Waveguides for 3D printers and CNC's

[wkhanna]

ok, so the WG has been sitting on my desk for two weeks now.......

i need to determine how to proceed with smoothing out the surface finish without disturbing the critical surface profile.
a few possible courses of action:

1.

i can use a filler, then sand down to the original profile

2.

i can just sand down the ABS surface as is

3.

i can put it an acetone fume chamber (which has the potential to produce an uneven surface as it 'melts' the top layers of ABS along with dissolving the phase shield support)

in all cases, the phase shield is a problem.
i am considering removing it, blending the surface to the required smoothness, & then reattaching it.

what are your thoughts on this approach, Brandon?






edit: the reason for the current surface finish has to do with diameter of the ABS filament & the resolution parameters that were programmed/available for the material. getting a better surface finish will require a smaller diameter filament & feed tips, & possible a different material such a nylon, all of which are not available to me at this time.

[Steve Manning]

edit: the reason for the current surface finish has to do with diameter of the ABS filament & the resolution parameters that were programmed/available for the material. getting a better surface finish will require a smaller diameter filament & feed tips, & possible a different material such a nylon, all of which are not available to me at this time.

Bill would that increase the fabrication time as well going for a higher resolution, smoother finish?

[300Z]

I wonder how these imperfections from the printing would actually affect the response?
Would be nice to see some before and after measurements.

[augerpro]

I second 300Z's comments. Bill do you have reliable measurement capability?

My current solution is to go to CNC in wood. In the future a buddy of mine is getting a Pegasus 12 3D printer. Claims 50 micron resolution, hopefully we can make that happen after dialing in the settings. It will use 1.75mm hot end. What are your specs and settings Bill?

[wkhanna]

I second 300Z's comments. Bill do you have reliable measurement capability?

i just purchased a calibrated UMIK-1, but am still working thru understanding & setting up REW.
with your help on set-up settings & component positioning i may be able to get you some numbers....?

My current solution is to go to CNC in wood. In the future a buddy of mine is getting a Pegasus 12 3D printer. Claims 50 micron resolution, hopefully we can make that happen after dialing in the settings. It will use 1.75mm hot end. What are your specs and settings Bill?

i also used 1.75 mm filament
there are some tweaks i could make to help improve surface finish, but getting a good blend on the contour is a challenge with this size filament

Bill would that increase the fabrication time as well going for a higher resolution, smoother finish?

not necessarily, as the higher resolution is only required on the final pass for the critical surfaces

[300Z]

i just purchased a calibrated UMIK-1, but am still working thru understanding & setting up REW.
with your help on set-up settings & component positioning i may be able to get you some numbers....?

You could use the tips here and here to get started.

[augerpro]

Bill was that a .4mm tip? What resolution was the machine set at?

[wkhanna]

Brandon,

our machine is a true "industrial production" specification unit.
unlike most smaller, low production units, its actual "resolution" parameters are governed by, among other things, tip diameter, slice height settings & material.

as we are still experimenting with & learning the capabilities of this machine, keep in mind your part was run at our current 'default' setting.
for your first sample, slice resolution was set at 0.010 inches. (note that there are a range of resolutions available with the same tip on our machine)

i currently have the ability go to 0.005 slice height with the smallest tip we have at this time.
along with other parameters i can manipulate, i can produce a much better surface finish.

[wkhanna]

You could the tips here and here to get started.

thank you, Leo.

[300Z]

thank you, Leo.

You're welcome.
Now if you feel like reading a bit this is a good help too. Much props to Jeff for putting it together.

[wkhanna]

that is a V good link, Leo.
thanks again.

btw, i just did reading on Recife......sounds like a really interesting place.....lots of 'old world' charm mixed with a booming tech industry.

[300Z]

Recife is an interesting city indeed but unfortunately it's riddled with what I call incompetence from the government, so many things that are wrong and wouldn't take much to fix but for some reason nothing is done about it. I've been here for a little over 3 years now, I like it but from all the other cities that have travelled here in the northeast Recife is the most disorganized (I'm a little OCD so perhaps take that into consideration), but I have met lots of cool and interesting people here tho.

[augerpro]

i just purchased a calibrated UMIK-1, but am still working thru understanding & setting up REW.
with your help on set-up settings & component positioning i may be able to get you some numbers....?

Calibrated mic is a good start, but I'm not too familiar with REW. I've used the RTA before is about it. Just be sure to use a gated MLS, and end the window before the first reflection. You don't need a terribly big baffle for this, just do a quick sim in EDGE or something to find a smooth response above, say, 1.5khz. Make sure you don't use a typical mic holder, they cause reflections right in the treble. Main thing is getting good comparative results, not so much absolute accuracy (for this particular task).

[wkhanna]

cool.
i just need to what driver i need to get & how you are mounting it.
we can handle these details via email, if you prefer.

[augerpro]

Email, here, don't matter to me. I'm using the SB Acoustics SB26ADC-C000-4. Tailor made for a waveguide. Just unscrew the 4 screws on the faceplate, and the faceplate comes right off. It doesn't hold the dome down or anything so there are no worries.

[augerpro]

So the phase shields and mic mount have arrived. Looks like really good quality, though I haven't checked with a caliper yet. If anyone wants to print the mic mount here is the file: https://www.dropbox.com/s/eijmpzwauz...reads.stl?dl=0

This was a 25 micron print in resin at leaderprototyping.com

[augerpro]

So the waveguides from Tim have arrived and the phase shields are in place. Looking good so far. To reiterate these are all 5" waveguides. Depth is varied for one pair. For the other pair the horn wall is made either tangent or non-tangent at the mouth. The tangent mouth ones have constant flare radius of 2.xx" (can't remember exactly), and the non-tangent mouth ones have a set radius of 4", so much shallower curvature. I have to make a flat baffle mount so the results can be compared to that, hoping to get a friend's 3D printer working this week to make that.

[300Z]

Looking good. Is that an RS28A?

[Carl V]

I'll be designing the first guide to be used with the SB Acoustics SB26ADC-C000-4. Once the best practices are established, I'll move on to my Transducer Lab ceramic tweeter. After that I'm hoping someone will supply a few good candidates, like a good Scanspeak.

Unless something has changed.....

[300Z]

I forgot about that. Thank you.

Unless something has changed.....

[Evil Twin]

There is no doubt that properly implemented, waveguides may be of substantial usefulness in extending the performance of conventional dome tweeters...






I encourage your further development efforts...

[augerpro]

Measurements start tomorrow! Finally! I stumbled across a new source for waveguides and hope to be able to tear through some iterations as fast as possible. Finish quality is excellent, although not smooth, similar to 100-200 grit sandpaper. Some sort of laser-sintered nylon process I've never heard of. Anyway, I had the four designs that Tim supplied me in CNC'ed wood above printed on this new process. So some comparisons between the manufacturing method should be helpful.

Right now all the designs are 5" diameter mouth, with two variables to play with: depth (1" vs 3/4") and whether the mouth is tangent to the baffle or not. Depending what the results are I'm going to play with wall radius more, and with and without phase shield. Once I find "best practices" at 5" I'll design some 8" waveguides and see if those hold. If so, I can easily extrapolate in between and just print a couple samples to verify the prediction, instead of doing a bunch of variations at every size.

[300Z]

I like it, looks really good. Can't wait to see the measurements.

[augerpro]

Ok here are the first results. This is data dump, haven't really looked at the results myself in detail yet. A little about my measurement setup: I use MLS measurements with 6ms of clean signal for a quasi-anechoic response, no smoothing, and a calibrated mic. Pictured is the baffle, I don't have my big baffle anymore, so I optimized this the best I could, it does appear that under about 4khz the baffle is having a strong impact. Now IF the waveguides are doing there job, this effect should go away. This first batch is the nylon sintered ones, with phase shield. I'll be posting each update at my Soma Sonus site, so that will be easier reading than this thread.

All waveguides are 5" diameter mouth. The rest of the info is listed below.



Flat baffle for reference, 0-10-20-30-40-50:



3/4" deep, tangent mouth, 0-10-20-30-40-50:



0-30-60:



profile:



3/4" deep, non-tangent mouth, 0-10-20-30-40-50:



0-30-60:



profile:



1" deep, tangent mouth, 0-10-20-30-40-50:



0-30-60:



profile:



1" deep, non-tangent mouth, 0-10-20-30-40-50:



0-30-60:



profile:



Next I'll see what it is like without phase shields, play with phase shield placement a bit, lower radius non-tangent mouth versions possibly even a straight cone, and a roundover on the non-tangent versions.

Any other measurements you'd like to see before I cut the phase shields off?

[ergo]

Cool to see a progress with this. The tangent mouth seems to have a small benefit on axis in a 2..10kHz rang. It's interesting that a bit off axis the small hump at 4..7kHz disappears.
Thanks a lot for sharing the results!

I have not had any time during the spring to make experiments with the milled WG-s. I did pull a trigger ordering one of these yesterday

http://shop.prusa3d.com/en/3d-printers/59-original-prusa-i3-mk2-kit.html

So once I get it sometime in August probably it will be interesting. I did also order the 4 material upgrade which arrives later in autumn

http://www.prusaprinters.org/update-multi-material-upgrade-2/

that would allow adding a water soluble support structures to the model, so in theory even printing the WG with the phase shield becomes possible, but it's probably more practical still to do it separate.

[augerpro]

I heard that Mk2 is a damn good machine. Should get you quality prints.

Yeah I'm not sure whether difference is the mouth, or something related, like the fact the flare radius is much tighter on the tangent mouth specimens. They are 2.2-2.6", while the non-tangent are both 4" radius. I'm going to try some things to tease out the difference. Also the larger radius will meet the dome at a shallower angle, so maybe it is too shallow? That would be the possibility I like the least since it implies that every dome shape will need to have the radius tweaked by experiment to match.

[augerpro]

Playing some more with the phase shield. This is on the .75" deep, tangent mouth guide. Let me know if the bolder colors are easier to read or not.

First the original version, with .47" wide phase shield:



Widdling the PS down to roughly .38":



Smaller yet, .25" wide:



And no phase shield:

[augerpro]

Another phase shield comparison, this time on the deeper 1", tangent mouth waveguide.

With original .47" wide PS:



With approx. .38" wide PS:



With approx. .25" wide PS:



And last, no phase shield:

[augerpro]

Moving on to the shallower .75" deep, non-tangent mouth waveguide.

Original .47" wide phase shield:



Skipped the .38" wide PS and went straight to .25" which seems like the sweet above:



And no phase shield:

[augerpro]

As people have discovered before, what is going on right at the throat is the most critical factor. I've got some other phase shield designs to play with, but I'm also going to have the wg that overhands the surround made soon, I think that will be another critical test. I also wonder how the throat angle effects things, but it might take going to the larger wg's to provide room to play. As it is the tangent mouth wg's around low/mid 50 degrees, while the non-tangent wg's are around 40 degrees.

[ergo]

It's interesting that the phase shield effect seems largely affecting only the range from 15kHz up. Would be very interesting if at age of 40 or 50 there will be an audible difference too or not. Microphones do hear better then we do up there for sure.

[Zvu]

I have Visaton WG148R - 5.83" waveguides. I can call my friend to copy the curve and send it to you to try it if you want to experiment with it.

[augerpro]

Zvu> looks like an elliptical profile. I was going to play with that today. I would be curious what the throat angle is if you can get that info?

[Zvu]

I wouldn't know how to get the info you need but if there is a way to measure it, i can try that.

[Zvu]

Hey Auger,

I mounted Morel DMS30s (DMS30s, MDT30s, DMS32s, MDT32s and CAT308 are the same tweeter but different cosmetics) in a Visaton WG148R. It's mounted in 13 liters cabinet. Front baffle dimensions are HxW - 33cm x 17.6cm. Measurements done by Holmimpulse from distance of 1m at tweeter axis, gate is set at 270Hz. Vertical distance between driver centers is 15cm.

Here it is:



I measured Morel with and without WG148R, on axis so there is diffraction dip at 4KHz:



Acoustic centers offset between Seas midwoofer and Morel tweeter is just 5mm at Z axis with tweeter in WG148R.

I simulated an xover to test in a couple of days, there is just a few components on tweeter:

[augerpro]

Wow that worked out really well! Do you have any pics?

[Zvu]

Of course...

Images not available


It was quite simple and fast. Mounting went smoothly because Morel has flat aluminium front plate. I haven't used any filler between waveguide and tweeter front plate because there's no need for that. X-over for tweeter is quite simple - 5 components.

I'm at work at the moment and all i have in my phone are pics of my notebook monitor taken during measurements - i was excited about the results so i had to send it to my best friend: now, right now, this moment and not a moment later... Values of xover are a bit off and so is on axis frequency response, but close enough to see if i'm on the right track. Grid is 2dB. Off-axis smoothed to 3dB/o just to see what are trends when going off axis. Since it was quick and dirty (measurements done this monday at 00:00h, i get up at 06:00 and i really need my 8 hours sleep). So, blue is on axis, about 30deg is bordeaux, and with green the loudspeaker was off axis so much that i could barely see the dome. I'll do more thorough measurements when i wind the coils to the exact values and buy resistors i need. I'll upload screenshots when i get back home.

Images not available


Here are few pics of test cabinets.

Images not available

[augerpro]

deleted post

[augerpro]

More test results! After seeing the results of the elliptical above, and promise of the occluded surround, combining the two should be the bomb right? Wrong:



So what wrong? Well for one, with a throat that is smaller and .08" above the tweeter faceplate, forces a slightly different wall profile with a shallower radius. So that could be it. And in contravention of good scientific method I changed two things: I also made the throat diameter just a bit smaller than the prior design so it ended up above the edge of the dome where the voice coil attaches. I suppose I'll have to try it with the same as prior designs I posted, but really, I'm not sure following this path matters much, just going to an (golden) elliptical waveguide provided all the benefits already.

I noticed before that it seems like a lower angle throat that widens rapidly (small wall radius) seems to be the best. To test this more directly I made two waveguides that start as conical to set the throat angle, than a roundover (fillet actually) is applied to transition from throat to mouth. One throat angle was set at 80 degrees and the other at 100. Results:





So that does add some good data to support that conclusion. Somewhat along those lines I also made another version of the elliptical from the earlier post where I made the vertical wall radius smaller: 1" versus 1.5" of the original. Here are vertical and horizontal measurements, with and without phase shield:









I'll have a lot more to say once I compare that with the original. But I think we're VERY close to an end result!

All of those prior ellipticals were golden ellipses, where the vertical diameter is 61.8% of the horizontal diameter. I thought I would try another variant that wasn't as "squished" but still based on Fibonacci ratios. One ratio would be 23.6%, and subtracting that from 100 leaves 76.4%. I've seen this used in stock trading indicators, but if I understand how Fibs work this isn't a real one, but hey what the heck let's try it anyway:



Very good, though not better than the golden ellipse. Forgot the vertical responses, sorry.

Best so far:

[augerpro]

Here is a direct comparison of the first elliptical (which I'll call A), and the one above (which I'll call C). These are with the phase shield, 0 and 40 degrees:



Both are really good, seems to me splitting the difference would be just right. So I'll have that one printed. Certainly looks better than factory, especially when you consider the baffle diffraction signature is virtually gone with the waveguide:



Take another look at C of just the 0 and 60 degree responses:



Going back to the first post, I think I've met the criteria for this first design: intended to mate with a 12-15cm woofer at @2.5khz, with the difference between 0 and 60 degrees off axis at the crossover being about 3.5dB. Looking just at the on axis response, the tweeter has a natural bump around 5khz, and the waveguide makes the top octave wiggle a bit. Moving off axis the response quickly becomes smooth and straight. The phase shield looks best, but even without it these ellipticals look pretty nice. Thoughts so far? Anything else I should try? If you guys agree I'm going to move on to some 8" waveguides now that it *appears* I have some best practices.

[Zvu]

Lookin' great there. I'm quite intrigued how will the 8" waveguide measure since there are not many of those out there.

It would also be nice to see CSD plot... If anything changed after adding a waveguide.

[augerpro]

Yeah 8" will be fun, and a real test of well certain design ideas scale up, or not. I'll take a look at generating CSD from my impulse response.

I'm also going to do a version that is intended for LR2. It will have to be wider than above for the same crossover, and acoustic center will be more important. If someone wants to mock up a speaker and test this out (I don't have woodworking ability right now) I can send them a waveguide for pretty cheap.

I can also do specific tweaks to the design pretty easily if someone wants to experiment. Prices are cheap right now for the waveguide, might want to take advantage while you can.

I also want to move on to Satori tweeters. They appear to attach the faceplate the same way, so it really makes it easy. If anyone has a loaner out there, let me know!

And again, if anyone finds this useful, a donation would be a huge help. I'm already well over $500 into this, can't keep up that burn rate forever, I'm not Tesla! My paypal is whippersnapper02 at yaho, even $10 here or there helps.

[augerpro]

Here is the C elliptical with LR4 @ 2khz acoustic response, at 0 and 40 degrees, to give you an idea:



Here are some files if anyone wants to play with this a bit. If you're using Fusion 360 the .f3d file is the one you want. Otherwise the others are .stl's with and without phase shield, and with and without a circular flange for easier mounting. Probably easier for prototyping work. I'll make a router template when I get a chance, that would make routing the ellipse easy for those who don't know how to make a template directly from the part.

.f3d file: https://www.dropbox.com/s/lxbau322t6...%20v2.f3d?dl=0

Elliptical flange, with phase shield: https://www.dropbox.com/s/oo2w1id8e6...ndard.stl?dl=0

Elliptical flange, no phase shield: https://www.dropbox.com/s/q3wa1byn4f...ndard.stl?dl=0

Circle flange, with phase shield: https://www.dropbox.com/s/rv6jbsvprz...ircle.stl?dl=0

Circle flange, no phase shield: https://www.dropbox.com/s/r2avxyyzxr...ircle.stl?dl=0

[augerpro]

if anyone finds this useful, a donation would be a huge help.

Thanks Ed!

[augerpro]

Here are some pics of the conical 80 and 100 degree waveguides from a few posts back for reference. The seam you see is where the straight cone ends and the fillet begins. They are a little different because I wanted the fillet radius to be similar: