Small 2-way speakers with linear on-axis and power response characteristics (Scan Speak and SB Acoustics drivers). H&V off-axis measurements included

[a4eaudio]

Distortion: 25% infill cabinet, ...

I would change the terminology to "25% 3D printed wall infill" to be clear. "Cabinet infill" implies you are stuffing the cabinet.

...Edit:
Amir's port measurements also show two peaks, though at a bit lower frequencies and the first is not nearly as pronounced as I'm seeing.

Is this still without lining? The port resonance can only resonate if given energy at that frequency. Once you line the cabinet (assuming you haven't in the images above), the output inside the cabinet at 1.2kHz will be damped and your resonance will be less.

 

[technodanvan]

I would change the terminology to "25% 3D printed wall infill" to be clear. "Cabinet infill" implies you are stuffing the cabinet.


Is this still without lining? The port resonance can only resonate if given energy at that frequency. Once you line the cabinet (assuming you haven't in the images above), the output inside the cabinet at 1.2kHz will be damped and your resonance will be less.

Fair points! I'll try to do a better job about the describing what I mean by infill percentage.

Correct, the interior of the cabinet is completely bare as, at least up til now, I wanted maximum energy going into the plastic so we can see the impact. Goal is to create a cabinet that uses the lowest density wall infill percentage possible while not impacting frequency response. So far, it kinda looks like this particular box may not benefit from additional infill percentage. Granted, the walls are 18mm thick and have a few features that inherently stiffen them a bit so we should probably not assume 10% wall infill is good enough in other situations. It is unclear whether additional bracing is necessary. The stuff I printed was far too tight to install so I'll be working on that when I can. I think at a minimum adding a TPU brace to damp vibrations in the side panels would be prudent, but if it can't be measured then maybe not?

 

[BossBunos]

Hey thanks for the reply! The port diameter is variable in this case as I used the port optimizer to model it. The minimum diameter is 38mm (center of port) while the ends are 57mm. Length is 125 mm...so the answer might be yes, in this case? 57mm is approximately 1/2 of 125 mm. This was intentionally modeled at 55 Hz by the way, based on some earlier comments in this thread that suggested the standard port was tuned a little low.

I need to reread augerpro's studies. I could try printing this with holes in it, my understanding is that sometimes helps with resonances. I could also print some or all of it with TPU to make it a bit flexible - I think a manufacturer does that for some reason. Again, need to read up on it.

Below is a revised graph showing the port (red line) and the QNF response for both the 10% infill cabinet and 25% infill cabinet. I did not go out of my way to match volume or anything between tests, but I think the takeaway is that the woofer response is pretty unaffected by infill density, even without any bracing present. There might be a slight difference around 30 Hz but that's pretty quiet anyway.

I would have shown the 25% port response too but I apparently didn't save that data this morning and cannot retest until later today. I do not believe there was anything remarkable in it though.

View attachment 547100

Something that may have improved though is distortion down low, but it's slight even when we're talking 100+ dB. This is probably where I should have taken a few tests at lower volumes and tried to level match a bit. I also don't think my mic interface is set up correctly for distortion measurements, so there may be some from that. It should be low though and probably inconsequential for this woofer.

Distortion: 10% Infill cabinet, no bracing.

View attachment 547101

Distortion: 25% infill cabinet, no bracing (note the response is a few dB louder than the above, so naturally some distortion is higher as well)

View attachment 547102

Edit:

Amir's port measurements also show two peaks, though at a bit lower frequencies and the first is not nearly as pronounced as I'm seeing.



View attachment 547103

Edit again:

Amir's distortion measurements may also be useful. Doesn't really look like the port resonance shows up much, but the bump I have in the impedance plot occurs here at just over 300 Hz (at least, I think they're related).

View attachment 547105

I would check what wavelengths correspondent with those peaks and see if you can translate that to some dimension of the port or inner cabinet sizes. As it is now the secondary resonance is only -5 db compared to the main output of the port. This is considerably worse than the original port that has one peak at -10dB. Maybe you could remeasure with some filling in the cabinet. Good luck!

 

[technodanvan]

For those unfamiliar with 3D printing I had Claude (AI) generate the below image so I didn't have to steal one from elsewhere. The construction of the cabinet walls come down to two things - the shell of the wall (aka wall loops) and what is inside the shell (the infill). Wall loops are simply a count - the more loops the thicker the shell, and this impacts all exterior surfaces including what might be considered minor features like screw holes, the T-slots I made, etc. The infill structure could be made in any of a few dozen ways, but a lot of people rely on the 'gyroid' structure which supposedly offers (roughly) equal strength in all directions. Whether this is a benefit to us is something that should probably be tested at some point (if it hasn't already), but it's what I'm using for a baseline. Downside of gyroid is it can take a slicer software longer to process (irrelevant to my beast of a computer) and it takes longer to print compared to some simpler structures like 'grid'.

The other part of the infill is the density of it, the more dense it is the less air is inside the wall. That's what I had been concerned with up til now - I printed a 10% wall infill and 25% wall infill, thinking that the denser one may have a beneficial impact to cabinet resonances. However, the brief, amateurish testing I did suggests they were pretty much identical. Another guy on YouTube took this a step further at 30%, 60%, and 100% and had similar results, though I don't know how many wall loops he used. I think this is for a couple of reasons:

• My test subjects both used 6 wall loops, which is probably on the high side.
• The structure is not simply flat like a wooden cabinet - it has features for the screw holes plus the t-slots. I think these features shore up the cabinet a bit - especially when the baffle and rear are bolted on (eight bolts on each).
• The speakers are already reasonably small anyway and use 18mm thick sides (as specced in the original wood cabinet)

For my own designs (upcoming, not related to the Mechano23) I'll try to reduce the overall thickness of the wall to 12mm to see what happens - this could allow greater volumes or a slightly more 'svelte' appearance. But for the Mechano23 we don't want to stray too far from the spec so I don't see a reason to do this.

I also don't see any reason to make a simpler cabinet just to test the efficacy of the T-slots. Perhaps another time.

However, I do see value in reducing the wall loops if we can - this would result in faster print times and less material used. Consequently, I'm reprinting the 10% wall infill cabinet with only 3 wall loops instead of the original 6 to see what the difference may be.

As a reminder, all tests have been performed without any bracing, wall treatments or stuffing up until now. I'm working on the braces still as they were simply too tight to work. I'm considering reprinting the port in TPU to see if it would damp those resonances. I'd prefer to integrate it in a brace somehow but I don't know how to do that with the port optimizer output - it generates a mesh that I can load into CAD but I'm unfamiliar with how to work with it as I've just done modeling with solids.

Anyway, it's all in work in progress and I appreciate all input I've received thus far. You all are a great help!

 

[technodanvan]

I've played around with a few things the last couple of days in an aim to limit resonances previously seen. I think I'm at the point where I need to finish assembling a crossover and test the whole speaker.

Summary of what has been done:

• Tested various levels of acoustic stuffing in each box type.
• Tested denim insulation in the same.
• Tested combinations of acoustic stuffing and denim in the same.

In general, various levels of stuffing have taken care of the standing waves despite no inner treatments to the box (no butyl rubber or similar). Next step is to attempt to take care of what I think is a panel resonance due to the unbraced side panels. There I got stuck for a couple days as all the braces I had were simply too tight to install. Original plan was PETG 'beams' the go into the T-slots with a K- or X- shaped TPU bracket that connected them. I have since reversed this order with TPU beams and a PETG connecting brace. I'm not confident this did much based on the impedance plot. It's too hot to run frequency sweeps right now (the a/c is basically running nonstop in my office) so I'll try to do that in the morning.

This is the impedance plot for the 25% wall infill cabinet, 6 wall loops, with TPU/PETG brace, with stuffing. That 280 Hz blip remains pretty much unchanged from other plots. Note however the tuning drops to roughly where I had aimed for (55Hz) so that's nice to see.

I am printing slightly smaller PETG beams to try a fully rigid brace to see if that makes a difference. Might be able to test that tomorrow too. I think if it remains I'll just leave it alone. It hasn't been particularly visible in other measurements which means it probably isn't audible anyway. I'm probably too caught up in the weeds on this.

Test subject below.

So yeah, after a couple more tests I'm thinking it's best to move forward with the crossover and test the full thing to see what that controls. Unless anyone else has some ideas before I move forward?

 

[BossBunos]

I've played around with a few things the last couple of days in an aim to limit resonances previously seen. I think I'm at the point where I need to finish assembling a crossover and test the whole speaker.

Summary of what has been done:

• Tested various levels of acoustic stuffing in each box type.
• Tested denim insulation in the same.
• Tested combinations of acoustic stuffing and denim in the same.

In general, various levels of stuffing have taken care of the standing waves despite no inner treatments to the box (no butyl rubber or similar). Next step is to attempt to take care of what I think is a panel resonance due to the unbraced side panels. There I got stuck for a couple days as all the braces I had were simply too tight to install. Original plan was PETG 'beams' the go into the T-slots with a K- or X- shaped TPU bracket that connected them. I have since reversed this order with TPU beams and a PETG connecting brace. I'm not confident this did much based on the impedance plot. It's too hot to run frequency sweeps right now (the a/c is basically running nonstop in my office) so I'll try to do that in the morning.

This is the impedance plot for the 25% wall infill cabinet, 6 wall loops, with TPU/PETG brace, with stuffing. That 280 Hz blip remains pretty much unchanged from other plots. Note however the tuning drops to roughly where I had aimed for (55Hz) so that's nice to see.

View attachment 548238

I am printing slightly smaller PETG beams to try a fully rigid brace to see if that makes a difference. Might be able to test that tomorrow too. I think if it remains I'll just leave it alone. It hasn't been particularly visible in other measurements which means it probably isn't audible anyway. I'm probably too caught up in the weeds on this.

Test subject below.

View attachment 548239

So yeah, after a couple more tests I'm thinking it's best to move forward with the crossover and test the full thing to see what that controls. Unless anyone else has some ideas before I move forward?

Really nice! I have the same resonance in the impedance sweep. My cabinet has a horizontal full width/depth brace. I read somewhere (not specifically for this driver) that fixating/supporting the magnet might remove the resonance. Using a brace/panel with some foam pressing on the magnet. I have not tried it

 

[technodanvan]

This is interesting, I'll have to evaluate if these last few attempts don't do much. I did determine the TPU/PETG brace did less than a 100% PETG brace, so we've ruled out that additional level of complexity. That seemed to move a resonance a bit but not eliminate it. I'm wrapping up the design of a new crossover brace for the lower half though as I definitely feel the lower side panels exciting still. After I try that I have some butyl sheet I might add in a few places. I'd be a little surprised if the resonance is the frame since it's plastic but weirder things have happened I'm sure.

 

[klapper]

Nice to see so many Well engineered 3D print approaches. I was unsere if a fully printed baffle would change the sound so i came up with a wood/3D Print mixture. Unfortunately, this makes my baffle very thin, which limits the complex shapes I can create. But i am still VERY happy with the result

A few other things in CAD I did this morning...

View attachment 545001

 

[technodanvan]

That's a sharp looking speaker you have there!

I just got back to my office after an extended trip, hopefully I will have some new measurements later this week.

 

[technodanvan]

It's been a minute since I could work on these. I decided wanted to wrap up cabinet design for the Mechano23. Still working on that impedance blip but I have an idea or two for that.

In the meantime, I have printed a new cabinet (15%, 6 wall loops), new baffle (that needs reprinted as I neglected to resize the bolt holes for inserts), and new gaskets. The gaskets are now designed to surround each driver to fully isolate them from the baffle, plus it could add a nice ring of color depending on the TPU used, seen here in white on the scrap baffle I made.

Yes, I'm going with a faceted baffle despite the design calling for flat edges. I just think a faceted baffle looks a bit more visually interesting. I will run some sweeps to verify nothing dramatic has changed.

 

[DearSX]

Wow this is a great project!

 

[BoysRBackinTown]

It's been a minute since I could work on these. I decided wanted to wrap up cabinet design for the Mechano23. Still working on that impedance blip but I have an idea or two for that.

In the meantime, I have printed a new cabinet (15%, 6 wall loops), new baffle (that needs reprinted as I neglected to resize the bolt holes for inserts), and new gaskets. The gaskets are now designed to surround each driver to fully isolate them from the baffle, plus it could add a nice ring of color depending on the TPU used, seen here in white on the scrap baffle I made.

Yes, I'm going with a faceted baffle despite the design calling for flat edges. I just think a faceted baffle looks a bit more visually interesting. I will run some sweeps to verify nothing dramatic has changed.

View attachment 551616

View attachment 551617

Great! When you have would you share the stl files?

 

[technodanvan]

I'll share the files when I have fine tuned everything to work on my printers. No guarantee further fine tuning would not be necessary for different printers of course, but I want to be sure we have a complete working set first!

I should point out that I'm a bit delayed right now as I work on another project in hopes of completing it in time for IowaDIY in mid-October (preliminary renders of that project shown below). This is similar in that it is 3D printed and I'm using a lot of what I learned from the Mechano23 enclosure development, different in that it is, well, pretty darn different (and is completely my design with different drivers, crossover, tuning, etc.). I also have some significant home remodeling going on that is stealing my time. Always busy these days! Consequently, I probably can't share the final Mechano23 files until the November/December timeframe.

 

[johnp98]

I am about to build two pairs of these, one for myself and one for a friend and I have looked at lots of the builds linked at the beginning of this forum, so thanks for those!

I just had some final questions:

Components
I already have the parts for one of them, but need to order the parts for the second pair. Just making sure there are no changes or upgrades to the original design that was reviewed here: https://www.audiosciencereview.com/...chano23-open-source-diy-speaker-review.54066/
M23 components at solen.ca
Ind. 0.33mH link CA$5.10
Ind. 0.56mH link CA$6.73
Ind. 1.5mH link CA$10.58
Ind. 0.39mH link CA$5.81
Res. 12ohm link CA$0.41
Res. 15ohm link CA$0.41
Res. 8.2ohm link CA$0.41
Cap. 6.8uF link CA$2.39
Cap. 0.82uF link CA$6.09
Cap. 27uF link CA$1.73
Cap. 18uF link CA$6.39
Cap. 3.3uF link CA$1.41
Cap. 2.0uF link CA$2.48
Midwoofer SB Acoustics sb13pfcr25-4 link CA$34.44
Tweeter Scan Speak h26069200 link CA$59.75
Any changes to these?

Port
As for the port, I was going to 3D print this port as recommended previously here: https://www.audiosciencereview.com/...axis-measurements-included.41757/post-2553417

Unless others have thoughts on a better port that can be 3D printed.

Build
As for the build itself. I was going to follow the construction guide made by @Woody99
Anything that people recommend otherwise?

I was probably going to extend the depth of the speaker just a tiny bit to allow for more volume for a bit more bracing (probably window bracing in the middle and maybe a strip going vertically on the sides and back?), but I am open to suggestions.

Also the edges I was probably going to round the edges to reduce diffraction. I assume that's a good idea if I have the means, correct? Or does it change the frequency response at all?

Should I use gasket tape for the woofer and tweeter? Or not worth the hassle?


Anything else that I should consider? Thanks!

 

[technodanvan]

I’m on mobile so I apologize that I can’t give quite a complete answer, but here are my thoughts:

First, I believe the parts you have listed are correct…unless you want the THD reduction filter which is also described in this thread. I seem to recall you omit one of the original caps and install the filter in line with the woofer.

Second, are you in Canada? The States? If the latter you may want to rethink ordering parts from Canada as I want to say there are new tariffs in place? Someone please correct me if I’m mistaken!

The woofer is less costly at TLHP, but perhaps not if you’re only buying two since they’re coming from (presumably) France. Be sure to log in the to see the price, I want to say they were like 15 euros apiece last I looked.

I don’t think changing the box depth will hurt, but I also don’t think additional bracing is all that necessary. We are fairly certain the resonance Amir identified is not the cabinet at this point.

A lot of people have done the round over but I’m not sure any have measured it. Keep in mind the crossover was designed for no edge treatments so there may indeed be a change. A positive one? Maybe, hard to say without measurements comparing the two. I will try to do that with my faceted baffle if I can but it’ll be a while.

I think a gasket is a worthwhile endeavor. The woofer actually comes with one IIRC, but I seem to remember it being quite thin and just not great overall. I’m sure it’s fine but it bugged me that it seemed someone cheaped out on it a little.

 

[XMechanik]

Anything else that I should consider?

There's also a crossover modification that reduces THD around 2.4kHz link. You can simply add a 5.1uF/0.1mH parallel resonant circuit to the midwoofer filter output. It might also be worth considering changing the tweeter resistor to 12ohm, which will slightly reduce the Sound Power characteristic slope. If you keep the original resistor (15ohm), you can get rid of the 0.82uF capacitor, then the whole crossover has only one component more comparing to the original version. For the modified resistor value however, I think it's better to keep the 0.82uF cap.

 

[johnp98]

I’m on mobile so I apologize that I can’t give quite a complete answer, but here are my thoughts:

First, I believe the parts you have listed are correct…unless you want the THD reduction filter which is also described in this thread. I seem to recall you omit one of the original caps and install the filter in line with the woofer.

Second, are you in Canada? The States? If the latter you may want to rethink ordering parts from Canada as I want to say there are new tariffs in place? Someone please correct me if I’m mistaken!

I am in Canada; this is the only time it has been advantageous for building speakers!

I don’t think changing the box depth will hurt, but I also don’t think additional bracing is all that necessary. We are fairly certain the resonance Amir identified is not the cabinet at this point.

Perfect, thanks. Then I won't overthink it.

A lot of people have done the round over but I’m not sure any have measured it. Keep in mind the crossover was designed for no edge treatments so there may indeed be a change. A positive one? Maybe, hard to say without measurements comparing the two. I will try to do that with my faceted baffle if I can but it’ll be a while.

I think a gasket is a worthwhile endeavor. The woofer actually comes with one IIRC, but I seem to remember it being quite thin and just not great overall. I’m sure it’s fine but it bugged me that it seemed someone cheaped out on it a little.

Done and done. Thanks!

 

[johnp98]

There's also a crossover modification that reduces THD around 2.4kHz link. You can simply add a 5.1uF/0.1mH parallel resonant circuit to the midwoofer filter output. It might also be worth considering changing the tweeter resistor to 12ohm, which will slightly reduce the Sound Power characteristic slope. If you keep the original resistor (15ohm), you can get rid of the 0.82uF capacitor, then the whole crossover has only one component more comparing to the original version. For the modified resistor value however, I think it's better to keep the 0.82uF cap.

Sorry, I might be dense with this stuff (I've only built one set of speakers before).
But what part(s) would I delete / change:
Ind. 0.33mH link CA$5.10
Ind. 0.56mH link CA$6.73
Ind. 1.5mH link CA$10.58
Ind. 0.39mH link CA$5.81
Res. 12ohm link CA$0.41 - So now I would get two of these?
Res. 15ohm link CA$0.41 - So I can change this to 12ohm?
Res. 8.2ohm link CA$0.41
Cap. 6.8uF link CA$2.39
Cap. 0.82uF link CA$6.09 - Get rid of this one?
Cap. 27uF link CA$1.73
Cap. 18uF link CA$6.39
Cap. 3.3uF link CA$1.41
Cap. 2.0uF link CA$2.48

And then what would I add, as there appears to be a few options:
0.1mH inductor: https://www.solen.ca/en/products?query=0.1mH inductor
5.1 uf cap: https://www.solen.ca/en/products?query=5.1 uf cap

Sorry for not being able to figure it out. I am also happy to build the base version if this is too difficult.

@XMechanik if you ever build a larger speaker with more output designed for home theaters, I am sure it would take the hobby by storm. Just putting it out there

 

[XMechanik]

And then what would I add, as there appears to be a few options:

Original crossover:


For the modified version (link), add a 0.1mH/0.1ohm coil and 5.1uF cap to the list. Could be any source, not necessarily Solen.

Adding parallel resonant circuit for THD reduction:


Note that the 820nF (0.82uF) capacitor is not installed. It can be removed from the list.

Alternative mod option: changing the tweeter resistor (15->12 ohm) and adding parallel resonant circuit for THD reduction:


Note that 820nF cap is kept but the 15ohm resistor will no longer be needed. It can be removed from the list but you will need additional 12ohm resistor (a total of 4 per pair, as another 12ohm resistor is used in the midwoofer filter).

 

[johnp98]

Ok I am going to build the boxes in two days with someone who is retired and has a woodworking shop so I suspect I'll get much nicer results with him involved.

I tend to fixate on any possible improvements until it is made, and then I just enjoy it. So here are some questions I have about the box:
I was going to use 3/4 plywood as that is what I have leftover from another project, but it is just 1mm thicker than the original design. I could extend the depth a bit as noted previously.

I also have this weird leftover material from making another speaker that is like mass-loaded vinyl glued onto a thin foam layer. I used it as almost a constrained-layer damping (CLD) effect where I glue the foam to the interior walls and it mass-loads the walls and is fantastic at dampening the walls themselves.

It is about 5mm in thickness so if I added it to every interior wall, then if my math is correct it would take up around 1L in volume.
I was also probably going to brace the walls with straps as outlined here.

All of that takes up internal volume, though.

So I could extend the depth of the speaker but then as noted #580, if the depth increases, then the internal nodes get closer to overlapping so I don't know what people would recommend.

What would people recommend?
Keep the original measurements and don't sweat the lost volume with it being 3/4 wood? (I am happy to be told that I am over tinkering and just going to make things wose)
Increase the depth and add the CLD / mass load vinyl and bracing?
What if I just kept the dimensions but then still added the CLD and bracing? What are the effects of lowering the internal volume (it would go from 8L to 7L), as I know it changes the tuning frequency, but I will be rolling these off to a subwoofer, so how low the bass goes is not important to me.
Some hybrid?

Thanks for any thoughts! I am excited to get these built!