A K-402-Based Full-Range Multiple-Entry Horn

[ClaudeJ1]

I was wondering how you got 15's to fit. Thanks.

[Marvel]

Yup, that's what I thought you did. Bill Waslo has this in his spreadsheet unity horn designer.

Very nice, Chris.

[Chris A]

It's actually pretty easy, once you locate the anchor points on the horn: that part's a bit tricky.  The Crites CW1526C woofers mounted to their MDF pads using 14x1 1/4" self tapping screws.  Modeling clay fills any gaps in the ports between the horn and the mounting pads. You can also use epoxy if you wish to make it permanent.

 

Chris

[DrWho]

Uhmm...not in the case of the dual diaphragm BMS driver.

 

 

I believe the total radiating surface area increased when he went from the coaxial BMS to the BMS Mid + K69 tweeter.

 

The coaxial diaphragms also need to deal with the air pressure generated by the MF driver showing up right in front of the HF driver. This causes modulation of the HF unit's coupling, which is a form of modulation distortion. When the two drivers are separated into separate horns, then you don't have that effect.

 

 

One other somewhat related comment....

 

The ring radiators suffer from a distortion where the radiating surface area changes with excursion. The BMS designs are excellent in this regard, but the resultant distortion is still there. It's a minor thing, but you can definitely hear it. Moving to a K69 on the tweeter will have removed that ring radiator distortion and replaced it with all the "beauties" of the K69 diaphragm.

 

 

I personally wouldn't replicate Mark's setup for my own listening environments, but I don't think it's entirely without merit.

Edited February 2, 2016 by DrWho

[DrWho]

But the harmonic distortion levels didn't reduce by the same amount...relative to the fundamental frequency boost reduction. 

 

 

A typical THD versus Level plot will look something like this:

 

The negative slope on the left is dominated by the noise floor of the measurement. As we increase the signal level, the signal to noise ratio improves, and the THD+N level gets better. At some point, the non-linearities in the system kick in and the distortion has more energy than the noise. Eventually we get near the clip point where the non-linearities become much higher order and we see a steep increase in distortion with a small change in input signal level.

 

I've seen systems where the minimum looks more like a plateau, or maybe there are multiple plateaus in a design - the shape of the curve is usually dominated by the architecture. Plateaus are extremely interesting when it comes to distortion because it means the non-linearity is non-linear....

 

I'm usually dealing with circuitry at work, but speakers have their own complex distortion mechanisms. Usually there will be multiple slopes of distortion in a design (this example has 3). The relative magnitude of distortion is level dependent, and at a system level we're seeing the culmination of all those mechanisms summing together. Everything else in your signal chain is part of that equation too.

 

 

 

Anyways, all that to say a distortion plot is really hard to read to understand what's happening in the system. I've been spending a ton of time trying to understand how two similar distortions can sound different - a large part of that is definitely the harmonic structure, but I don't think that's the entire story either. There are time domain effects that aren't readily apparent in steady state measurements, and then there is the complexity of asymmetric source material. A very low level signal can easily be riding into a heavily non-linear region. I'm starting to think them DC coupled fanatics are onto something....

[ClaudeJ1]

I personally wouldn't replicate Mark's setup for my own listening environments, but I don't think it's entirely without merit.

 

If you've got the space, I'm sure it's impressive more than visually. It's like being in a suite in any modern sports stadium. First class with a different size and angle. 

[ClaudeJ1]

 

 

 

 

 

 

 

I'm usually dealing with circuitry at work, but speakers have their own complex distortion mechanisms. 

 

 

Me too, and I'm dealing with CPLD's with 6 power supplies on the same PCB while slamming out 500 amps to a GTO SCR which controls 2000 amps in a train. Speakers are much more pleasant to deal with, at least they make nicer noise than a subway train.

 

It's a non linear world we live in. If you think anything in this world is linear, you are looking at a very small portion of the curve.

[Chris A]

...+ K69 tweeter.

 

The K-69 is a poor tweeter, IMO, as is the K-69-A.  If the diaphragm areas that you're talking about in the coax version are insufficient to make FMD inaudible, I'd say those listening will soon need hearing aids--if in a home hi-fi environment.

 

he coaxial diaphragms also need to deal with the air pressure generated by the MF driver showing up right in front of the HF driver. This causes modulation of the HF unit's coupling, which is a form of modulation distortion. When the two drivers are separated into separate horns, then you don't have that effect.

 

Then I think that you need to reconsider the thread that you're in if you believe that is actually a factor in this design.  I'd recommend you forking a new thread on that rather than now arguing that point at this juncture here.  This thread actually assumes that you've gotten past those kind of arguments and buy into point source performance.  Otherwise I believe that the outcome of trying to make those arguments now will diverge the focus on the current design to the point that this thread will no longer reflect its title.

 

Chris

[DrWho]

I would think the adiabatic properties of air would be the most significant whenever bandwidth is increased within the same size throat. Berenek gives us an easy calculation to approximate the distortion:

% 2HD = 1.76 x sqrt(It) x (f/fc) / 100

(https://books.google.com/books?id=6UzaBwAAQBAJ&pg=PA90&lpg=PA90&dq=beranek+throat+distortion&source=bl&ots=s27fE9RBLZ&sig=1d-YD2o4-KrCOhEthnoosrJQnuU&hl=en&sa=X&ved=0ahUKEwjH-fznl9jKAhVL7iYKHdy6AvgQ6AEIHDAA#v=onepage&q=beranek%20throat%20distortion&f=false)

The equation is for an infinite exponential horn, but Fc can be moved around to approximate the flare rate at the throat of a different expansion.

 

What happens when multiple tones go through that system? Dividing the bandwidth into two different 2" throats cuts "It" in half because each throat is seeing half the total acoustic power. And is not Danley's argument that entering further down the horn path approximates a lower flare rate? A lower Fc increases distortion in this equation - and I think you gotta use the port throat opening for this equation, not the horn cross-sectional area at the port entry point.

 

I'm calculating 1W creating 0.4% 2HD at 1kHz in this design.

A 2" throat is 0.002 sq meters. Let's use 1W and assume an efficiency of 25% -> which gives It = 125

Fc = 500Hz'ish? (conical is faster at throat than exponential)

F -> let's use 1kHz

 

At 4kHz that's 1.6%....already above the level we call clipping in electronics world. Reduce the input level to 100mW and you're still looking at 0.5%.

 

Normally I wouldn't care about 0.4% THD, but here it introduces modulation distortion since other frequencies are sharing the "same pressure". Also, cramming more frequencies at the same time through the same throat increases the total sound power. You can't just look at the level of a single sine wave. That "It" component in the equation becomes a factor in multiple entry designs whereas it's not a factor when using separate throats (total sound power is increased at the throat).

 

 

I only bring it up because you're dismissing Mark for throwing away point source behavior in favor of nothing. Mark is a sane guy (for the most part, hah) and there are some equations to show that he is gaining something.

 

 

Some more interesting articles on the subject:

http://doc.utwente.nl/58981/1/Schurer94modeling.pdf

-Compensating for the nonliearity with feedforward DSP - who's feeling like cranking some crazy math?

 

http://heim.ifi.uio.no/~sverre/papers/2009_AES-126-horn.pdf

-Effects of the rear chamber (at 99dB SPL) - maybe a larger rear chamber or a lossy rear chamber is something to consider?

[ClaudeJ1]

 

...+ K69 tweeter.

 

The K-69 is a poor tweeter, IMO, as is the K-69-A.  If the diaphragm areas that you're talking about in the coax version are insufficient to make FMD inaudible, I'd say those listening will soon need hearing aids--if in a home hi-fi environment.

 

he coaxial diaphragms also need to deal with the air pressure generated by the MF driver showing up right in front of the HF driver. This causes modulation of the HF unit's coupling, which is a form of modulation distortion. When the two drivers are separated into separate horns, then you don't have that effect.

 

Then I think that you need to reconsider the thread that you're in if you believe that is actually a factor in this design.  I'd recommend you forking a new thread on that rather than now arguing that point at this juncture here.  This thread actually assumes that you've gotten past those kind of arguments and buy into point source performance.  Otherwise I believe that the outcome of trying to make those arguments now will diverge the focus on the current design to the point that this thread will no longer reflect its title.

 

Chris

 

In looking at Roy's raw curves of the K-69, your comment has merit, but you are right about derailing this thread. I had single Sound Physics "Runt" (which was the name of PWK's toy poodle, BTW) and it's now known as the SH-95 from Danley's new company. It has the same great tweeter as the SH-50 (BMS4550), along with twin 8" drivers. It's only about 95 db/watt, but boy it's very smooth in it's response. I owned only one of those for a while as a center channel between TD-1 mains. I was amazed that it went down to 40 hz., even lower than the TD-1's but about 9 db less efficient. If you look at William Cowan's web site, you will see that he has copied this Runt design and lives with it every day, last time I checked.

Edited February 2, 2016 by ClaudeJ1

[ClaudeJ1]

A  rear view of the horn itself with drivers attached:

 

 

How about some more details, like dimensions and such. I'm sure you at least have a sketch of this on paper somewhere, eh?

[Chris A]

Nope.  No sketches, just the dimensions that I've already discussed.  The port sizes were adjusted to be 1/10 the area of the woofers, but it turns out that isn't really that hard-and-fast.  Perhaps more on that subject later.

 

The distance of the beginning of the off-axis ports from the throat was also discussed above, and is something that you can simulate within Hornresp, albeit grossly as it turns out.

 

I sized the drivers in relation to the horn and its real estate in my head and by using measured dimensions of the KPT-305 box for fit purposes.  Once the drivers showed up from Crites, I had an easy time determining the details on how they were going to fit (but I knew that they would fit, however). I just put them face down on a piece of MDF (actually melamine board), and drew an outline around them, then cut the pads out from the first pattern. 

 

Even the back closed box volume calculations were approximations, once I found how insensitive the performance of the woofers/off-axis ports were to that volume--it just doesn't matter much.

 

It was amazing to me how easy that it was to do--once I had an idea of how to cut the ports through. But I did a great deal of thinking about it before cutting anything.  Perhaps that's the real difference in why it was so easy once the fabrication started. 

 

The next phase isn't so easy as it turns out (I already knew that, too, before cutting anything).  I'll let you know when there's something to talk about in that area.  I think everyone now knows what the real issue is, and it isn't about cutting MDF and making boxes. It's that part that I'm working on.  There is at least 10x more to consider, perhaps 100x more to think about.  Nothing about it is easy.

 

But fortunately, all of it is easily held in one's head at one time.  It's really nice to work on a design problem that's that easy to completely conceptualize--for once. 

 

[i spent >30 years working on large and exceedingly complex design problems that were too big to do that, and had to use much more powerful analysis tools to not forget anything and to do the trades more intelligently and optimally--and simultaneously.  That was much more complicated and challenging mentally..  The current project is quite simple in comparison.]

 

Chris

[Don Richard]

How is the project coming along, not much news lately. Or have you been listening and enjoying?

[Chris A]

I've been listening to it as I design, trying to identify and avoid issues before making those errors, and trying out some controlled experiments to save on material costs--some of which can be fairly expensive if one is not careful. 

 

There actually is quite a bit of visualizing that is required in addition to selecting the process to control heat and making sure that dimensional tolerances in the finished product are still being held.  The tooling process seems to be a series of one-shot affairs to get each part right...and if not right, expensive materials turn into refuse.  Being able to reuse the tooling many times is also part of the design problem. I'm hoping for some initial results within a couple of months. 

 

There is at least one other from this forum that I believe has decided to build a New Center from a K-402 assembly as a starting point.  That process is of course very straightforward and low risk, albeit at a higher overall cost because of adding a second set of overhead costs and profit margins that's tacked on--as well as apparently introducing a minimum acceptable price and profit margin, or some closely related issue that's not being voiced.  Those are issues that don't apply in my case.

 

I've also been carefully looking at the REW data again.  I find that I have to periodically remind myself that most of my listening impressions and data are really not apples-apples with the Jubs, since the Jubs are in the corners and the new horn isn't - instead being located along a wall and elevated off the floor, and using a K-69-A instead of a TAD 4002  This is actually extremely good news in that the results are very close to each other, nevertheless.  That's something that wouldn't be true if their loudspeaker room positions were reversed.

 

I'm finding that not designing for certain conditions and customer needs that I don't need makes all the difference.  Many are apparently not asking themselves the questions: "Why do I need that?  Can I design without that and get better listening results?"  Problems originate in imitating other designs without first understanding what function those design details were made to address.

 

Chris

Edited February 18, 2016 by Chris A

[Chris A]

I had my daughter and son-in-law over recently.  My daughter isn't an audio system enthusiast, but she brought up the subject of what I've been working on, so I played the first cut of this DVD-A for them.

 

She mentioned that the clarinet sounded like she could reach out and touch it (you should know that her music scholarship instrument was clarinet).  While I've read that same sort comment in this forum before, this was a first for her and one that came unsolicited. Remember that I've owned Jubilees for about 8 years now, 7 with TAD 4002s.  She never mentioned anything about the sound with them before that, to my knowledge.  My son-in-law (himself a pro musician with his own band) just listened intently and nodded when she said that.

 

That's an indication that this new center is a pretty big step up, not only for the center speaker, but in terms of its effect on the overall sound experience. Remembering Beranek's Law, I try not to overstate this sort of thing, but in the case I thought that I could share that experience.

 

 

Chris

[Don Richard]

Thanks for the news - I figured you were ironing out details.

[Khornukopia]

I've been listening to it as I design, trying to identify and avoid issues before making those errors, and trying out some controlled experiments to save on material costs--some of which can be fairly expensive if one is not careful. 

 

There actually is quite a bit of visualizing that is required in addition to selecting the process to control heat and making sure that dimensional tolerances in the finished product are still being held.  The tooling process seems to be a series of one-shot affairs to get each part right...and if not right, expensive materials turn into refuse.  Being able to reuse the tooling many times is also part of the design problem. I'm hoping for some initial results within a couple of months. 

 

Are you explaining a modification to the original design of your New Center speaker, or are you describing the development of a production line for a thoroughly tested, state of the art, Multiple-Entry speaker?

[Chris A]

I need about four more for my main 5.2 setup (plus maybe one), and perhaps a couple or maybe even three more for a setup in another room which is available for hauling around without having to borrow from the main system.  That's a few.  Since they are so much more compact than Khorns and Jubs, I'll be getting some space back, and their listening axis can be much lower which is also quite handy for setting up a surround sound array to be in-line with a screen/TV that's more at sitting height.

 

But it's best not to start counting chickens yet, before I can prove to myself that I can fabricate the horns reliably, and that they perform as I expect them to sound in the main listening room.

 

Chris

[ClaudeJ1]

But it's best not to start counting chickens yet, before I can prove to myself that I can fabricate the horns reliably, and that they perform as I expect them to sound in the main listening room.  

 

A worthwhile endeavor that merits some care. Meahwhile, I just sold my center SH-50 (ported), finding that an old TD-1 is good enough for voice channel. Instead of 50x50 coverage it's 60x40. and 3 more for surrounds. So all Danley stuff for the moment while I work on my K-402 setup. I'm far behind you on this but will catch up soon.

[Chris A]

Instead of 50x50 coverage it's 60x40. and 3 more for surrounds...while I work on my K-402 setup.

 

One of the things that I really can't overstate is the coverage (and smooth directivity throughout) of the K-402s...which is spectacular.  I mentioned this point to my son-in-law, at which point he walked from side-wall to side-wall and also found that the integrated soundstage image was stable and solid anywhere in the room once you get about 2-3 feet from the center loudspeaker.  I believe that he was very impressed.  That's the benefit of having 100 x 60 degree coverage with the front three loudspeakers. 

 

I really can't wait to hear the difference with surround channels once I can replace the present bi-amped Cornwalls with this new multiple entry horn design. Cornwalls don't have anywhere near that kind of even (i.e., no directivity hiccups with frequency) in horizontal coverage. 

 

Chris

Edited February 20, 2016 by Chris A