Beyond the Modula NeoD CC (split from original thread)

[Evil Twin]

I sense a disturbance in the force.... is it possible that you intend to "revive" this thread or even the unfinished business regarding the design? Even attempt some "improvements" with new midwoofers or tweeter?

Once you were the teacher, but now I am the master... I warn you, old man, from experience, that these transformative events usually go badly... perhaps you should reconsider, as this update might not be one even the Emperor would consider on a good day, while in his prime.

[---k---]

I think I just saw a ghost!

Welcome home Jon. I hope this turns out to be an extended stay.

[Bear]

Welcome home Jon. I hope this turns out to be an extended stay.

Hear! Hear! ;x( ;x( (Says the noob who spent the better part of his Christmas vacation reading through the archives here!)

[Ray Collins]

The Phantom walks among us....

Can it be???

He's back????


Ray

[Dean100]

Jon,

Its great to see you back. Wish you the best on "catching up" with life. Hopefully, you can ease back into it and also find time for your hobby.

The guys here have been holding down the fort very well. The knowledge that is available here at htguide is incredible.

Looking forward to seeing your future plans and ideas with the Modula NeoD CC.

Just recently finished some bass bins to go with my Nat P's, will do a post soon about the bass bins and how much the sound of the Nat P's have improved being high passed at 275hz to a pair of Aura NS10's. :T

[JonMarsh]

Jon,

Its great to see you back. Wish you the best on "catching up" with life. Hopefully, you can ease back into it and also find time for your hobby.

The guys here have been holding down the fort very well. The knowledge that is available here at htguide is incredible.

Looking forward to seeing your future plans and ideas with the Modula NeoD CC.

Just recently finished some bass bins to go with my Nat P's, will do a post soon about the bass bins and how much the sound of the Nat P's have improved being high passed at 275hz to a pair of Aura NS10's. :T

Thanks for the kind wishes. Yes, the Aurasound NS10s are a great buy at the moment, and with their very linear midrange performance, you have a huge amount of flexibility choosing your crossover point. Will be looking forward to your post.

~Jon

[Bear]

[..]That said, I'm still wondering if there's a little more left in the design, so I will be exploring that, with an updated midwoofer candidate and tweeters, plus some crossover tweaks.

WHEN that's done, then I'll go back to the M8ta-3, and eventually the M12ta. It's a toss whether that or the Saint-Saens gets attention first- I have all the drivers for the M12ta configurations to evaluate, some of the cabinet stuff, but also have much of the Saint-Saens materials on hand. [..]

Sounds like a great agenda. In so long as you keep the pace slow enough for it to be fun for you and affordable for me, I'll be there with you. At least for the first two. The M12ta may be too big for my bungalow, and the concept of something named after Saint Saens just frightens my foundation (both in terms of weight and in being shaken apart!).

Bill

[JonMarsh]

Sounds like a great agenda. In so long as you keep the pace slow enough for it to be fun for you and affordable for me, I'll be there with you. At least for the first two. The M12ta may be too big for my bungalow, and the concept of something named after Saint Saens just frightens my foundation (both in terms of weight and in being shaken apart!).

Bill

I understand that thought- that's the intent, after all. I have Corian panels and other bits, and if I can work up the nerve to plop down for the Aura NS10's, well, I mean, geez, I've had the RD52's and the Fountek ribbon tweeters for years, literally. Of course, their value has done better than stocks of late, so I guess it was the right call.... ;^)

I'm still expecting the M12ta to not be much bigger (footprint wise) than the M8ta. But heavy? well, taking the two piece approach should help, the bottom module hopefully won't be too much heavier than a complete M8ta. Of course, that's like saying that the small Saturn VUE SUV is at least a little under 5,000 lb- it's still damn heavy compared to my Honda Element, with less cargo room and poorer gas mileage.

We'll see. I want the NeoD CC to reach some kind of pinnacle of development just so I don't kid myself because one of these bigger projects will play louder- this is about finesse. Because of what the NeoD CC does between about 500 Hz and 3 kHz, I want the same performance in the M8ta. It will be a little tricky updating the front panel after the fact, but an aluminum plate template, my Hitachi 12V router, and a steady hand should get the job done. I'm still debating whether i'll use the RS225's I have on hand or use the set of W22's I've got lying around. In fairness, I should do a crossover for both if possible- the only major difference likely to be the resonance filter, due to different breakup modes.

[Bear]

We'll see. I want the NeoD CC to reach some kind of pinnacle of development just so I don't kid myself because one of these bigger projects will play louder- this is about finesse. Because of what the NeoD CC does between about 500 Hz and 3 kHz, I want the same performance in the M8ta. It will be a little tricky updating the front panel after the fact, but an aluminum plate template, my Hitachi 12V router, and a steady hand should get the job done. I'm still debating whether i'll use the RS225's I have on hand or use the set of W22's I've got lying around. In fairness, I should do a crossover for both if possible- the only major difference likely to be the resonance filter, due to different breakup modes.

I'm obviously with you on finishing off the Neo D! In the M8ta discussion, there was significant talk about the RS225 having the better bottom-end than the W22, and the trade-off for the better mid-range of the W22 not really being worth the trade-off. Given the addition of a mid-range into the mix, would the advantages of the W22 be even further diminished?

Bill

[JonMarsh]

I'm obviously with you on finishing off the Neo D! In the M8ta discussion, there was significant talk about the RS225 having the better bottom-end than the W22, and the trade-off for the better mid-range of the W22 not really being worth the trade-off. Given the addition of a mid-range into the mix, would the advantages of the W22 be even further diminished?

Bill

Possibly- though the the upper midbass to 1 kHz distortion of the W22 is lower.

But heck, the W22's are paid for, and they look cool... And they don't have a weird cone mode at 800 Hz, like the Scanspeak 8's.

Seriously, going back to the 8531, as mentioned above, they have an excellent motor and very low distortion; the optimum choice might be whether one is using the original enclosure size or a larger one- that's a point to explore. The trade-offs are sometimes subtle, sometimes not-
The 8531 has the best motor, then it's a bit of a horse race between the TB W6-1721 and the others. The Jantzen has a more open frame window than the HiVi, as does the W6; there are some on axis response things in the HiVi I believe are due to the reflections resulting from the relatively short rear window. The Jantzen is much better in this regard, has slightly lower distortion, and smoother more extended response, so overall should better performer in the ideal case, but 4X better? (price ratio approximately) The HiVi probably has the best bang for the buck; the TB W6 wasn't out when this design was being developed. The TB is quite good in distortion levels above 100 Hz; not quite the equal of the 8531, but at about 1/3 the cost. The 8531 may require a more complex or expensive crossover. The W6 may save on costs compared to the present version. All things to investigate and quantify. The W6 will have to be a series configuration, and will probably be sealed box.

In addition to the bang for the buck Seas tweeter, I may order the new D3004, for the ne 'cest plus ultra version. But folks, it gets a little crazy at some point. Of course, this is coming from the guy using these with a $3k integrated amplifier, and a $3500 SACD player. So what's a pricey tweeter among friends?

It's just that fundamentally, I think the Duelund approach is very interesting, and this implementation is a pretty good compromise design in size and complexity, and achieving a network design that would work with the driver's capabilities without requiring buckets full of drivers in parallel like most of Steen's designs did. And when you see what the commercial guys charge for moderate size bookshelfs, just two ways even with expensive drivers but just don't perform like this, well, a little sprucing up may be in order.

Just consider this to be like an "M" series BMW, even though I'm a Honda guy. Or, this is like the Acura version, whereas the current published design is like a high tech Accord or S2000.

Lots of things to figure out.

~Jon

[JonMarsh]

I currently live in Texarkana, AR, which unfortunately is about 8 hours away from Mobile. However, if you are ever in the area you are welcome to come by and listen to them.

This is a hoot- I went to High School in Texarkana TX.

[Bear]

Seriously, going back to the 8531, as mentioned above, they have an excellent motor and very low distortion; the optimum choice might be whether one is using the original enclosure size or a larger one- that's a point to explore. The trade-offs are sometimes subtle, sometimes not-

Definitely great points. For my own ends, I would think the point of the 8531 would be for use in a mini-tower version, rather than a monitor format since it takes a bit more air to open up the bottom end than in the stock cabinet. Following this logic, I doubt someone would want to go to the incremental expense of using the 8531s if all that they wanted was a high-performance CC.

In addition to the bang for the buck Seas tweeter, I may order the new D3004, for the ne 'cest plus ultra version. But folks, it gets a little crazy at some point. Of course, this is coming from the guy using these with a $3k integrated amplifier, and a $3500 SACD player. So what's a pricey tweeter among friends?

Hey, it's a buck cheaper than the D2904 tweeter in the Modula MTM Center! Speaking of ScanSpeak tweeters, is now the right time to whisper "Air Circ" in your ear when it comes time to update the M8ta?

It's just that fundamentally, I think the Duelund approach is very interesting, and this implementation is a pretty good compromise design in size and complexity, and achieving a network design that would work with the driver's capabilities without requiring buckets full of drivers in parallel like most of Steen's designs did. And when you see what the commercial guys charge for moderate size bookshelfs, just two ways even with expensive drivers but just don't perform like this, well, a little sprucing up may be in order.

Just consider this to be like an "M" series BMW, even though I'm a Honda guy. Or, this is like the Acura version, whereas the current published design is like a high tech Accord or S2000.

Since I fly a lot domestically (187k miles last year), I don't drive much anymore, but I get the analogy.

Bill

[powens]

This is a hoot- I went to High School in Texarkana TX.

Wow, its a small world! I actually don't live there anymore. I'm now in the DC area, but in 3 more months I'll be moving again.

I'm glad to see you back and hear you're finally getting a break.

Also, I have a quick question. What is the ideal placement of the NeoD? I recently built a NeoD center to match my L&R speakers. It sounds great overall, but it is a little heavy in the upper bass. This is especially an issue on deep male voices. I currently have it on the top shelf of my tv stand about 1 1/2' away from the back wall. My tv is mounted above the center speaker on the back wall. My L&R versions don't have this problem. So, I assume it is the placement.

[technimac]

Hi Guys,
Well, by now everyone's heard of the recession of 2008, leading to the crash of 2009. Well, it's often said that every cloud has a sliver lining- in my case, we are under severe travel restrictions to save money, and particularly in my case, travel will be cut WAAAYYY back. This is the silver lining, as I should have far fewer (if any!) weekends spent doing laundry and packing on Saturday, and flying out on Sunday. This should translate to more spare time, and the possibility of "catching up" on my life....
I'm still working on many other neglected domestic issues, but I think I'll be able to see the light at the end of the tunnel, and to that end I've ordered drivers and am setting up for a bit more work on the NeoD CC- specifically a replacement tweeter and modified crossover, and lower cost midwoofer setup test using RS180-4. I'm not going to make promises regarding the timetable- but the priority is tweeter & crossover update first, midwoofer options second.....

Jon, WELCOME BACK to the realm.....
Let's hope the Silver Lining bit will give you a chance to enjoy doing the things YOU love to do. :T
Now I actually have the impetus to finish a master bathroom remodeling job that my SO insists HAS to be done before I re-embark on my Modula NeoD LCR project. :cry:
Last February, Thomas cautioned us to "not hold our breath", that work would continue on the RS 180 version of the NeoD CC. 8O
Now that "you've resurfaced" it has given me hope that I might actually be able to morph this box of HiVi D 6.8's, RS 180's, RS 52's and D26's into the finest sounding LCR Stage this side of the 49th parallel. :W
Until then, we're thoroughly enjoying music and movies with our quad of Modula MT's.
Cheers, Bruce

[Dennis H]

Jon, while you've been gone, a few of us have been playing with the Duelund math. Here's what I posted in another thread. It makes it really easy to generate target files with the LspCAD transfer function component. It also opens the door for using filters other than LR2 and doing 'asymmetrical' designs. The 'big concept' I think is duplicating the poles (either highpass or lowpass) on all the drivers.

Back to the Duelund 3-way thing, we've been making this way too hard and
Steen Duelund, mathematician that he was, didn't help make it easier for
us mathematically challenged DIYers. I had some time today before the
baseball game and I had a lightbulb/eureka moment. I checked it out in
LspCAD and it appears to be true.

The whole thing is based on LR2 filters.

Pick the two crossover frequencies F-low and F-high.

The midrange has an LR2 highpass at F-low and an LR2 lowpass at F-high.
That's 4 poles total.

The woofer has an LR2 lowpass at F-low (2 poles). You need two more
poles to keep it in phase at all frequencies so you add another LR2
lowpass at F-high.

The tweeter has an LR2 highpass at F-high (2 poles). You need two more
poles to keep it in phase at all frequencies so you add another LR2
highpass at F-low.

That's it. Simple huh?

I haven't checked the 4-way yet but I'm guessing it's the same sort of
thing -- adding LR2s (either highpass or lowpass) at the other
frequencies to get a total of 6 poles for each driver. Woofer = LR2 at
woof/lo-mid, LR2 at lo-mid/hi-mid, LR2 at hi-mid/tweet. Etc.

[Fronn]

Welcome back Jon!

I'm loving my NeoD CC (as a center). I still want to do two more for L/R (I actually have two D26s that I picked up around the time I made the first, just in case), but it's hard to get motivated to sit in a cold garage during Minnesota winters.

[JonMarsh]

Jon, while you've been gone, a few of us have been playing with the Duelund math. Here's what I posted in another thread. It makes it really easy to generate target files with the LspCAD transfer function component. It also opens the door for using filters other than LR2 and doing 'asymmetrical' designs. The 'big concept' I think is duplicating the poles (either highpass or lowpass) on all the drivers.

Well, I would never discourage taking what ever road gets you to where you want to go, but let's recall my original description.... using LspCAD to create a target model...

If successfully implemented, the resulting crossover will have 2nd order bandpass characteristics for the midrange, and roughly flat delay Bessel 4th order functions for the woofer LP and tweeter HP. Note that depending on the value of the damping coefficient aleph, much different midrange spread and overall levels are possible- transfer functions that may be very difficult to realize with practical drivers. This acoustic target was chosen for the specific driver picked - particularly the Dayton RS52AN, which has a rather flat, wide bandpass.

This initial concept proposed a center frequency of 1500 Hz; this was revised to 1600 Hz to ease the low frequency requirements on the RS52AN. Because the 4th order LP and HP functions with this aleph are bessel 4th order functions, with the range of acoustic targets already built in to LspCAD, it should be possible to realize one of the more usable Duelund alignments in a three way configuration, without a MathCAD generated reference target. This is because Ingemar offers flat delay filters (minimum group delay) in both 2nd and 4th order. Flat delay is a characteristic of the lower order coefficient Duelund alignments. It turns out this is also a desirable approach because a "pure" Duelund acoustic function mandates that the acoustic centers of the drivers be aligned at equal distance from the listener. This is not possible with a flat baffle cabinet like the PE enclosures or other conventional box enclosures. I suspected that by using the midrange function aligned to achieve a flat net SPL response after the woofer and tweeter had been aligned to the necessary transfer function, it might be possible to achieve something very close, if not exactly congruent, to a Duelund three way transfer function. This proved to be the case, fortunately.

As a first step, the ability to realize this transfer function with "ideal" flat drivers was undertaken in LspCAD. This does not include having to use driver data with baffle step fall off and the other usual irregularities- those are dependent on the drivers chosen and the cabinet design. This is merely to describe the basic filter impelmentation and targets - much like the difference between a text book LR4 two way, and what you have to do in the real world to pull off the complete design.

The basic target functions are derived from the LspCAD Flat Delay functions- the LP target is the Flat Delay 4th order with a Fs of 1000 Hz, which hits the target -18 dB level at ~1.6 kHz, and the HP Flat Delay target with Fp set to ~2200 Hz, again hitting the -18 dB crossover target at ~1.6 kHz.

The initial midrange target is somewhat more difficult, as it must be generated from the combination of HP and LP 2nd order Flat delay targets; the LP set to 1400 Hz, the HP to 1600 Hz, and the target level adjusted for the peak at ~ -3 dB.

After the LP and HP networks have been optimized to the Flat delay targets, and the bandpass likewise for the synthesized target, response is close, but still with a dB or so ripple- this is because the flat delay target is not exactly equivalent to the chosen Duelund (aleph = 2.82, but very close. The final step is to optimize ONLY the bandpass components, and setup the target as flat response between 200 Hz and 10 kHz, to bring the midrange passband into compliance.

With the Duelund alignment chosen for the NeoD CC, with aleph = 2.82, these 4th order flat delay alignments for the high and low pass do the trick, and they're built in to LspCAD.

For other designs, if a different value of aleph is used because of different driver capabilities for bandwidth, like one of the M12ta proposals I was kicking around, then I'll probably end up importing the MathCAD data into LspCAD. The Excel sheets you did would be fine for that kind of evaluation, too, and creating target data.

You say Tomayto, I say tomahto.

[JonMarsh]

Welcome back Jon!

I'm loving my NeoD CC (as a center). I still want to do two more for L/R (I actually have two D26s that I picked up around the time I made the first, just in case), but it's hard to get motivated to sit in a cold garage during Minnesota winters.

Glad to hear you and others have built and are enjoying them- it's a heck of a pretty small box speaker, IMO, though it seems everyone has been surprised how heavy they are, considering the use of all neo magnet drivers in this version. Imagine how much more they would weigh with more conventional drivers? ;^)

[JonMarsh]

Jon, WELCOME BACK to the realm.....
Let's hope the Silver Lining bit will give you a chance to enjoy doing the things YOU love to do. :T
Now I actually have the impetus to finish a master bathroom remodeling job that my SO insists HAS to be done before I re-embark on my Modula NeoD LCR project. :cry:
Last February, Thomas cautioned us to "not hold our breath", that work would continue on the RS 180 version of the NeoD CC. 8O
Now that "you've resurfaced" it has given me hope that I might actually be able to morph this box of HiVi D 6.8's, RS 180's, RS 52's and D26's into the finest sounding LCR Stage this side of the 49th parallel. :W
Until then, we're thoroughly enjoying music and movies with our quad of Modula MT's.
Cheers, Bruce

well, there is definitely unfinished business here! have fun with your bathroom remodel- are you going to do a tile floor? It's not too hard as long as you get a good wet tile saw. And plan it carefully- I laid out the whole tile pattern in a 2D CAD program when I did one.

Let's see if we can't make a lot of silver lining out of the current world circumstances!

[JonMarsh]

Wow, its a small world! I actually don't live there anymore. I'm now in the DC area, but in 3 more months I'll be moving again.

I'm glad to see you back and hear you're finally getting a break.

Also, I have a quick question. What is the ideal placement of the NeoD? I recently built a NeoD center to match my L&R speakers. It sounds great overall, but it is a little heavy in the upper bass. This is especially an issue on deep male voices. I currently have it on the top shelf of my tv stand about 1 1/2' away from the back wall. My tv is mounted above the center speaker on the back wall. My L&R versions don't have this problem. So, I assume it is the placement.

Yes, it's a placement issue- pulling out more from the wall can help, but if you have an EQ build in to your HT processor, a bit of EQ would probably be much simpler.

Like any speaker, room placement and boundary loading have their affects- they sound different in my home than Tom's, too, as they positioned closer to the walls than he had them, a bit lower to the floor, and not with the same room modes, so the bass is smoother and better defined, and deeper, running on their own than they would do in his family room.

I have used programs like RPG Acoustics or MathCAD to calculate boundary loading issues- in general, it's a good idea to have golden mean rations between the height above floor, the distance to the back wall, and the distance to the other closest boundary (usually a side wall).

[Dennis H]

With the Duelund alignment chosen for the NeoD CC, with aleph = 2.82, these 4th order flat delay alignments for the high and low pass do the trick, and they're built in to LspCAD.

Dang Jon, I hate to be a Padawan disagreeing with the master but that's just not right. It's really not even close to a Bessel4. It's two stacked LR2 filters at the two crossover frequencies.

I wrote some Mathcad sheets (got an old copy from Danish and used this problem as a learning exercise) using the native S-plane math and used them to generate target files. Then I used those targets in LspCAD to optimize the frequencies of LR2 transfer functions. The LR2 frequencies of the three drivers were identical within the LspCAD precision and the errors from the Duelund curves were like a couple thousandths of a dB. Here are the notes I took at the time.

-----------------------------

Center frequency = 1000
a = 2.828 (recommended by Steen Duelund)
Approximate crossovers 400 and 2400 (eyeballed from the graph)
Curves exported to .txt target files

LspCAD optimization runs, optimizing Fc of LR2 filters to match the Duelund curves:

Lowpass optimized from 20 to 5K (~ -60dB)
LR2 lowpass filters at 414.2 and 2414.2
Average error = 0.001843 dB

Bandpass optimized from 20 to 20K (~ -50 and -40 dB)
LR2 highpass at 414.2 and LR2 lowpass at 2414.1
Average error = 0.001920 dB

Highpass optimized from 200 to 20K (~ -60dB)
LR2 highpass filters at 414.3 and 2414.1
Average error = 0.002069 dB

Conclusion: the Fc's of the filters on the three drivers is within the precision limit of the LspCAD active transfer functions to achieve the Duelund targets. The same Fc's should be used for designing all three filters.

[Dennis H]

Just a followup with a pic I had saved. Duelund is in black, Bessel4 is in red. Fc is adjusted so they're -6dB at about the same frequency.

[technimac]

have fun with your bathroom remodel- are you going to do a tile floor? It's not too hard as long as you get a good wet tile saw. And plan it carefully- I laid out the whole tile pattern in a 2D CAD program when I did one.
Let's see if we can't make a lot of silver lining out of the current world circumstances!

Tile floor? Did I write that between the lines?
Yep, I'm installing under-tile heat mat, 12" tiles on the diagonal in the center with soldier tiles around the perimeter. Didn't do the layout in CAD, just old-fashioned graph paper, because it's not a big space.
You're dead-on about the wet saw being an absolute necessity with all those diagonals to cut :W . In fact, at the moment I'm calling the wet-saw table my "office".
Also included is a raised, drop-in tub with tiled deck & step as well as a new door to be installed for the existing full-tiled shower.
And to make the job complete, tile is going up the wall to a slate border that's topped off with a twisted & braided rope cap - lotta fun running that baby around corners, trying to keep the pattern matched 8O.
Having a chance to get back to the Modula Neo D project is putting a glimmer on my horizon.
Cheers, Bruce

[TNRRClassic]

Wow have I been gone a long time. With job changes and a move, I haven't built a speaker for something like 1 1/2 - 2 years??? I was desperately wanting to build a trio of these speakers, when the big changes occurred. I have a US made tweeter, and was going to buy a pair of the Chinese ones for the L & R, but never got around to it. I guess now I have to wait for Jon to come up with something to replace the Vifa tweeters.

The great thing is that Jon may be able to spend a bit more time doing things other than his job.

[Brian Bunge]

Wow Jon! Great to see you back here! I'm glad you appear to be in good health and doing well.

I need to PM you about that defective tweeter I got from you!

[JonMarsh]

More Thoughts on Duelund Crossovers & Implementation

Well, as it turns out, my return to trying to post and work on projects may have been a bit premature considering what’s happened work shortly after that. I’ve had some weeks of consecutive business travel, and staying over at ThomasW’s before doing a “touch and go” in San Jose and leaving for another week on the road.

Be that as it may, I’d like to address the questions raised about how and why I described the development of a target transfer function using the Duelund concepts for the NeoD CC, and explore my perspective on the Duelund concept.

This post will encompass my thoughts about the why the Duelund crossover concept can be very useful and interesting, as well as visiting Steen’s own comments.

I’m going share my thoughts about the specific challenges of this approach in a smaller speaker, and what trade offs I chose to make. I’ll include a bit more detailed review of the Duelund concepts and characteristics, and why/how the LspCAD target example I described fits the particular Duelund crossover example I chose for the NeoD CC.

Please understand that I don’t consider anything about my description of my approach as “definitive”, other than explaining why I made the choices I did, and how I think they furthered the achievement of the spirt of this approach in a relatively simple, low cost implementation. (With regard to a Duelund design approach, one could say that’s like saying I’m trying to do a simple low cost V-12 engine, though…. There are certain conditions to the approach which are definitely demanding from various aspects of the speaker system.
Last, I would really be interested to hear about any other completed Duelund designs with outlines of the designer’s methodology, starting assumptions, and details about the end results. I’m looking at 2-3 more designs besides the NeoD CC trying out this concept in one form or another, including a possible “Indra” inspired project.


Equations and transfer function:
The Duelund style crossover, invented by the late Steen Duelund, (often referred to as the Mad Loudspeaker Professor) is a mathematic derivation for a three way crossover, using a shared damping coefficient aleph for each section, which sums in all cases to unity all pass, with a single phase rotation from the low frequencies to the high. By varying the value of the aleph coeffficient, different response types regarding the Low Pass and High Pass alignments and bandwidth of the midrange and it's level are described. With decreasing levels of aleph, the midrange element narrows in the range it covers and in level, and the network response can devolve to just a two way LR4 if the value of aleph is equal to the square root of 2 (1.414)

(Full set of equations, Fig 1)



In MathCAD the function HSum(s,α) defines the complete transfer function in terms of the frequency variable “s” and the damping coefficient “α”, with the following functions describing just the low pass, band pass (midrange), and high pass function.

All three functions (low pass, band pass, and high pass) are in relative phase with each other at any frequency, with a net total phase rotation of 360 degrees from the low bass to the treble, just as the phase rotation of a two way LR-4 crossover. In fact, as noted above, with aleph at 1.414, the three way devolves to a LR4 two way. High and low pass are 4th order functions, but with changing aleph, the pole separation changes and much slower initial rates may be seen at the turn-over frquencies- with some values of aleph, this INITIAL rate may be close to an LR2. The ultimate slope is still 4th order, though.

In the next figure the amplitude plots for LP, BP, and HP are plotted with several coefficient values: the square root of 2 2, 2, 2 X the square root of 2, and 4. This clearly illustrates the progression of the midrange bandpass level and width, along with the change in LP and HP alignments as the Q is lowered.

The case for α = 2 results in the sharpest cutoff slope that has no ringing in the step function, but the peak of the bandpass level is about -6 dB, and the HP and LP functions crossover at -12 dB; this requires a fair amount of bandwidth of the woofer and tweeter, but minimizes the workload on the midrange.

The case for α = twice the root of 2 is the variation I tried for the NeoD CC; the midrange level rises to -2.5 dB, and the crossover of the LP and HP are at -18 dB. The summation of the LP and HP at 1800 Hz adds to the midrange output to bring the level to unity- as long as the correct phase relationship is maintained in the realized speaker.
The last example for α = 4 shows the midrange/bandpass level reaching within 1 dB of the unity level, with reduced requirements of the bass and treble drivers. But the requirements here on the midrange are quite substantial- considering what is required for relatively flat summation, uniform response seems required to about the -15 to -18 dB level- which for the midrange is at least 200 Hz to 13 kHz. Few drivers can approach this requirement.

The red marker lines at several dB points (-12 dB, -18 dB, and -24 dB show the LP and HP crossover points as the alignments produce a wider and higher bandpass level, but at the same center frequency. This example was created with a nominal center frequency of 1600 Hz.

(Figure 2).



What does the phase response look like for these variations in the damping coefficient? This is shown below, where the a function extracting the phase from the transfer function is plotted for the same four sets of coefficients. The trace colors used for each coefficient example are the same as shown in the graph above- note that the phase shift shape for the square root of two α is the same as for the conventionally calculated LR4. With higher values of α, the phase turn is slower through the transition region.

(Figure 3)




That the bass, midrange, and treble are in the same relative phase at all frequencies is shown in the following graph, plotting the extracted phase for LP, BP, and Hp for the case of α = 2 X the square root of 2. The three plots overlay each other.



An interesting question that arises is how closely must a practical speaker follow the LP, BP, and HP calculated transfer functions for satisfactory performance? This was especially a concern for trying to develop a box version with a flat baffle, as the use of separate baffles for each frequency range with the offset manually adjusted, as Steen advocated, isn’t feasible.
Speaking of Steen, let’s review his summary of the advantages and disadvantages that he saw in this crossover path -



ADVANTAGES:
The summed amplitude response is always exactly 1.
The units are in phase with one another at all frequencies if “α” is greater than or equal to the square root of 2.
The step response for “α” greater than 2 is without ringing in the step response.
The turn of phase can be chosen as the starting point (center frequency of bandpass).
The amplitude slope is rounded, slowly reading the highest slope. The higher the “α” value, the slower the rounding.
The driver units have great overlap (correctional network is unavoidable).
When playing it is impossible subjectively to identify the single driver units as separate sources of sound.
It can be realized in full and doesn’t demand more of the drive units that it should be possible for them to deliver (depending on the value of “α”)

To this I would add that as designed, the drivers are always in phase, not in phase quadrature, as with some crossover alignments, so there is no off axis lobing or perturbations in the power response which occur for example, in the 1st and 3rd order Butterworth.

DISADVANTAGES:
The displacement of drivers to one another is critical as they work in phase. (Steen wrote: )” They must fit within 1 mm from listening position. You are therefore forced to house the units in three separate cabinets, so you can optimize the distance to your listening position.”
To this I would also note that it’s really maintaining a relative alignment of each driver relative to the phase angle caused by an offset; obviously this is far more critical distance for the midrange to tweeter transition than for the midrange to woofer transition; however, at the center frequency, all drivers must be operating with comparable phase for correct summation to occur.

(Figure 5)



It’s pretty obvious that with most available drivers following the calculated transfer function throughout the whole frequency range is not feasible- but how much is sufficient? Considering the overlap and levels required for fairly complete summation, my judgement is that the final acoustic output should track down to at least the -18 to -20 dB level. For the alignment with α = 2.828, I expect the bounding box below in green shows the required region for getting good summation, requiring good amplitude and phase tracking. This took some effort but the midrange summation was largely achieved, and verified by the comparison of total frequency response versus the measured midrange level. Fortunately for the midrange, the overall output levels are reduced, but still requiring a wide range. Possibly a narrower midrange alignment should have been chosen, but that may have posed problems for the chosen tweeter or the 7” woofers.
In summary, the key features of the Duelund concept which intrigued me was the achievement of a three way system in which the drivers were in the same relative operating phase over the critical regions of LP, BP, and HP summation, with only one relatively gentle phase turn of 360 degrees, and the possibility to have a step response largely free of ringing with moderate slopes.


LSPCAD Trial
The straight forward way to generate the amplitude targets in LspCAD would have been to export them from MathCAD in a text form file, which is readily done- but inspection of the high pass and low pass for the alignment I chose indicated some standard available functions may suffice. Verifying this in practice would give other LspCAD users an opportunity to examine this concept without requiring either MathCAD or Excel to generate targets, a potentially useful outcome. Examining the calculated LP amplitude curve, the attenuation is ~2.5 dB at 400 Hz, -11 db at 1,000 Hz, and -18 dB at 1600 Hz. Examining the LspCAD target functions, a fairly close fit in the critical midrange area is the Flat Delay 4th order with a turn-over frequency of 1 kHz; the level is a bit too high at the 400 Hz area, so the required amplitude of the midrange may be reduced there, which could be of benefit. Considering the the “critical” bounding box as described above, the behavior below the -20 dB level was not considered very important. This match was not completely exact, but close enough that I considered it worth evaluating- it would alter the curvature of the phase and amplitude slightly in the lower midrange, but could ease the workload on the midrange for flat summation.

(Figure 6)



The high pass target selected was a flat delay 4th order with nominal fc of 2200 Hz, again for the same reason, the relative levels in the area of midrange summation, with a -18 dB point at the design center frequency.

(Figure 7)





The midrange targets used for component calculations were chosen just to get component values in the ball park, as previously described.

(Figure 8)




The final midrange component values were optimized then to achieve a flat transfer function with the LspCAD optimizer. The results are shown in Figure 9, with the plotted LP, BP, and HP functions, and the summed response.

(Figure 9)




Not quite a perfect summing, with the tiniest bit of ripple, but then with real loudspeaker drivers and cabinets you would be incredibly charmed to get even close to this. The total phase transfer function follows a single slow 360 degree all pass function as expected, and most importantly, summing behavior in the critical region appears to be quite good. As such it appears quite usable for the system development, so I went ahead with this as an example anyone could use with just LspCAD, and with the value of “α” I think may be most commonly useful with available drivers for a three way system as an example.

If anyone thinks this is blasphemous, well, understand that I was trying to make a typical example more accessible. Using these targets with slightly shifted poles and zeros for the targets, but achieving the same general behavior with good summation is, in my opinion, an acceptable pragmatic modification for engineering utility.

Of course, the final networks have to include driver equalization and baffle step correction, as well as some adjustment in phase to compensate for the difference in acoustic centers from the baffle for the woofers and midrange.

My experience listening to these in my living room the last year and half has been interesting; they respond well to improvements in associated equipment and sources. Just last month to do some critical listening tests with DAC’s I even brought out my big gun speaker cables, “just in case”, and was surprised at the further improvement in focus and dynamics.

I’ve got a couple of more projects in development for which I plan to use the Duelund approach for three ways, including an update to the M8ta to a three way, the M12ta, and a new three way tower idea for smaller rooms influenced by the Peak Consult Zoltan and Dragon.







Think of a configuration like the Dragon, but driver sizes like the Zoltan.

[mkc]

Hi Jon,

Great post.

Actually, it is a funny coincidence. My son and I are in the process of making a Peak-consult Empress look-alike, but with cheap drivers, I might add. He think the black leather look cool. I have build some test cabs.

It' going to be a fully active design with LM3886 chipamps (bridged for the low frequency). Op-amps are going to be Analog Devices OP275. Good basic opamp. I used to work for Analog Devices, so I have a sweet spot, you know .

Best regards,
Mogens

[JonMarsh]

What drivers do you intend to use in the Empress Look-alike?

I've used AD parts quite a bit before- my favorite was the AD844, but it could be a bit of overkill in an electronic crossover- it makes a pretty good follower to the D/A stage of a DAC, though.

My "miniature" Dragon look-alike will be using Tang Band W6-1721 for midwoofers, W5-1685 for midranges in the MTM part, and probably the pair of D3004-662001 I have on hand for testing. I really want to see what they sound like. Sort of a sacrilegious combination, but I have some hopes, due to the fairly low distortion of the TB's, and their smooth extended response. Will make the network easier. Let us know how your project with your son turns out!

IFX is working on some new Class D chip amps, which have the possibility to put driver EQ and crossovers in FIR filters internal to the chips- I'm hoping to get some demoboards to try out in April or may, so I may be looking at an active three way attempt sometime, too, depending on what the Munich team shakes loose for me.

[mkc]

Hi Jon,

Yes, the AD844 is a great I/V converter. I used the AD811 in my CD player for I/V duty. I have a small collection of AD op-amps, so I might as well use what I have. I get the impression that we are a little bit alike - we tend to get more stuff collected than we get around using

As said, we are going to use some cheap drivers and the Peak-Consult Empress was chosen purely on looks.

The current plan is to use 27TFFC and a pair of 5" and 8" OEM drivers. I have no idea how well they perform, but I think they will do great for his purpose. Hopefully he will grow up and acquire a better taste of music. Kids... :roll:

The reason for going active is twofold. First, he has a network streamer, which provides a line level signal out. Second, given the fact that the drivers is "what we have" and properly have different sensitivity also makes active seem like a good idea. We will see once we get some measurements done.

Anyway, I'm looking forward to your more serious attempt at a Peak-Consult look-alike.

Best regards,
Mogens

[CraigJ]

IFX is working on some new Class D chip amps, which have the possibility to put driver EQ and crossovers in FIR filters internal to the chips- I'm hoping to get some demoboards to try out in April or may, so I may be looking at an active three way attempt sometime, too, depending on what the Munich team shakes loose for me.

Any updates on the IFX chips? it pans out, this would be most excellent. Also, any plans available for pming of the Avalon clone?

Craig

[JonMarsh]

Hello Craig,

A new silicon tape out "low power" version of the chip that is a system evaluation is being evaluated right now; the final silicon is scheduled for this fall. Discussion is ongoing with customers, including system architecture. I'm not officially involved with that group, but it's part of our overall division, and they've asked me to participate with inputs and later evaluations.

For me, that isn't tied to this activity at all, as I'll be doing a full passive crossover implementation, based on component selections and evaluations done with the NeoD CC. That means a lot of Jantzen caps, in various grades, this time.

The 2D sketches are being fleshed out, and based on my minion completing the driver testing soon, the detailed 3D model will be prepared soon- this is the critical element, as with my CAD software's model to sheet tools, I can prepare very comprehensive 2D build plans from the 3D model.

Pele prepared a 3D model based on the Soup concept of design for the SDX-2 Ardent project, but there are many area's which I will be doing things much more the way Avalon does than the Soup version does. Pele may build his based on the Soup, but then again, maybe not.





The front baffle must be cut out in various ways behind the drivers, and the bracing and internal baffle details will be different.

The devil is in the details. There will be two versions, one for a "budget build", and one where cost is not as serious a consideration. My plan was to have target transfer function definition and evaluation completed this week, driver testing completed by the end of this week, CAD drawings by 1-2 weeks from now, and the first test cabinets in the end of June. I will be getting the last of the new drivers to test on Friday, so we will probably be able to stay on track. :W The final definition of the target transfer function will await the results of midrange testing this week.

[JonMarsh]

Hello Craig,

A new silicon tape out "low power" version of the chip that is a system evaluation is being evaluated right now; the final silicon is scheduled for this fall. Discussion is ongoing with customers, including system architecture. I'm not officially involved with that group, but it's part of our overall division, and they've asked me to participate with inputs and later evaluations.

For me, that isn't tied to this activity at all, as I'll be doing a full passive crossover implementation, based on component selections and evaluations done with the NeoD CC. That means a lot of Jantzen caps, in various grades, this time.

The 2D sketches are being fleshed out, and based on my minion completing the driver testing soon, the detailed 3D model will be prepared soon- this is the critical element, as with my CAD software's model to sheet tools, I can prepare very comprehensive 2D build plans from the 3D model.

Pele prepared a 3D model based on the Soup concept of design for the SDX-2 Ardent project, but there are many area's which I will be doing things much more the way Avalon does than the Soup version does. Pele may build his based on the Soup, but then again, maybe not.





The front baffle must be cut out in various ways behind the drivers, and the bracing and internal baffle details will be different.

The devil is in the details. There will be two versions, one for a "budget build", and one where cost is not as serious a consideration. My plan was to have target transfer function definition and evaluation completed this week, driver testing completed by the end of this week, CAD drawings by 1-2 weeks from now, and the first test cabinets in the end of June. I will be getting the last of the new drivers to test on Friday, so we will probably be able to stay on track. :W The final definition of the target transfer function will await the results of midrange testing this week.

Just to indicate something more representative of what the SDX-2 Ardent will be configured, here's a screen shot in the current CAD development process. The internals are NOT based on the Soup clone, as Pele drew above (which in turn is based on the Avalon Opus Ceramique), but rather on structural details and methods more derived from the source. There will be detailed piece part drawings prepared, so as ThomasW says, I'm pushing too many pixels and not making enough sawdust. However, it's wise to plan once, design twice, measure three times, and cut once. :W

This is only a screen capture during design, not a rendering, but the differences should be obvious.

[Jed]

Looks very elegant, Jon. How are you going to cut the facets?

By the way, based on your reviews and a bunch of others I picked up an Ayre AX7e for a good price. Should be here next week for my listening pleasure.

[JonMarsh]

Custom jigs and a saw that cuts deep.... I've done this before, first time in the 70's. :W

Congratulations on the AX7e- I got mine used nearly brand new at a good price, too. I've been tweaking the power supply and regulators a little, but it's very nice out of the box as is. It uses the same devices as output transistors I was using as drivers in Pro sound I designed back in the 70's. Fairly fast. :W I recommend using balanced sources/cables if you have them- it is true differential balanced input to output. It does reward good cables- more than you might think- I took the Cardas Golden reference speaker cables and Ayre signature interconnects out of my main system, put them in with the Ayre when I was doing DAC A/B's recently, and ended up leaving them there- now I need to buy more cables. :W

Another curiosity- Cambridge Audio remotes use simlar codes, so my controller for the 640C I use as a CD transport will do the volume and input selection, too!

[bigg]

Hi Jon,
This new one looks awesome. So you've cut the facets before. Well, this will be an interesting challenge for this wannabe. When I printed out the picture it called it a Zoltan. What is the new project going to be called? Will it have it's own thread?

[JonMarsh]

The new project will actually have a web site, for a "first" for me, just because posting all the information in a forum manner would be such a b*tch. :W Not that I won't do some summary stuff in a post, but since I plan to document the piece parts in CAD and so forth, doing it by web site with the tools I use (Rapidweaver) is MUCH easier, especially with a lot of graphics information- easier to work on, too, as it's going together already. Just the difference in uploading the graphics amounts to hours of work.

The last two weeks of June I'm off to work on this project, so hopefully I will have some things to report and post in a few weeks. There's been a lot of prep work leading up to this, my goal is to get these finished and bring them to ThomasW's during RMAF if possible.

~Jon

[bigg]

New Zoltan or Zoltan II

Hi Jon,

Yeah a website sounds great! Sometimes it's so hard to find what you're looking for in the forum threads. I've actually spent hours staring at the computer screen to find a posting that reads in 30 seconds. Crazy huh?

Even though I don't have near the knowledge of a lot of you folks here and at PE forum, I'm very busy with several projects. Someday I hope to post some pictures of my speakers. Most of them will be gone soon.

I'm really interested in doing something like the Zoltan. I thought your project was going to be more like a small Dragon. That sounds cool too. I'm looking forward to your next great project. 8)

[bigg]

New Project

I think this is going to be an awesome project that many will want to build. :T The front baffles look to be the greatest challenge. How thick are those front baffles going to be? :?:

[JonMarsh]

They are laminated/layered up from materials which will add to a total of 4". I have a saw with an unusual motor and pulley design, small pulley and broad flat belt, which does fairly deep cuts. Even with that it will take some custom built fixtures.

This project will be documented in a depth I haven't done before, other than the M8a bookshelf article for Audio Xpress magazine. When I was doing the M8ta, I didn't expect anyone else to build them, and so didn't spend much time documenting the construction process, though of course I had my CAD drawings prepared from 3D models which were adequate for me to work from. I think over a dozen sets of these have been built to date, something that rather surprised me.

You can get some clues about what's involved by reviewing the M8ta thread, but there are many, many details changed for the design and construction of this project.

• The bracing system is quite different, not resembling the M8ta or the Soup Avalon project (done by someone else; Pele's favorite), and should be both simpler and stronger.
• The front panel is at the core of the project behavior and construction, and will be fabricated in steps, for my premium working pair with a slightly more exotic materials approach than just MDF.
• Crossovers will be internal, and the bottom removable; as with the M8ta, the cabinet will use a relatively low tuned downward firing port.
• Rabbeting, fixtures, and braces will be used for dimensional stability and trueness in the assembly of the main cabinet.
• Threaded inserts will be used for mounting all drivers and the bottom panel.
• Veneering is planned at present to use a mixture of techniques. Of course, veneering is optional; the old truck bed liner technique or satin polyurethane over black paint may be a much easier and adequately protective road for many. I may do the RS180/RS100/27TBFCG versions that way, just because of how time consuming veneering is, and to have a working example.
• The same midrange configuration will be used for both cabinets, with the RS100 mounted in an aluminum plate to avoid rear masking issues.

Anyone considering this project after I (hopefully!) successfully complete and document it should consider one point carefully- the typical DIY'er seems to look at the driver expense as being the main cost for a project, whether budget, midrange, or high end. If you get to a point where you're truly looking for finesse and not just sheer dB of output, and doing something with more construction complexity than a standard box, the drivers are only a small part of the overall expense. In this project, which ever set of drivers and crossovers you select yourself, the tools needed, including clamps, routers, router bits, saw, custom built fixtures, etc, are not a small expense. IF one is already a serious wood working hobbyist, then that portion may be relatively incremental. But for many it won't be. This won't be a project for a hand held circular saw, a jig saw, and a $75 dollar router from Home Depot. :W Of course, like the NeoD CC, the crossover will be a substantial expense- Steen Duelund always said to budget more money for the crossover cost than for the drivers. In a two way, that can often be avoided; in a three way, with good quality components, it seems inevitable.

On the web site in development I will be spelling out what I used in the way of tools and fixtures, and a little about my reasoning; I'm trying to keep cost reasonable while at the same time assuring productivity and high quality results. Since this is DIY, it's always a given than constructors may innovate in tools and techniques, in all likelihood come up with better ways to do things than I thought of. I'm not a professional wood worker or craftsman. :W OTOH, I built my first "serious" woodworking speaker projects in the early 70's, walnut finished "clones" of the AR-LST that were quite attractive and nicely done if I say so- certainly by the standards of that time. This is what the real AR-Lst looked like, if you're not familiar- unfortunately I don't have any pictures from that time- didn't even own a camera- most of my money was tied up in my band equipment.



I have a long history of doing speakers that are not straight boxes. :W My wife made me get rid of the last ones I had in the early 80s, she couldn't deal with the size (dual 10 woofers, four way with 6" midrange, dome midrange, and 3/4" MB dome tweeter, in a faceted front panel cabinet; front panel was about 4" thick. Anything sound familiar here? By the way, I got rid of those by giving them to Charles Hansen, who used to work for me when he was a physics student at CU. The rest is history.

For that time, and my age (very early 20's), the AR-LST "clone" was a sophisticated project to do, using Long dome mids and Long tweeters. It's not a beginner project; the SDX-2 Ardent isn't either, unless one is really up for a cruel learning curve! :W

[bigg]

SDX-2 Ardent

Thanks Jon for your detailed response. I know it's a steep learning curve. All of my speaker building has been a real uphill climb. But with fantastic rewards! I was curious about the front baffle because I thought I could start building up the front baffle to get the thickness I needed while I'm working on other projects. And to have it ready for cutting when the time came. My tools may not be up to the task, but I have a friend with a real large and seriously equipped woodworking shop. He has offered his assistance before. The only problem is that he travels a lot. I also plan to better equip my shop eventually. Thank you. :thanku:

[---k---]

Jon,
You better watch it or our big bad moderator is going to have to yell at you to start a new thread.

[ThomasW]

Yeap.....

[bigg]

SDX-2 Ardent

Hey I found it. I thought the whole thing disappeared for a moment. I was wondering what materials were going to be layered for the new 4 inch baffle. Has that been determined yet? Thank you for your time.

[dwk]

I'm also excited about this project, and give a hearty thanks-in-advance for the level of effort and documentation you're putting in to this.

I think your cautionary note is one that could well be factored out and made a sticky. Even looking at the 'budget' version, materials look like they'll run at least $1k per pair, possibly more depending on final xover. This is in line with the NeoDCC, but still may be a bit surprising to some.

For my part, I'm looking at this project as something of a stake-in-the-ground. I've been experimenting with my own projects for ~15 years or more, but have never produced a 'finished' product. Time to put up or shut up, and the fact that it's ambitious is of course part of the attraction. Plus, it's a excuse to finally get a vacuum press :T Work won't probably start in earnest until the fall, but there's lots of shop prep needed before then anyway.

They are laminated/layered up from materials which will add to a total of 4". I have a saw with an unusual motor and pulley design, small pulley and broad flat belt, which does fairly deep cuts. Even with that it will take some custom built fixtures.

Can I interpret this as an indication that the BT3K will be sufficient for the facets? The fact that you used one for the M8 project was (I think - memory is getting fuzzy) a part of what drew me to the saw in the first place.

[ThomasW]

Jon's going to use a techinque similar to the one ColoradoTom used to build his M8a-TA

https://www.htguide.com/forum/showpost.php?p=250091&postcount=112