Waveguides for 3D printers and CNC's
Collapse
X
-
- Bottom
-
Great work Brandon! The result of the better versions look very good indeed and should produce some nicely balanced directivity of the final speakers. The results do confirm again though that this subject needs a lot of trial error to get right. KEF guys use some hard core computer simulations to come up with stuff like the tangerine waveguide, but getting this close to similar performance with hobby level tools is great achievement.
Also thanks a lot for sharing the F360 model. You've used some new techniques to form these which I found interesting and useful
I've been off speaker building topics (building decking and roofing instead) Just today I received a note though that the Prusa 3D printer finally started to move towards Estonia. So that and coming autumn/winter period will hopefully boost the hobby activity again eventually.- Bottom
Comment
-
Thanks Ergo! Yeah I should have some fun designs for you to print and play with! I have three 8" guides being made right now, all elliptical. Hopefully what I've learned so far scales up in size, would make every size after very easy.
I'm really liking F360. It's such an art finding clean and elegant ways to design that will be amenable to changes later without having to build the whole thing again.- Bottom
Comment
-
Big update! This will be my last as the money to go on has run out. This is a big data dump as I haven't gone over it for trends, so I could use members help here to tease out best practices. I have several variants of the 5" guide including a shallower 1/2" deep version. I end with two 8" waveguides.
Above I had mentioned that a waveguide between "A" and "C" should be ideal. Here is that waveguide "E" with and without phase shield:
Early on I had discovered the terminals on the tweeter sometimes kept it from sealing against the back of the waveguide. So I wanted to try a thin shim to get a good seal. Here is that result with a .03" thick shim:
I wasn't sure if the difference was due to sealing or because the ever-critical throat geometry changed, but figured trying a thicker shim might shed some light on that. Here results using a .06" and .09" shim:
Now the phase shield is moving away from the dome as increase the shim, so in the spirit of good scientific method I felt I needed to measure with the phase shield back closer to the dome. Now this is not back to default distance since I had to cut the phase shield off and then using some bluetac moving it closer to the dome about 1/16".So it is close to default but not exact:
On the no-phase-shield version I did only one measurement with the .06" shim, I'll try to get others if I have time:
My attempts with an occluded surround have been poor but the last version I made the throat smaller, terminating above teh former of the diaphragm where the earlier version terminated above the surround. I went back to that larger, more promising throat here:
Here is the shallower 1/2" deep guides. The first is simply a shallower version of the earlier A,C,E type of guide:
As I go shallower, or much wider -like the 8" guides- the horn curve radius gets larger and/or the throat entry angle gets wider. Smaller radius and narrower (@80 degree) throat angles seemed better from earlier experiments. To accomplish that I tried the elliptical profile of the horn wall again. Here is that result for the 5 x .5" guide, I have not had the 8" version printed unfortunately. Both with and without phase shield:
And finally the 8"-ers! One is 1" deep, the other 1.5". With and without phase shield, and tried a few shims.
Here is the 1" deep guide:
And the 1.5" deep one:
The shims have a clear result worth investigating. Phase shields pretty much always have an advantage, but this has narrowed since going strictly to elliptical waveguides. Other things worth studying are how some of these features effect the directivity in the top octave, how the resonant peak is knocked down or amplified (I'm talking 20-23khz, at 24khz the soundcard cutoff causes all sorts of spurious spikes so ignore that), and the effect on mid-band smoothness (for example look at the 40 degree measurement across designs, some are straight and quite smooth, others very curved and/or bumpy). I hope to keep going, I think some designs are already the equal, if not better, than the Kef and Revel posted above.- Bottom
Comment
-
Here are results for the SB26 5" waveguide version F, and the SB19ST 5" waveguide version A. I'll get the TW29RN and N26CA out over the next couple days.
SB26ADC:
The midband is much smoother over previous results due to the .02" lip I incorporated after shims and putty showed the faceplate was not sealing well against the mouth of the waveguide:
Speaking of sealing, in actual use you want to seal a couple gaps on the tweeter that do not seal like you might expect. I would use blue tack putty, easy on ,easy off:
The SB19ST:
Unsurprisingly the phase shield was slightly worse with a fabric dome. OTOH I didn't expect that on axis dip even without phase shield. I believe the shape of the dome requires another look at throat angles as this a small diameter but tall dome. Nearly a 180 degree hemisphere. I'm *guessing* I need to relax the throat angle some.
One thing I've noticed so far is that everything looks much better on the 8" waveguides (results coming soon). I guess I assumed that a little waveguide would cause little problems, but that does not appear to be the case. In fact just going to say, 6" wide waveguides might solve a lot of the top octave issues I've had to solve. But I had to start somewhere!- Bottom
Comment
-
Your result keep getting better and better - great work !
My past experiment have all also indicated that the 'bigger bulge' of the soft tweeter dome gives troubles for waveguide use and the metal domes with lesser bulge less so. The XT25 dual ring radiator seemed have least issues with the throat matching, but that type of tweeter has other issues that are avoided with a good hard dome.- Bottom
Comment
-
Ergo> have you done anymore work on your own waveguides?- Bottom
Comment
-
That topic has been in hibernation along with the snowy winter around here I have the Prusa 3D printer up and running and I've practised with that by making vacuum hose fitting, handles for a saw, and misc other bits and pieces. I'm hoping to get back to both Ardent build and waveguide topics as spring comes along. It's time to put Prusa though a trial in printing a full WG just to see what comes out.- Bottom
Comment
-
Cool. Let me know and I'll send one of the latest designs instead of that older one you have.- Bottom
Comment
-
BlieSMa T34B-4 in 165mm diameter waveguide
Finally got around to doing some waveguide work. Thanks to Brandon for all the inspiration that helped push to get around to doing this.
Anyhow
The driver: BlieSMa T34B-4
Waveguide: 165mm diameter - profile is a section of a radius of bigger circle. Printed on Prusa i3 MK2.5 - 0.2mm layer with linear advance, material PETG
Test baffle - 220mm x 700mm baffle as intended for the final project (2.5way with 2x Satori MW16P)
Test conditions:
Mic to baffle distance for off axis measurements: 600mm
Mic to baffle distance for THD: 315mm ( to be compatible with hificompass.com test
- Bottom
Comment
-
CLIO output signal had an High Pass at 250Hz just in case (I did not want to kill my friends driver)
First a 0 to 60deg plots with relative wide dB range to make it similar to Brandon's SoundEasy measurement plots
Then a 0 to 100deg with 20dB less scale in vertical
And this is the 3.9uF capacitor added in series with driver to create a HP
THD at 1.41V input (mic at 315mm)
THD at 2V input (mic at 315mm)
- Bottom
Comment
-
In the off axis plots above I've chosen the 20deg off axis curve as reference - so that one appears a bit bolder. With Modula 2 ways this is about the way I listened to those - never turned them to exactly point to the ear. Soundstage seemed better that way
Also as seen there is no phase shiuld in front of the dome. Decided to try without and seems it is ok as is as the response is very smooth in 10..20k range where that would probably influence most. But maybe it would influnece the ca 7.6k small dimple... maybe I'll try to print one still.
Overall not bad result for a first guess try.... but this first guess has a lot of pre work from Brandon in it of course.- Bottom
Comment
-
Sorry for the absence, working a ton the last few months. But the good news is I've started a new website somasonus.net to host all my DIY stuff. It's barren right now, but I have posted the best of the waveguides with downloadable CAD files for 3D printing or CNC. Let me know if anything renders weird on the website, it really changes depending on what device I'm on.- Bottom
Comment
-
Cool. I've downloaded a couple of your files, but haven't done anything with them yet :-( . I 'think' I really want to try a MW13P + SB21 waveguide as the MT in a 3-way, but may not be able to scare up the time to do the design.
Have you collected any formal(ish) designs using your waveguides?
BTW - I'm still in 'Denver' (actually way SE) and have a Monoprice Ultimate and a 6040 CNC if you're looking for some local support. Your Statement Monitors are still sounding great in the theater space - I managed to score a Statement Center from the PE forum to round out the front stage.- Bottom
Comment
-
Cool. I've downloaded a couple of your files, but haven't done anything with them yet :-( . I 'think' I really want to try a MW13P + SB21 waveguide as the MT in a 3-way, but may not be able to scare up the time to do the design.
Have you collected any formal(ish) designs using your waveguides?
BTW - I'm still in 'Denver' (actually way SE) and have a Monoprice Ultimate and a 6040 CNC if you're looking for some local support. Your Statement Monitors are still sounding great in the theater space - I managed to score a Statement Center from the PE forum to round out the front stage.
What kind of build volume is that CNC? I'd like to get a Shapeoko.- Bottom
Comment
-
These guys are using my waveguide with a Scanspeak tweeter. Not elliptical, but round, which reminds me, I forgot to put the round one up! The got killer pricing for CNCing in aluminum.
What kind of build volume is that CNC? I'd like to get a Shapeoko.
I have a bit of a budget for hobby stuff that will have to be split between audio and shop/cnc, and just picking up a shapeoko is an idea I return to occasionally. I ultimately don't think I can accept belt drive though, and an XXL is still not really the right work envelope - I think a 2'x4' style setup is more useful than the 32"x32" capacity of the xxl. I've also basically spent all the $$$ to get my design running and am limited by time and focus, so dropping almost $2k isn't terribly appealing.- Bottom
Comment
-
the 6040 is a small/midrange machine. roughly 16"x22"x3.5" work volume. I am (slowly) building up a larger ~50"x32" setup of my own design, but I suspect it'll be next summer at the earliest before it's operational. It's at the 'rolling chassis' level at the moment, needing to get the rails leveled and secured, and the ballscrews mounted. I have a basic electronics kit as well. The main unknown at the moment is the Z axis assemble and potentially the spindle choice.
I have a bit of a budget for hobby stuff that will have to be split between audio and shop/cnc, and just picking up a shapeoko is an idea I return to occasionally. I ultimately don't think I can accept belt drive though, and an XXL is still not really the right work envelope - I think a 2'x4' style setup is more useful than the 32"x32" capacity of the xxl. I've also basically spent all the $$$ to get my design running and am limited by time and focus, so dropping almost $2k isn't terribly appealing.- Bottom
Comment
-
I don't know if you have looked at them, but the I saved a considerable chunk of change when I built my machine from these guys. http://www.cncrouterparts.com/index.php- Bottom
Comment
-
Oh, very familiar with them. They're reasonably priced compared to big industrial machines, but for hobbiests on a budget they're something of a 'cadillac' solution. A 2x4 with linear rails from CNCRP runs $3500 I think, and even their 'bearings on steel' version is $2500. I think I'm into my project for about $1k so far, and will probably be looking at another $500 to finish depending on spindle choice. If I had any intention of making money with the machine, the I'd agree that CNCRP is a great path, but for someone like me where it's just for fun and will likely sit idle way too much of the time, it's tougher to justify.- Bottom
Comment
-
- Bottom
Comment
-
Ok first results are SB26 5" version H. It doesn't get much better, so I'm calling the 5" done. Note whether a phase shield is included.
For fun I wanted to go with a more round mouth, so here is a 5" but with mouth aspect ratio of 1:.80
Not really anything to gain. But I will try a strictly circular mouth just to be sure I didn't go down the wrong path when went with elliptical mouths early on when the first circular mouths weren't very good.
Now the Bliesma T25B tweeter. I really like that the best response are without phase shield. I did a 5" with my normal 1:.618 aspect ratio mouth and another using 1:.80, two 6" of varying depth, and one 6.5".
A note about the high frequency spike @48khz: this is a combination of the mic calibration no longer being very accurate and the math of the FFT. None of these tweeters actually spike the high.- Bottom
Comment
-
Am I missing something or does the phase shield only seem to help the SB26ADC above 15kHz and beyond 30* off axis? I'd expect the phase shield to make the tweeter seem less bright, but my eyeball isn't translating the chart into a more practical effect.Welcome to Rivendell, Mr. Anderson.- Bottom
Comment
-
- Bottom
Comment
-
It hasn't been that case in previous results, but in this particular waveguide, yeah, not much difference between the two.- Bottom
Comment
-
That last photo puts these things in perspective....
These are quite large!
Beautiful quality though. Are you using the 3D print to silicone mold and cast from ABS?- Bottom
Comment
-
No, these are SLS nylon prints from a commercial vendor. A laser fuses powdered nylon together. Similar to SLS resin, but powder instead of liquid and MUCH stronger. Can't wait until this technology is available to hobbyists, but that might be a few years.
BTW the T25B is quite a small tweeter, almost automotive size.- Bottom
Comment
-
And yet again: thanks!! I may look at a new round of prints. How much trouble was it to remove the grill from the Bliesma? I'd assume that the Al version works similarly to the Be version, but that may be an inference too far.Welcome to Rivendell, Mr. Anderson.- Bottom
Comment
-
Very easy, just gently pry the screen off. It uses some sticky substance to stay in place. Easy to put back on.- Bottom
Comment
-
I've started a gofundme for this project now that I'll have to depend on commercial printers. Initial goal is for finalizing designs for the SB21 and SB26 families, and continuing development for the Bliesma T25B. Fundraiser: Waveguides for CNC and 3D printing!- Bottom
Comment
-
I'm not seeing version H on your website. Am I just missing it? Given that the phase shields have not stood the test of time on my earlier prints, I'd like to go ahead and get a fresh pair printed without the phase shield.Welcome to Rivendell, Mr. Anderson.- Bottom
Comment
-
I was going to put it up on my site after a series of test prints with my 3D printer to make sure even hobbyist printers will do well with the design. Those tests resulted in a couple tweaks for a final version J that I need to measure and verify first.- Bottom
Comment
-
Cool! Thanks for the diligence and rigor!Welcome to Rivendell, Mr. Anderson.- Bottom
Comment
-
Might be making big jump forward in waveguide design. I started using mabat's ATH program to generate the files that ABEC uses to simulate waveguides, including a representation of the tweeter diaphragm itself as acoustic source! Here are the simulated results of the T25B in ABEC, followed by my actual measurements:
BTW: I believe the off axis hump @ 15khz is the natural response of the tweeter itself, so not captured by the ABEC sim. I've measured the domes of the T25B, SB26ADC/CDC, SB21SBC, and Dayton RST28A. If all works out well, it might be nothing more than sim once, print once.- Bottom
Comment
-
This looks like a very good match between model and reality indeed. I've looked briefly into ABEK several times too, but it's always looked like much more time is needed to learn it enough to get any results. Superb that you are making that leap. Community as a whole will benefit for sure, so thanks a lot for the efforts- Bottom
Comment
-
- Bottom
Comment
-
Welcome to Rivendell, Mr. Anderson.- Bottom
Comment
-
Test baffle effect maybe? This starts to show up <4khz in this size waveguide anyway.- Bottom
Comment
-
More Bliesma T25B and SB26 results! Reminder these are 7ms of clean signal, no smoothing. Depending on waveguide size, baffle diffraction starts to show up @ 3-4khz. Basically if you see the responses getting wavy and crossing each other as you move off axis, baffle diffraction is showing up and the waveguide is losing pattern control.
- Bottom
Comment
-
- Bottom
Comment
-
- Bottom
Comment
-
Probably 30-ish hours. I don't know about sanding these things, it's all new to me. I think the current situation is if you must have a smooth finish either use a commercial outfit and print in SLS nylon, or use a CNC. As technology advances hobbyist 3D printers will also get pretty smooth, but it's not currently there as far as I can see. I mean you could try a tiny nozzle which would enable higher resolution in the slicer, but print time would get much longer. I should try a test with .02 nozzle.- Bottom
Comment
-
- Bottom
Comment
Comment