3rd order butterworth, proper implementation

I am not sure what I am meant to be agreeing with. Evan a half decently built yet imperfect first order passive crossover speaker implementation in a real room has significant advantages over higher order crossovers in speech intelligibility and the ability to hear, for example, individual instruments or voices in a group.

That is why for the past 30 years I have mostly implemented first order active and passive speaker systems for professional and domestic systems, even though it goes against conventional wisdom.

One surprising benefit of first order crossovers in competently designed public address speakers is greater gain before acoustic feedback. I was once a technical writer for a professional audio magazine (Connections) and know that that statement will invoke strong disagreement - from people who haven't tried it!
 
Correct to my post #36 above:

Many (most?) people don't realise that only a first order crossover (whether passive or active) can pass audio without altering how it sounds when the outputs of the filters are recombined - the output sounds different to the input for all orders of crossover higher than 1.

Must agree most folks don't seem to realize what higher-order xovers, anything beyond first order, necessarily do to the time-domain when recombined.
I've seen clever low-order xovers beyond first order, that can reduce the time-domain damage, but none that can eliminate it.
(By xovers, i mean mainstream passive, analog active, or IIR replications with DSP.)
And I think for a speaker alone, not considering the room, it really is that simple.

But I've also seen how some like to combine various higher-order xovers' lobing patterns with how the lobes will interact with a room, so it wont seem as simple to them.
Just different perspectives on how to reach our particular end games, i guess.
 
One surprising benefit of first order crossovers in competently designed public address speakers is greater gain before acoustic feedback. I was once a technical writer for a professional audio magazine (Connections) and know that that statement will invoke strong disagreement - from people who haven't tried it!

Thanks for posting the details of your crossover and the reasons for your design choices, very interesting. I assume it sounds great, those drivers otimally crossed over with low order slopes. I couldn't find any info at all on your tweeter though, can you describe its character/compare it to a common tweeter? Something I've read is the B139 can have its Fs rise significantly (double or more) over time, is yours still down around 25hz?

I had no clue you could increase maximum gain before feedback by using first order crossovers... it seems like something which wouldn't have an effect. Do you have a theory why it happens?
 
I am not sure what I am meant to be agreeing with.

I believe Allen means that by stating 'those same characteristics conspire against higher order crossovers too...' you implicitly accept they also apply to 1st order electrical and acoustical crossovers.

Evan a half decently built yet imperfect first order passive crossover speaker implementation in a real room has significant advantages over higher order crossovers in speech intelligibility and the ability to hear, for example, individual instruments or voices in a group.

That is your opinion, not fact.
 
I wouldn't say the system sounds great by todays standards; the KEF drivers belong back in the '60s and '70s. I did this design to satisfy the client's nostalgia for their system owned for >40 years, and because I had 'revived' several pairs of KEF Concertos in the mid 1980s and knew how dramatically better they could sound with a better crossover than KEF's SP1003.

I didn't realise the tweeter has been discontinued. Here is its product data: Peerless Drivers - Tymphany

I don't have the measured data on the old B139s handy. I do recall that the system resonance was 28Hz. The driver fs doesn't impact on the crossover design in this case.

A first order crossover doesn't increase gain before feedback in a PA system, rather it avoids the lobes and resonances of typical "constant directivity" PA speakers.
These two innate characteristics of some real world directional loudspeakers give rise to break frequencies for feedback that are highly spatially position dependent.
 
Last edited:
Quote:
Evan a half decently built yet imperfect first order passive crossover speaker implementation in a real room has significant advantages over higher order crossovers in speech intelligibility and the ability to hear, for example, individual instruments or voices in a group.
That is your opinion, not fact..

It is an opinion based on an understanding of the behaviour of loudspeakers in real rooms, the behaviour of the ear/brain auditory system of real people, and how speech perception works. Based on >20 years of success of the application of this opinion in professional and consumer audio systems, I consider it a fact. I have made a couple of presentations to the Audio Engineering Society - Adelaide Section that have included a discussion on why this is so and presented practical applications in field visits.

I am not clever enough to understand how it helps discrimination of individual voices or instruments in human auditory perception, however as I mentioned earlier David Griesinger did try to explain it to me once in 1998. I consider David to be the world authority on spacial auditory perception of audio systems and few people if any come close having so much published research on the topic. http://www.davidgriesinger.com
 
Last edited:
I wouldn't say the system sounds great by todays standards; the KEF drivers belong back in the '60s and '70s. I did this design to satisfy the client's nostalgia for their system owned for >40 years, and because I had 'revived' several pairs of KEF Concertos in the mid 1980s and knew how dramatically better they could sound with a better crossover than KEF's SP1003

Yeah, they're only so good. Nostalgia helps things sound better too.

I had a pair of Infinity SM112s for a long time before I got into higher end stuff. I think that because it had a 10 inch woofer, it was the best performing of the series. It was crossed at 900hz to a 4" midrange which crossed to a 1" "high output" polycell dome at 5khz. 3rd order butterworth mid to tweeter, not sure of the woofer. Maybe the 15" version with two vertically aligned 4 inch mids crossed at 500hz would have been better than them, but maybe not (the cabinets weren't braced, none of them in the series unfortunately... I braced mine which helped a bit, so I would've braced the 15 which would have helped more). I measured the mids Fs in 3 pairs of 112s. Surprisingly they were all matched pairs to within a couple hz, ranging from 400 to 440hz. I'm not sure how well they would have worked crossed at 500hz, so my 112s might have been the best model.

The speakers were extremely dynamic and high resolution. Electrostatic like. Not so much the woofers, but definitely the mids and tweeters. The mid was probably 105db/w with an extremely well damped rear chamber. I think it would've been 108 if the rear chamber wasn't stuffed (I didn't realise how extremely efficient they were until I took them out to measure Fs). The tweeter was 100, recessed into the shallowest waveguide, 1.1" at the thinnest and 1.3" at the face plate, probably good for 11khz and up, keeping dispersion correct at the highest frequencies. The woofer was around 96, which I found out after stuffing the empty cabinet and the reflections off the back were absorbed. They sounded lean after lol. I equalized it, sounded better than taking out the stuffing and not. I powered them with the power amp section of a Pioneer SX-950 receiver, preamp varied but mostly PC sound card ranging from Sound Blaster AWE 64 gold (ISA) to Audigy 2 with custom ASIO drivers.

Good times
 
It is an opinion based on an understanding of the behaviour of loudspeakers in real rooms, the behaviour of the ear/brain auditory system of real people, and how speech perception works. Based on >20 years of success of the application of this opinion in professional and consumer audio systems, I consider it a fact. I have made a couple of presentations to the Audio Engineering Society - Adelaide Section that have included a discussion on why this is so and presented practical applications in field visits.

Which is still opinion: one not universally accepted, and while I have no doubt everybody here respects your personal experience, many others have similar amounts in the field which does not align with your own. I have already given an example of Dunlavy designs which by configuration (as John himself clearly pointed out) were not flexible in terms of listening position, either in terms of listening distance, or vertical axis. Now that is a scientific fact, demonstrated with hard data (and you only had to listen to those models closer than intended, or more than about 10 degrees off the tweeter's vertical axis for the issues to be very audible indeed). QED, the use of 1st order filters in and of itself is no guarantee of any nebulous concepts of room interaction, speech intelligability or anything else for that matter. Would that it were, life might be a little easier. 😉
 
My posts have been about the importance of one aspect of speaker design that is most often ignored, namely power response. In a real room* a speaker's power response is important for the perception of musical quality, articulation and speech intelligibility. *For the purposes of my discussion, real room means a space of dimensions such that multiple first order reflections arrive in the first few 10s of milliseconds, as opposed to a hall where they fall outside of that window.

John Dunlevy had to comply with same laws of physics as everyone else and, like everyone else's designs, his designs only sum correctly at one point in space. Any large format system needs to listened to at a distance commensurate with the size of the system.

Dunlavy made a range of designs and not all of them need to be listened to at a large distance. The ABC in Australia used Dunlevys on OB vans, for example, where extreme off axis listening is impossible to avoid, and those monitors were designed with the same principles as his large ones, MTM arrays with first order crossovers.

Unfortunately in the pursuit of the obsession for a flat frequency at one point in space, it is often forgotten that people rarely if ever sit where the test microphone was, don't even listen in anechoic rooms, have two ears that can't be in the same place, and have an auditory system that integrates energy over time before, or at least as a part of, processing the air pressure stimulus into a perception of sound.

Here's a simple mind exercise. Take an entirely acoustic experience, eg listening to a piano. Is there a flat transfer response from the piano to the listener? Is there a single point in space that the origin a transfer function could even be measured from? Does it still sound like a piano? Is there an electronic reproduction system that can capture the sound with the same fidelity as the acoustic experience?

By definition the transfer function is perfect for that experience.The irrefutable answers to this thought exercise are no, no, yes, no. Repeat for any acoustic source you can imagine and see if you can find a different set of answers.

My point is that all other things being equal, removing irregularities in the power response of a speaker system helps it sound more natural in a real room. If other things are not equal, then all all comparisons are moot!
 
Yeah, they're only so good. Nostalgia helps things sound better too.
I wouldn't say they are only good. In fact they are more musical than most of what is sold as hifi in stores today until you get into serious money; but they are fundamentally flawed as I pointed out before.

I think a good DIY measure for tweeter performance is to listen to cymbals and high hats. Every chance I get I will skulk around a drummer playing a kit acoustically, trying to absorb the natural percussive sounds, same with piano, or vibes, etc. (Luckily I haven't been arrested yet...)

I have yet to hear an Infinity Polycell tweeter, or for that matter a metal dome tweeter, that can come close to reproducing a real acoustic sound, no matter how well they "measure". Digital (CD, SACD, PCM, whatever) also struggles, typically turning a 'TISS' sound into a 'TUSH' sound, although digital has come a long way over the past 40 years.

When I owned a high fidelity store I kept some simple instruments in the audition rooms so that customers could compare real sounds with the sounds created by loudspeakers and reproduction chains. Speakers often turn a triangle's 'TING' into a 'TISH'. Speakers that can get the fundamentals right are much easier to live with.
 
I was reminding myself about the KEF acoustic butterworth design:

This is the KEF R101 with 5" driver, similar to the LS3/5A:

377516d1382147610-kef-concerto-crossover-rebuild-kef_r101_crossover-jpg


This the KEF 104ab with 8" midbass:

377517d1382147610-kef-concerto-crossover-rebuild-kef-104ab-jpg


I am fairly sure KEF arranged the electrical slopes for a smooth 18dB/octave overall.

486597d1433396292-falcon-acoustics-5ab-crossover-5ab-t27-kef-cs1-electrical-png
I wonder if there are some errors in the values of the inductors. Please see attached documentation showing the KEF Constructor Series CS1 crossover. As far as I can see there are only some extra capacitors at the input reducing the lowest bass output. FWIW: The CS1 was using the B110 SP1057 and the T27 SP1032
 

Attachments

  • Kef constuctor series CS1 corssover.jpg
    Kef constuctor series CS1 corssover.jpg
    232.4 KB · Views: 112
Last edited: