6th Order L-R. I find it to be an easy to hit slope

[fjhuerta]

Lately, I've found out I can easily hit 6th order L-R, quite easier than I can get to 4th order...

I mean - 4th order is doable, but using a couple more pieces, 6th order looks a lot flatter and easier to work with.

The topology is quite simple. It's a third order electrical and a conjugate, on both woofer and tweeter. I have to invert the polarity of the tweeter. That's it. I get very good phase tracking and an almost perfect response.

My question is - why is it that I don't see it mentioned too often? Is this topology somehow wrong? Has anyone else used it before? Opinions?

[JonMarsh]

This always depends on the acoustical roll off of the drivers, and what electrical filter you have to combine with the drivers to get a net acoustic target transfer function.

Just becuase something isn't widely used, is not a reason to not investigate it- most often it may be becuase of cost reasons.

For example, in many commercial speakers, drivers are used as close as possible to their bandwidth limits, so an LR-4 may be realized with only a 2nd order electrical. This keeps costs down- but doesn't provide very much power reduction to the driver, and is using it in a range where it doesn't function very well- sound commerical practice, I suppose, but perhaps not the best audio practice. I'm sure arguements could be made both ways.

In the case of the Modula MTMs and some other designs I do, a CE filter is used to realize an 8th order LR for the first 50 dB of attenuation. I also experimented with only using a 6th order, but the component count was the same, and it doesn't offer the same protection to the drivers. But that doesn't mean it's not usable in a given application.

With what you describe, combining a 3rd order electrical with drivers, I'm inclined to question why/how you are using drivers in the transistion band when there own roll off is greater than 12 dB/octave? Or are you acually using the conjugate network not just for impedance flattening, but also to add some roll off? The latter sounds likely, based on my understanding of filter topologies.

[nerd of nerds]

This always depends on the acoustical roll off of the drivers, and what electrical filter you have to combine with the drivers to get a net acoustic target transfer function.

Just becuase something isn't widely used, is not a reason to not investigate it- most often it may be becuase of cost reasons.

For example, in many commercial speakers, drivers are used as close as possible to their bandwidth limits, so an LR-4 may be realized with only a 2nd order electrical. This keeps costs down- but doesn't provide very much power reduction to the driver, and is using it in a range where it doesn't function very well- sound commerical practice, I suppose, but perhaps not the best audio practice. I'm sure arguements could be made both ways.

In the case of the Modula MTMs and some other designs I do, a CE filter is used to realize an 8th order LR for the first 50 dB of attenuation. I also experimented with only using a 6th order, but the component count was the same, and it doesn't offer the same protection to the drivers. But that doesn't mean it's not usable in a given application.

With what you describe, combining a 3rd order electrical with drivers, I'm inclined to question why/how you are using drivers in the transistion band when there own roll off is greater than 12 dB/octave? Or are you acually using the conjugate network not just for impedance flattening, but also to add some roll off? The latter sounds likely, based on my understanding of filter topologies.

Do you have any reading or articles you could recomend for CE filters? i'm very curious about them (haven't designed an xo yet, but still would like to read about them for future projects).

[Dennis H]

LR6 is Dan Wiggins' favorite.

[JonMarsh]

Do you have any reading or articles you could recomend for CE filters? i'm very curious about them (haven't designed an xo yet, but still would like to read about them for future projects).

Cauer-Elliptic filters are used more in communications applications than in audio-

Do a search in the Mission Possible and you'll come up with lots of discussion- the search function is your friend. I know of no book written describing CE filters for speaker crossovers. I first used CE filters in Audio for class D amplifiers in the late 80's.

[fjhuerta]

Jon, the conjugate is used for helping roll-off a bit on the woofer.

I suppose that since I try and use low crossover points on the tweeter, I already have a 12 db/Oct or higher roll-off, so hitting 6th order is simple with a 3rd order electrical. I use the CR there for a bit of high end EQ.

Thank you very much all for your info - I'll try to compare a 4th order and a 6th order on an upcoming project.

[Jed]

Thank you very much all for your info - I'll try to compare a 4th order and a 6th order on an upcoming project.

Please share, what are you working on?

[fjhuerta]

It's an MTM tower. It used to have a Neo3 PDR and a pair of Focal 5N421 woofers.

I tried working with the Neo3's, but could never get any sort of useful measurement out of it. So I'll be modifying the baffle a bit, and will use the same Focal woofers with the titanium MB Quart tweeter from Madisound.

I don't know how low I can make the MB Quart work, so maybe a 6th order filter will help me get to 2 KHz, at least.

[JonMarsh]

The MB will be quite clean down to 2 kHz; the FR supports going lower, but the distortion performance doesn't.

Since the response of the tweeter goes somewhat low, getting a 6th order response without a lot of components may be hard- you might as well look at the CE design used in the Modula MTM (Missions Accomplished). OTOH, I'd think carefully about taking an MTM up to 2 kHz; classic theory says the CTC of the outside drivers with a 2 kHz crossover should be no more tha 7"; that would indicate the basket to basket distance on your 5's should be under 2"- the MB isn't that small. I susepct you'll wind up with a CTC more like 10", which would indicate no higher than 1400 Hz. With an LR8 equivalent CE like the Modula MTM, that should be doable. (the Modula MTM is at 1200Hz, using 7" midwoofers).

I'm talking about ideal conditions, of course.

[fjhuerta]

The MB will be quite clean down to 2 kHz; the FR supports going lower, but the distortion performance doesn't.

Since the response of the tweeter goes somewhat low, getting a 6th order response without a lot of components may be hard- you might as well look at the CE design used in the Modula MTM (Missions Accomplished). OTOH, I'd think carefully about taking an MTM up to 2 kHz; classic theory says the CTC of the outside drivers with a 2 kHz crossover should be no more tha 7"; that would indicate the basket to basket distance on your 5's should be under 2"- the MB isn't that small. I susepct you'll wind up with a CTC more like 10", which would indicate no higher than 1400 Hz. With an LR8 equivalent CE like the Modula MTM, that should be doable. (the Modula MTM is at 1200Hz, using 7" midwoofers).

I'm talking about ideal conditions, of course.

Hi Jon!

I've never worked with CE filters - this could be a good project to try them. Do you think the MB's can support such a low frequency using LR-8?

[JonMarsh]

Yes, I do- Avalon originally ran them down to 1 kHz with just a LR-4 in the Eclipse loudspeaker. That was too low, IMO, but even so they sounded pretty good unless you pushed them. Linkwitz runs the Millenium Excel down to 1400 with an LR-4. The MB doesn't quite have that much Xmax, but you'd be surprised what it will do- it's just that the motor has a little more 2nd order non-linearity than we'd like by current standards for SOTA when you push it below 2 kHz. HD3 is a pretty good guide where to avoid pushing it- and here it's still pretty good when you consider that the big "bump" is just hitting -50.

[fjhuerta]

Thanks a lot Jon! In a couple of weeks (when the cabinets are ready) I'll begin working on this project.

[fjhuerta]

Project complete!

Here's the project. L-R 6 beat out L-R 4 without a problem.

The drivers were:

Woofers: Focal 5N421
Tweeter: MB Quart

(MTM Configuration).

The woofers were a bit.. strange to work with. They could get pretty high without any roll-off, and the very slim size of the baffle meant I ended up with an extra 2 or 3 dB at around 1KHz.

Likewise, I decided to add a bit of a "BBC Dip" to the speakers, and optimize them for off-axis response (since I'd be using them in my living room, and people aren't always seated exactly in front of them).

Here's the FR. It has an upward slope so that



the power response looks a bit flatter.



Impedance means a tube amplifier wouldn't mean such a good idea, although it never drops below 4.5 Ohms, or so.



Phase tracking at the x-over point (2KHz) is pretty much OK, I guess.



X-fer function is a bit strange. I had to give the tweeter a bit of a bump with a high Q to flatten its response. The only thing I can think of is that diffraction is the culprit here.

[fjhuerta]

The circuit is here.



And the finished speakers (bear in mind - this is a 3 year old cabinet, modified to accept the new drivers... so it doesn't look that good (well, it doesn't look as bad as the pics show, either!).

[fjhuerta]

Has it been 4 years already? wow.

I got the speakers back - they look exactly the same. I used to love their slim design, now, not so much. But still, I gave them a listen.

I suppose I've learned quite a bit since that time, so I'm planning to redo the crossover completely. It's going to be a fun and simple project, and I suppose, I'll learn from what I used to do wrong.

I do not like at all the transfer functions, or the impedance plot - that tells me I did something wrong. So that's where I'll start.

Why didn't I listen to Jon when he said I should have gone for something closer to 1.4 KHz?

Let's see what I can do... as I said, it's a great opportunity. These were my first DIY speakers, and it was pure luck that brought them back to me... for $0.

[fjhuerta]

Well, it's over. I finished the speaker...

6th order is pretty nice. I had forgotten all about the filter. I managed to take 3 or 4 pieces out of the original crossover, and saw that, during all these years, I've really learned to simplify things

The sound didn't change that much - well, it did open up a bit (I had used a non-calibrated mic when I built them). They aren't "audiophile" speakers - not to me, anyway, anymore... funny how things can change in 5 years!

[---k---]

That is cool that you go the opportunity to go back to a project after all these years. I wish I could hear the before and after.