Walnut H9800

[John W]

As always... just beautiful John!

From the pics it looks as though you decided to omit the fillet that starts out really tight in the bell and eventually grows to a full 0.75" radius at the mouth?

I pondered a couple methods to include this, eventually ending up with the modified approach shown in this picture. The radius is applied to the side pieces tapering about 4 inches back to the mouth. This provided a bit of the curve shown in the plan and smooth taper for the mouth. I was considering adding a little putty initially but have shelved that for now.

[Earl K]

From the pics it looks as though you decided to omit the fillet that starts out really tight in the bell and eventually grows to a full 0.75" radius at the mouth?

For the sake of clarification ( for the rest of us /OK, maybe mostly me ) ;
(i) What do you mean in your above statement ?

(ii) For Instance ;
- Isn't your referenced .75" measurement, the internal height ( that's somewhat parallel ), taken between the bottom & top portions of the internal bell ? ( ie ; the height of the internal diffraction slot that exists before the bell opens up )
- & "Really Tight" equates to what internal height, if it's not .75" ?

With my CNC, I couldn't keep this fillet in wood and had to resort to adding bondo as the fillet edge became knife edge thin and blew out...

- I can see that if this thin flat ribbon ( fillet ) was very narrow / then it would want to peel off ( or "blow off" as you say ) / especially with grained woods.
- Is this fillet area , the only area where the internal diffraction slot is truly parallel ( measured top to bottom ) ?

I wonder how leaving it out affects the sound?

- As I understand diffraction ; one does need a well defined knife-edge to create the necessary low ( impedance ) pressure zone to bend those small (HF) wavelengths.

(i) Rounding down that internal knife-edge to a more gradual transition ( curve ) should reduce HF diffraction by some measurable amount. This will move the characteristic of the horns" HF polar pattern, a bit more towards that of a true exponential ( so, a rounding creates a bit beamier horn ) / which may in fact improve the H9800(s)' imaging capabilities ( to some extent ) .

(ii) Rounding does add path length, which will usually aid in LF loading .

(iii) Here's a comparison of measured performance for Johns' horn versus an old response study for the H9800/435Be from Project May .
- The vertical scales are different ( Johns' is a little more "dramatic" because his scale is set to 8 db ).
- The horizontal scales are now about the same, after I resized the pics.
- There are some notable FR differences . ( I do like Johns' version, a bit more .)

- Just some ramblings & observations . ( John, I hope you don't mind my 2 cent analysis )

[John W]

Looks like we were both writing at the same time.
I think Widget was writing about the radius where the vertical and horizontal curves come together.

Thanks for the side-by-side plots. I was just eyeing them before, but this is much clearer.

[scott fitlin]

Gorgeous woodwork.

Let us know how they sound!

[Earl K]

I think Widget was writing about the radius where the vertical and horizontal curves come together.

- Okay, that makes sense .
- I can see how making those rounded intersections between the horizontal & vertical parts ( immediately after entry to the horn but before the flat area of the diffraction slot ) / would be quite difficult . I like your approach. Smoothing with Bondo also seems an appropriate fix .

- I'll admit that I had to look up the definition of "fillet" , to even attempt to understand Widgets' question to you ( it appears, the dictionaries definition of "thin, flat, & ribbon-like", had me focusing on the wrong part of the horn ).

[Mr. Widget]

I think Widget was writing about the radius where the vertical and horizontal curves come together.

Yep... your solution to approximate it is quite clever... in my pedantic attempt to replicate the H9800 exactly, I found that the complex set of intersecting curves were just too difficult to pull off in wood... it is quite possible that your solution will be more than satisfactory. It certainly is an attractive looking final piece.


Widget

[Mr. Widget]

- I'll admit that I had to look up the definition of "fillet"...

Oops... I try to speak plainly so that our foreign friends can easily follow along with the help of Webster... I guess I didn't realize fillet was a technical term, but in retrospect I guess it isn't used around most people's dinner tables. In my circle of friends, designers and engineers, it is a rather common term... but then we do tend to bore outsiders with our dinner conversations.


Widget

[John W]

Actually “fillet” would be used around most peoples dinner table, just in a different context.

[Mr. Widget]

Actually “fillet” would be used around most peoples dinner table, just in a different context.

I imagine that there are those who'd call a Filet Mignon, a Fillet.


Widget

[Maron Horonzakz]

At my dinner table, fillet would describe the blending of a fuselage and wing of a airplane.

[jtgyn]

G'Day John,
Great work!!
Regards Scott

[matsj]

Have you had some time to test the new horns yet ?

mats

[John W]

Not much new to report.
My current plan is to construct a couple cabinets with LE14H-3 drivers and configure them as 2-ways. I was going to wait for final testing and setup till this step was done. Starting with the compensation network I'll play around with a few different crossover options before deciding whether to keep the crossover active 2-way or try make a full passive crossover version.

I was able to order a pair of LE14H-3 directly from Harman. One is in the mail, but the other is on back order till the end of April or so.

I'm considering a couple cabinet designs for the woofer, mostly along the lines of the 1400 array. I want to mount the horn vertically, but need to work out a good way of transitioning the vertical horn to the rest of the cabinet.

[John W]

I’m just finishing up a new 2-way project to compliment my H9800 walnut horns. The two drivers are 2435HPL and LE14H-3. The walnut veneered boxes have curved sides and are roughly 3.25 ft3 vented out the back. I plan on tuning them to 28hz, and still need to adjust the 4in x 8.25in ports, they are little long. The design is modeled after the 1400 array, but I decided to extend the top up a little for a couple reasons. The first was a tad more internal volume and the second is to allow the horns to be securely positioned either vertically or horizontally.

Of course there is one problem as you probably noticed, UPS lost my other woofer. I don’t know how they stay in business. JBL is handling it nicely and sending a replacement, but unfortunately they are on backorder for 3 or 4 more weeks.

[John W]

My initial test was running bi-amp with an active crossover at 800hz, and the top end using the Timbers compensation filter. Listening in mono-mode the one box is sweet. The compensation filter really smoothed out the timbre on the 2435. I can’t wait for the whole package.

Here is plot in the listening room, about 3 feet back. And a couple construction