"Nephila" F6/AMT design...

[Wolf]

Just showing the preliminary data for the upcoming pictures at Meniscus' shindig next weekend.





Note- the effective port length is actually about 3.5" in the tower, 2.5" diameter.







These really sound good! The rest of the pictures will wait until next weekend, following crowd perspectives I hope.
Thanks for looking,
Wolf

[jrhollander]

Looks like you modeled with and without the notch. Did you build both, could you hear the difference?

[mattsk8]

Can't wait to hear these Ben, interested in both that woofer and that tweeter!!

[Wolf]

Looks like you modeled with and without the notch. Did you build both, could you hear the difference?

Actually no- the 2 tweeter plots are only to see the difference between the electrical transfer function and the acoustic rolloff. The notch is present in both tweeter snapshots, just not shown in one of then

I have not listened without it, but- I would not have as flat a response without it as it peaks and rolls off quick. The variable is the cap, and anything from 0.75uF to 1.5uF in the notch changes the character in sim very dramatically.

Wolf

[Wolf]

Okay- now starting with the grit...

When I got started with this project's process, I had a set of door-prize Airborne RT5002 AMT's from InDIYana this past Spring. I narrowed down the woofer specimens to a couple, and chose the F6 due to its wonderful midrange, higher sensitivity, and extended upper bandwidth response. I found 3rd party FR plots for the RT5002 in an issue of Voice Coil that I was given at an MWAF event. It also turns out that the issue was available via PDF online, so I was able to glean quite a lot of info about the units before diving in. The 5002 has a sensitivity of around 100dB on its own, which left plenty of headroom for tailoring the response to my requirements since it did not have the flattest of FR plots. It also rolled off hard on its own, and had a peak of around 15kHz and then rolloff in both directions.


Vance also had an HD plot of the units, which helped in where to make the transition.


It took 5 iterations of types of xover experiments to get what I wanted, and the 3rd adaptation of the 5th sim was the winner. Since the AMT rolls off so hard on its own, it was difficult to get the response to favor much more than a 1st order xover on it to get good phase alignment. I finally achieved a second order topology with the last one, and was able to mold it as needed. The electrical transfer function yields a BS2 at 2.2kHz, which is about right according to the HD profile above.


Adding in the inherent acoustic rolloff yields a BW3 @ 3.8kHz as far as overlays go.


The F6 is a gem, so I didn't want to kill all it had to offer. The 900-1.1k region is a strong resonance that needs dealt with so it does not get honky. What you see below is measured parts values of what I had, but a 47uF/0.47mH/8 ohm notch is what the parts spec values were before tolerance. The coil DCR was kept to 0.6 ohms both in sim and in reality, as this can affect tonal balance when you stray too much from sim. Once the peak was deemed benign, the response was able to hit a pretty easy rolloff to meet the AMT at 3.4kHz/BW4. So the xover is basically a BW4 at 3.4kHz. (I misspoke Saturday when I said BW5...)


Final sim:


Box:

When I measured the F6, I got the following:


...and I used the Vas spec I already had in Unibox from a previous simulation awhile back. Turns out that was wrong at 25.4 ltrs. The initial box sim had a tuning of 50 and a box of approximately 1.0ft^3, walls lined; Nice and flat to 45Hz. When something didn't sound right at first listen, I dug in a bit....
HiVi spec lists Vas at 16.4 ltrs, and a Mms of about 18g. Since my Mms was calculated to be wrong in the simulation, I ratcheted the Vas spec (since it was the only unaccounted variable in the simulation) until I got closer to 18g, and found the Vas to be a lot closer to 13 ltrs- HALF of what I initially used. To top it off- the simulation now looked like what I was hearing as well. It had a rising peak to ~50Hz of about +4dB! I played with the box alignment to find that if I stuffed it mildly I now had a sim tuned to 45 and an F3 of 39Hz when using actual figured/measured port length, and the response was actually flatter in box-sim than it was before. I used 3oz of poly in the bottom cavity, 2oz in the port cavity, 3oz in the cavity below the woofer along with some Sonic Barrier, and Ultratouch lining only in the cavity behind the woofer. I pretty much always stuff the bottom sector of my towers to avoid the column resonance issues. It's likely just shy of a pound of fill per enclosure. So- I'm sure this is about what I got at 20W and 100dB:




Since I used a knockdown cab, I had to shift the baffle upward to get the AMT closer to on-axis listening. I used a piece of 1/2" MDF 6" in height to achieve this. I also added more bracing to the box than was standard, including backing the port, and tying in the top and bottom panels. The 'I-beam' behind the port is sculpted out to allow maximum airflow. The section behind the woofer uses the U shaped brace, and the longer height of the dual-end options. I added another +-brace to tie in the sides behind the woofer.





F6 rear of hole was chamfered for breathing room between the screw mounts. I used an edge-guide to get the rebate for the AMT precise, and surface mounted the F6 to get as close of a CTC as possible without overlapping the AMT's exit opening.

Relative baffle positions here:



I also chamfered the front edge of the cabinet from the same place just below the woofer all the way around to the other side just below the woofer. This left the upper baffle portion a little wider than the box and gave a neat illusion in terms of view, while the chamfer flush-trimmed the edge below. To get rid of the rabbeted edge at the rear of the knock-down cabinet, I used a table-edge bit (thumbnail) to end at the side panel and eliminate the MDF edge-grain to the sides and top of the cabs. This looks much better to me.
Full 1" Radius roundovers finish off the front to rear or Z-axis corners. Obviously some of these cuts have to be done before you glue on the baffles, so if you plan on copying the aesthetic, please think about the proper cutting/gluing order.

I have always isolated the tweeter from the woofer cabinet when it has been of the 'ribbon' variety, including AMT/Planar/etc.. As it turns out- I found out Saturday that there is indeed some sound emanation from the back of the AMT baffle. Partially because it resonates like a tuning fork, and also because the lead-in wires to the AMT are NOT SEALED. This means there is in fact a 'leaky-back' response profile of some sort. I plan on leaving this open when I affix the rear portion of the upper baffle and the right-angle supports to finish off the strength required up there. I didn't at first think there was rearward sound radiation, but there is, as also observed by Bryan Keane on Saturday at Meniscus.

The port was an interesting development. I recently became reacquainted with a picture of a Bryston and Axiom speaker where the port exit had a varying set of radii. Being it was a 2.5" Diameter port, 3.5" long, I had a wide enough chance to try something like it. I kissed the edge with a 3/8" roundover bit, and then marked my 'corners' with lines parallel to the cabinet edges using a square to be the same dimensions of the circle. I then used a 3/4" roundover by hand-routing inset 1/8" depth and just touching the corner lines with the outer edge of the bit. This yielded somewhat of an 8-lobed port exit, reminiscent of the namesake- Bonus!


The xover was made of fairly high-end parts, utilizing an OS flat-pack box from Meniscus. The PE Energy buyout terminal cups fit the B4N cutout, and the Neutrik NL4PMR jack fits the HiVi T20 recess. I did have to shave off a bit of the B4N through-hole so the funky shaped heart-cup fit into the hole, but otherwise a very good fit! I installed channels to either inner end of the box so the xover could slide in and out if necessary.



Easy-Peasy outboard xover!


I later used the terminal cup's auxiliary hole for a switch-mount, to facilitate the switch between 2 capacitors of my own choosing.
(Picture)

...as they currently stand:


Schematic to come later....
Thanks for looking!
Wolf

[Wolf]

Here's the schematic!
Wolf

[Wolf]

Of note:
The contour on the top-end of the AMT can be adjusted to taste if someone has the desire. Anything from 0.68uF to 1.5uF in place of the 1.0uF cap will change the response from more peaked to flat. I settled at the 1.0uF to be a decent compromise, and not be too low or too high in output and be about the same amplitude as most of the speaker system's produced bandwidth.

Try to follow DCR values, as shifting these will change the tonal balance.

The vertical off-axis response of the 5002 is not that great, being a ribbon variant, but as you step further away from the speaker it does improve. Obviously sitting is the best position for this shorter tower.

I optimized for on-axis response, or just slightly above the AMT axis for best results, so toe-in is a must for the best presentation.

Lastly- I don't plan on changing the xover at all. So this is the final circuit layout just above in the last post.

Later,
Wolf

[Wolf]

Our hosts now stock both drivers for this project.
http://meniscusaudio.com/hivi-f6-p-1493.html
http://meniscusaudio.com/airboorne-rt5002-p-1494.html

Later,
Wolf

[Wolf]

More grit, from when I was at Jeff's place....

FR(400+Hz):


Nearfield woofer, showing tuning is actually 40Hz as built (not complaining, I like it this way!):


Port response:


Speaker system HD (Pretty nice, IMO!):


What I really find interesting is that the bump in F2 is actually showing in the FR of the port response.

So there it be, all the neato squiggles....
Wolf

[jrhollander]

Looking forward to hearing them in April or July

[Wolf]

Likely both...
Wolf

[jrhollander]

OK thanks for the heads up I guess I know what category to avoid

[Wolf]

Barely into the 200+ these are...
Wolf

[Wolf]

Our hosts now stock the kit for this project!

http://meniscusaudio.com/nephila-bare-b ... -1519.html

Later,
Wolf

[Jeff B.]

That's going to be a nice kit, Ben. I found it to be a very neutral speaker with good resolution. Even the bass tuning was right on the mark. Excellent balance. I was very impressed. You did a great job on this design. Thumbs Up.
Jeff