralinear KT88 amp for the woofer, a triode connected EL34 for the mid, and a SET for the tweeters. Allowing people to change out the tubes to taste of course.

Brilliant! Maybe we should discuss crowdfunding a startup company to bring it all together. I smell money and plenty of it.

 

[Purité Audio]

Tubes in an active crossover, I personally wouldn't unless you enjoy playing 'find the faulty tube'.
Keith

 

[Thomas savage]

Put some front oriented bread sized ports on them so they double up as toasters with the salt content of processed white bread the toasted bread crumbs left inside the speakers will act as peizoelectric filters.

Looks like Keith's PSI speakers have beat us too it