Metlako: A Small, Affordable Two-Way Unity Waveguide

my brother in law has a colleague who owns a 3d printing company (he's a surgeon, they make bones)

accidental post there. What i was going to ask is, should i try and make a) the unitised waveguides b) metlako v2 c) metlako v1 d) all 3?

futher edit to say, should this go well, and cheaply (in the UK) i'm happy to facilitate futher copies. They should be extremely cheap (like much less than 100 GBP).

If you can get it to print, Metlako V2 is kinda amazing. Lots of output and smooth response, and a simple crossover.

For the past few months I've been trying to come up with something similar to Metlako V2, that's easy to print.

It's close, I think it should be ready for the printer today. It's basically a two-way Unity horn that's not much bigger than the UICW, using two 5" midbasses instead of four.
 
Last edited:
My first printed horn, Iwata 600 in white pearl.
That looks crazy nice. What's the story behind these (equipment used etc)?

Was the glossy finish achieved with a milllion hours of sanding + layers of paint?

If I was doing projects like Patrick's, aiming for cheap + not too hard to make, and getting the occasional crack or whatever, I would:

- use a bit of sanding + filler
- then hide all the remaining imperfections under a layer of paint

...either something a bit textural (e.g. hammerglaze) or something visually uneven (e.g. stippling, camo pattern).
 
Have a Creality CR-10 S4 purchased almost three years ago now. Haven't been feeling well for a couple of those years and it wasn't used much. Major PITA
But once feeling a bit better took on the task of sorting things out and replacing hardware that wasn't manufactured correctly or correcting in another fashion.


All metal hotend, mine uses a cheap e3d V6 clone, Titanium heat break, Z frame stiffeners. Added a Bondtech clone 4:1 dual drive extruder to the hotend. It's based on the CR-10 Heavy Duty
This Direct Drive HD varient, without filament sensor, drag chain, etc.


Solid bed mounts and self leveling system, unless included are my minimum recommendations.
I added X and Y belt tensioners, Y stepper stabilizer bracket, 1000w heatbed and a Wham Bam magnetic build plate w/PEX. I was #32 on their Kickstarter.

Currently pushing out PETG at 90mm/s and require another hotend upgrade to the e3d Volcano. Extrusion rate is my limiting factor atm. Want to use 0.8 - 1.2mm nozzles. Taulman recommends the cheap Chinese standard brass nozzles on their commercial printers for use with T-Glase PETT. Buy 0.4-0.6mm cheap brass nozzles and drill them out with a set of tungsten carbide drill bits from Harbor Freight. 10 pairs ranging from 0.2mm - 1.25mm for $8.99
20 Pc. Carbide Rotary Micro Bit Assorted Set

Use the 0.8 -1.25mm for angel hair to spaghetti like extrusions. They are laser etched, but measure first



Good solid CoreXY printer with a 10"x 10" x 10" build area for $500 SX2 SX2 3D Printer – Tiny Machines 3D LLC
or the beastly (and I'm ready to pull the trigger) SX-4 with a 400mmx x 400mmy x 500mmz
SX4 3D Printer – Tiny Machines 3D LLC


CoreXY is faster and more accurate. What commercial machines are based on... and get all the upgrades you can installed up front. Saves money lol... have enough spare parts/ extra things to build another machine minus the frame :)


Manage it with an Octopi server (raspB3+ and cam). On my android devices manage it with Printoid or directly through the web interface. Using Cura, but am wearing a little thin at how it slices sometimes. Think Simplify3D is on my list... like Bjorn's book that AMAZON still has yet to release! Errr
 
Last edited:
Post processing is very little on these transparent rose PETG (Prima EasyPrint) candy bowls. Sitting on top of my Galaxy Tab S3 for scale. The small one is 75% scale.



Made the Ctron vase on Thingiverse using 200g YOYI Translucence Green #5 petg.
Call it fluorescent Green with unique optical properties that tends to refract gold colored light. Looks amazing in a window, built tool tuff. Told neighbor to drop it, so he did. It just bounced, not a scratch :)


get the retractions set right and temps set correct for petg. Going from 2.95mm retract to 3.05mm retraction made a big difference in reducing that 'fine as frog hair' fuzz on my gimmick. Set temp to the lower side of the range scale... but as Patrick has mentioned even color change from the same manufacture has an effect on the glass transition point. I have some Tttyt3D pla that is recommended to work in the 205-230c range. It won't extrude, like not melt even at 210, must be pushed to 225-230. The highest temp material I've run yet is eSUN translucent purple petg which extrudes wonderfully at 242c

Purple Turtles. The second one is ironed on the top, so smooth as a baby's bottom. Interesting what the camera reveals



Not even going to try ABS until enclosure is built. Patrick what air temp are you shooting for when using this stuff with a heat lamp?
 

Attachments

  • 20191023_131959[1].jpg
    20191023_131959[1].jpg
    671.7 KB · Views: 626
  • 20190930_001335[1]ac[1].jpg
    20190930_001335[1]ac[1].jpg
    276.2 KB · Views: 627
  • 20191028_161646[1]ac.jpg
    20191028_161646[1]ac.jpg
    797.4 KB · Views: 619
Last edited:
That looks crazy nice. What's the story behind these (equipment used etc)?

Was the glossy finish achieved with a milllion hours of sanding + layers of paint?

If I was doing projects like Patrick's, aiming for cheap + not too hard to make, and getting the occasional crack or whatever, I would:

- use a bit of sanding + filler
- then hide all the remaining imperfections under a layer of paint

...either something a bit textural (e.g. hammerglaze) or something visually uneven (e.g. stippling, camo pattern).

Simply printed in 5 pieces on Zortrax M200 using quality PLA, then glued using CA glue and metal pins, then filled with finnish putty then outsourced to experienced sprayer.

Now I an print in one piece on large printer in just one night. My next attempt will be printed mold and expanding polyurethane foam.
 
7Lk5y47.jpg


ABrblAq.jpg


F0ZHhAM.jpg


I've honestly lost track of how many Metlako versions I've made now. I think this is #4, but it may be #3.

The idea with this Metlako, was to have a waveguide which could go into a sonotube.

The use of a sonotube should extend the directivity control a little bit lower than it would normally go. From my experience, a big ol' roundover on the enclosure will extend the directivity control about half an octave lower.

EdXHRJu.png


eXGpdhy.png


Polar response of the tweeter, with and without smoothing, with EQ and xover

uapHwMF.png


Here's the response with the midbasses.

I wound up trashing this Metlako for a few reasons:

1) This Metlako uses a different set of midbasses, the Dayton Reference 125 midbass. Due to the higher excursion, the Dayton was smacking into the mounting plate.

2) The Dayton midbasses cost more than twice as much as the MCMs and performed worse, though part of that was my (bad) mounting plate design.

3) Biggest issue by far, is that the diffraction slot isn't working well with the dome tweeter. The response above 10k is not great.

You win some, you lose some. Metlako V3 isn't terrible, but it's inferior to Metlako V1 and V2, so it's going into the dumpster.
 
passive x-over

Patrick , if one doing passive x-over what filter do you suggest? me thinking 2nd order on the tweeter and 1st order for the low @1.5k.Is it sound about right?
The tweeter look tiny and low sensitivity, would it be the limiting factor since there are 4 woofers? The PE said the tweeter handle 80Wrms.
 
For the most part, I nearly always use this topology, or close to it:

http://www.melaudia.net/zdoc/jml_crossovers_etf04.pdf

Here's the math, from the presentation:

settings of a 2 ways system according the proposed method
1..........Only Butterworth 3rd order filters are used.
2..........The relay frequency Fr between the low-pass and the high-pass
is defined at -5dB.
3..........The cut-off frequency Fl (at -3dB) of the low-pass filter is
calculated by: Fl = 0.87 x Fr
4..........The cut-off frequency F
h (at -3dB) of the high-pass filter is
calculated by: F
h = 1.14 x Fr
5..........If the 2 drivers are aligned at the same distance of the listener,
then the low-freqeuncy loudspeaker has to be moved toward the listener
of a distance equal to 0.22 x wavelength at Fr
.
6..........The polarity of the high frequency loudspeaker has to be
reversed.


It's a great presentation, and I recommend reading the whole thing.

But if you want the short version, we're using 3rd order filters on the high pass *and* the low pass. A 3rd order filter introduces a delay. To compensate for that delay, we're moving the woofer closer to the listener. (This is done automatically, because the woofers on the waveguide are closer than the tweeter.)

Where things get tricky, and the reason I'd really recommend using active crossovers, is the fact that the response of the midbass and the tweeter ISN'T flat.

Because the response isn't flat, you have to use some strange slopes to achieve an acoustic filter that's 3rd order.

In Metlako V5, I used a 2nd order Linkwitz Reiley filter on the midbass, and a Butterworth 3rd order on the tweeter, along with some EQ.

In the UICW, I was able to get close to a 3rd order low pass with no crossover whatsoever! Basically the inherent filter that's caused by the midrange taps, was very close to the 3rd order filter needed by the overall design.

To make a long story short: I'd start with somewhere around a 1st or 2nd order low pass on the midbasses, along with a 3rd order highpass with the tweeter. This design requires EQ, and going active is highly recommended. The $80 cost of a MiniDSP will be cheaper than a passive xover.

Then again, if you're into tube amps, it will need a passive xover. (Unless you want to buy two tube amps!) In that scenario, a passive xover will be the way to go. But I haven't made a passive xover for any of the Metlako speakers, they're all active.
 
If you have a microphone, you can make a crossover for a Unity horn in about one hour. Here is how I do it:

k8pLxZn.png


yqD3C1X.png


I use a combination of EQ and crossovers to achieve an acoustic response that's 3rd order on the midrange and the tweeter. The reason that I can get away with these steep slopes, while keeping everything in phase, is because the midbasses are closer to the listener than the tweeter is.

In this design, Metlako V5, I *am* using a little bit of delay to make the crossover better. But the difference is subtle and probably inaudible, the delay makes things a *little* better. I'm too lazy to look up the settings, but IIRC, the delay on Metlako V5 is something like one tenth of a millisecond, which is equivalent to moving the midbass closer by about 3.4 centimeters.

D19Fjin.png


Here's the filters and the response of Metlako V1. This crossover is quite a bit different than Metlako V5. You'll notice that Metlako V1 is using a first order filter on the tweeter, and the xover point is much much higher than Metlako V5.

The reason that this works, is because the tweeter response isn't flat. It rises at low frequency. Because the tweeter response is not inherently flat, a first order filter way up at 3500Hz will yield something close to a 3rd order rolloff at 1350Hz. It's basically two ways to achieve the same acoustic filter.
 
Simply printed in 5 pieces on Zortrax M200 using quality PLA, then glued using CA glue and metal pins, then filled with finnish putty then outsourced to experienced sprayer.

Now I an print in one piece on large printer in just one night. My next attempt will be printed mold and expanding polyurethane foam.

Very nice.

When I tried to fill gaps in a (much larger backloaded type for a 12") horn with PU foam I sprayed/misted a bit of water on the surfaces first, apparently helps the hardening process.

PB, What's the difference between v2 and v6?
 
Very nice.

When I tried to fill gaps in a (much larger backloaded type for a 12") horn with PU foam I sprayed/misted a bit of water on the surfaces first, apparently helps the hardening process.

Yep, I should have mentioned that, I do it with all the multi-part waveguides. I mist it with a spray bottle.
PB, What's the difference between v2 and v6?

7Lk5y47.jpg


I kinda lost track of all the variations. Off the top of my head, this was somewhere between V3 and V5. This was one was one of the poorest of the bunch. (Of course, this is relative, it outperformed a lot of commercial waveguides.) Basically the diffraction slot didn't work really well with that tweeter, and the response above 10khz was beamy and rough.

KikkdoE.jpg


Metlako V6 (or was it V5?) is probably one of the best "bang for the buck" waveguides I've ever made. Really nice performance, but also fairly easy to print. I was honestly surprised that I was able to fit two 5" woofers onto a waveguide that fits onto a print bed that's 8" x 8". I'm not entirely sure how I pulled that off. This waveguide is a straight up clone of the 18Sound XT1086, but it's about 80% as big. I really need to post some pics and upload the STL files to github. It's a nice one.

Probably the only reason I didn't wind up using it, is that I've been kinda obsessed with using curved baffles to extend the F3. I made another waveguide that fits into a cylinder, similar to Metlako V3 (or is it V4?), and that waveguide simply knocked it right out of the park. Really REALLY great performance, and though I shy away from subjective analysis, it sounds FANTASTIC. Of course, I've been too busy building waveguides to document anything... I put the kibosh on that one because it's incredibly difficult to print, and mounting it in a cylinder is a complete p.i.t.a. I'm working to come up with an enclosure shape that will mimic what a cylinder does, but is easier to build along with a waveguide that will take advantage of the baffle shape.

Basically the idea is to have a waveguide that's small, like 8" x 12", but that behaves as if it's twice the size. This is doable when you put the waveguide in a cylinder, or you use a big roundover.

yZAGjjA.jpg


To give you an idea, here's a polar response of the waveguide in question. This waveguide is 8.5" wide. But note how the directivity control goes about as low as a 17" waveguide, and it also collapses in a really benign way. IE, some of the big waveguides have a habit of collapsing poorly, basically they have a narrowing in the midrange and then they widen abruptly, and suffer from pattern flip. (This one DOES suffer from pattern flip on the other axis, which is part of the reason I'm redesigning it...)
 
Member
Joined 2007
Paid Member
That last set of polars looks fantastic Patrick. I think if you can get the performance of a 17" wide wave guide in 8" wide you are onto a very commercially viable product as there is a strong preference among more normal people for narrow baffled speakers.
 
Sound really seems to HATE symmetry. Here's an example of what I mean:

tDnJzgh.png

HCA5rsF.jpg

Unsmoothed polars
ym97YoR.jpg

Smoothed polars


Here's a waveguide that I kind of made on a goof, 2.5 years ago (Synergy Beolab)

71rL2tX.jpg

yZAGjjA.png


Here's the new 'Bend it Like Bateman' waveguide, that I made a week ago.

The performance isn't quite as ridiculous as the one from 2.5 years ago, but it's interesting that the asymmetry of the waveguide seems to be having a beneficial impact on the frequency response and the polars.
 
Last edited:
rsEZmUs.png


If I was going to put a coax on a Unity horn, it might look like this.

LDx2oJx.png


The depth is 3.5". The depth is largely dictated by where we want the midrange taps to be. (See the beginning of this post.)

JYiavT3.png


Here's where your coax would go. The tweeter fires through the throat of the horn, and the output of the midrange fires through that ring shape.

1aQf7PZ.png


Okay, this diagram is super important.

It is *critical* that the area of the midrange tap is expanding.

Here's a few reasons why we want that midrange tap to expand:

1) this is still a horn. If the area of the midrange tap is constant, we're going to get a resonance. In the original Unity horns, from 20 years ago, the area of the tap was constant. The reason that it worked was because the taps were short. For instance, if your midrange taps are only one inch deep, it's ok if they're straight. But if your midrange taps are three inches long, or six inches long, or TWELVE inches long, you want the area to expand. If not, you're going to get a big ol' resonant peak and that's going to ruin your speaker. Long story short: you want those midrange taps to be expanding. If you're a complete madman, you can mass load the ends of those taps. That's a story for another day.

2) If the area of the midrange taps are expanding, the horn behaves a lot more like a horn. If the area of the midrange taps are NOT expanding, you're going to maximize diffraction off of the slots, and you'll probably get some reflections off of the transition from the midrange taps to the rest of the horn. Also known as "higher order modes."

If it's not clear from this post, the reason that the AREA of the midrange taps is growing, is because the diameter of the midrange taps at the throat and at the exit of the taps is higher.

For instance, at the entrance of the midrange taps, the diameter of the midrange taps is 2". At the exit, the diameter is something like 4". This means that the *area* at the exit of the midrange taps is 4X higher.

18s6Sxv.png


Here's a cutaway giving a better idea of how the area of the midrange taps is constantly expanding.

2jj9eK1.png
Any development on this? I am eager to find out. Thanks,