Waveguides for 3D printers and CNC's

[augerpro]

I started a little project that might interest some of you. I feel we're really getting short-changed by manufacturers like ScanSpeak and SBA when they supply waveguide tweeters to the retail companies but not us. 3D printing inspired me to think maybe there was a solution. Once I got up to speed on the CAD software (Autodesk Fusion 360 is awesome for the untrained) I've started moving forward with some designs. The goal will be to create application specific waveguides. Applications will basically define the woofer size intended, acoustic offset, low frequency cutoff needed, and directivity at the crossover that matches the woofer DI. I see three or 4 ultimate designs: one for 12-15cm woofers, one or two for 15-17cm woofers, and one for a 22cm woofer. 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.

The first application is a 5" guide intended to match a 12-15cm woofer and has a 3.5dB spread from 0 to 60 degrees at 2.5khz - about the same as the intended woofer. AC depth of these woofers are about .75" - 1". I'll be doing both a .75" and 1" deep guide to gauge the impact on DI at 2.5khz. I'll also being two versions at each depth, one where the flare is tangent to the baffle face (and so flare radius is out of my hands) and one where it isn't and allows me to enter my own radius. For the designs so far the radius is about 2.1" for the tangent version, and I specified 4" radius for the non-tangent version.

A phase shield will be in all designs, though I will test without just to see the difference. I basically copied the size the SB tweeter uses.

I'm starting with round guides but will move on to elliptical also.

I intend to make all of the resulting designs available for free to anyone. My hope is to quantify the actual differences between design parameters and end up with very high quality designs for handful of popular tweeters for the entire DIY community.

BTW if anyone has access to an industrial printer for say, resin, and can help out (I'll pay material and shipping costs of course) let me know. Retail prices for the service are still high enough that it isn't feasible to pay for 20 prototypes that will end up in the garbage afterward. In an unfortunate turn of events the person that was going to no longer can. Luckily another hobbyist will help making some in wood with a CNC (which I can easily attach the printed phase shields to) so design and prototyping will still move forward, but in the future the 3D print will probably be the most accessible so I'm hoping to source that at reasonable cost.

[Steve Manning]

Cool idea Brandon ....... do you know how the big boys fabricate the ones they offer? I would suspect some kind of injection molding more than likely, just to keep the cost down. Do you know how much it would cost to have these done on a 3D printer? I know from doing the Minerva project, your looking at ~$100/hr for CNC time. To get a good finish on something like these you could be talking a few hours of machine time, which could make that route cost prohibitive. Looking forward to what you come up with.

[augerpro]

PLA or ABS is $100 minimum. Resin is lot higher than that. But like any new technology the quality will go up and prices down over the years. I *hope* to do this project well enough with enough popular tweeters that people will still be using the guides 10 years from now and printing prices will be low enough to encourage that. As far as CNC you would really only be routing the waveguide into a baffle, it wouldn't be a standalone piece, so I think that would probably be relatively cheap and the better option for now. The phase shields are cheap to have printed in resin and the end user could just glue them to the wood waveguide as I will be doing.

Now that you mention injection molding, I wonder if there reasonably priced services out there for small quantity orders, like group buy size?

[ergo]

Very good initiative!

Transducer Lab tweeters seem especially appealing for this due to their flat front plate. But this far I don't think I've seen anyone meeting those with WG, so not know yet how well one can marry those frequency response smoothness wise.

I don't have a 3D print option, but I do have a CNC. Vectric recently released a new version of the VCarve Pro CAM where they have a new feature for "Moulding Toolpath". By the look of it it should be a much better option for waveguide milling as well compared to the too basic 3D mill options they had this far. So I might be trying this out too.

https://vimeo.com/165263229

[augerpro]

Dan N had mounted the TL to waveguide in his Echelon speaker.

Once I know what format the CNC needs (TNA is helping me with that) I can send anything you want to try.

[ergo]

I installed the Fusion 360 also. Looks like an interesting option - has the CAM section in it as well.... For now I tried to export as IGES and STEPS from F360 and import into Rhino3D. Both options work ok. From Rhino to VCarve I know the process already, so that would be faster until I can find bit time to learn the F360. I assume you can share the file via the Fusion system as well? I'll send you a PM with my Fusion Account (e-mail).

[Steve Manning]

[ATTACH=CONFIG]26327[/ATTACH]

I installed the Fusion 360 also. Looks like an interesting option - has the CAM section in it as well.... For now I tried to export as IGES and STEPS from F360 and import into Rhino3D. Both options work ok. From Rhino to VCarve I know the process already, so that would be faster until I can find bit time to learn the F360. I assume you can share the file via the Fusion system as well? I'll send you a PM with my Fusion Account (e-mail).

Hey Ergo ..... you can actually send an invitation to the files directly through Fusion 360. It's part of the concept behind the system, being able to share and collaborate stuff directly, without having to email files all over the place.

[augerpro]

I'm on vacation but I'll send the files out when I get back home.

[Bear]

It's been over a month since the last post, but I'm rooting you guys on!

[ergo]

I got the Fusion project from Brandon and was able to open and play with it a bit. Fusion also has a CAM option, so I briefly tried the workflow of creating a milling file and exporting that to WinPCNC I use to drive my machine. Seems that this can work....
But now last few weekends have gone to building a firewood shed and other jobs outside. Autumn weather is closing in. Today it was +3deg (37F) in the morning already and there been plenty of rainy days too, so these sort of jobs take a priority. I'm hoping to get around to this again though in this or coming week.

[augerpro]

Yeah that time of year to tie up loose ends for me too. Rest assured I'll be back at it soon. All the 5" guides are designed with variations in flare radius diameter, depth, and phase shield or no. I started the ellipse guides also. Found a good design process so they don't take much time at all to whip up. I didn't know where to begin with aspect ratio so I just used a Golden Ratio.

[wkhanna]

cannot make any solid promises at the moment....but....

i have access to our new printer at work.

i might be able to get a few ABS prototypes printed for you.......

Fortu 380 MC


note: coffee cup shown as a size fuducial

[augerpro]

Cool! PM me with your email address if you want to do a test run to see how well it does.

[wkhanna]

pm sent

[wkhanna]

once a few designs are finalized & vetted for performance, molding may be a cost effective method for small scale production?

Smooth-On

[wkhanna]

This line of resin based is good stuff TASK

for example: TASK 14

a printed model can be used to caste the mold.

[augerpro]

once a few designs are finalized & vetted for performance, molding may be a cost effective method for small scale production?

Smooth-On

Probably yes.

[wkhanna]

yes.......slow work does in fact take time.......

however, i have a New Years's present for Brandon.

fresh off the presses, or printer, as it were.......

the white material seen is for support of the blue ABS during the printing process & will be dissolved away in our cleaning tank.

[Steve Manning]

Cool Bill ...... how long did that take to print?

[augerpro]

Awesome! Which one is that?

So what is the support material made of? Why does it lay the ABS base layer (blue) first and then support material, then back to blue ABS? How did you remove remove the blue ABS base layer?

Sorry for all the questions, I've been doing more research as a buddy of mine was considering getting a 3D printer.

[JonMarsh]

Very cool, Bill! :T

Must confess I'm interested to hear the answers to Augerpro's questions..

[wkhanna]

the file i printed was "5x1 tangent mouth SB26 v8"

the print took just under 4 hours.

first a 20 mil piece of clear acetate is laid on the bed
then one layer of ABS is put down to provide a solid footer

the white material is PVA (Polyvinyl Alcohol), a water soluble polymer that supports the pre-cured ABS
complex prints with overhangs can only be made by printing a support structure
otherwise, the print would warp or collapse

once the print is completed, it is manually, albeit somewhat carefully, peeled off the acetate

the finished part is then put in a solution of warm, aggetated water & what is essentially sodium carbonate & hydrogen peroxide, which accelerates dissolving the PVA

[BobEllis]

Very cool, Bill.

Thinking out loud, would it make sense to print two halves of a mold and cast the waveguides? The phase shield might need to be printed separately and incorporated into the casting. My son prints parts for cosplay props and casts the second part on. He says that the cast part needs much less finishing than the printed mold plug. It seems like a thickened epoxy cast in a two part mold would be less work than cleaning up the PVA.

[augerpro]

Bob I think molds are very well the way to go after prototyping yields some high performers. Even CNCing into wood would be cheaper and easier for people, either way the phase shield will have to be a separate piece attached by the end user. I've been pondering how to do that, maybe a little dimple in the waveguide wall so people know where to glue it and it ends up level and at the right distance from the dome.

Bill thanks for the explanation! So some more simpleton questions if you don't mind? So why don't you just continue with the ABS when you lay the base layer? How thick is the PVA layer on the bottom after this? I know when I had one printed the guy also did a base (roughly 1-2mm) that he said he needed and that would have to be removed. I didn't find a way to do this without having an uneven gouged face for the tweeter. If I understand what you are doing, your method should help that? Good base layer, followed by a thin PVA layer, then the actual ABS for the part, which should result in a flat, smoothish bottom after the PVA is removed?

[Steve Manning]

Bill .... the surface looks like it has ridges, as in, not super smooth. Is that the case, or is it just a lighting issue?

[NyxOne]

Waveguides for 3D printers and CNC's

the file i printed was "5x1 tangent mouth SB26 v8"

the print took just under 4 hours.

first a 20 mil piece of clear acetate is laid on the bed
then one layer of ABS is put down to provide a solid footer

the white material is PVA (Polyvinyl Alcohol), a water soluble polymer that supports the pre-cured ABS
complex prints with overhangs can only be made by printing a support structure
otherwise, the print would warp or collapse

once the print is completed, it is manually, albeit somewhat carefully, peeled off the acetate

the finished part is then put in a solution of warm, aggetated water & what is essentially sodium carbonate & hydrogen peroxide, which accelerates dissolving the PVA

I've just machine a similar waveguide (5x0.75) on my humble CNC and it took about 10 minutes for the whole process running at 120 IPM with a precision set to 0.004 inch giving a super smooth finish. I knew 3D printing was long but 4 hours? Whoaaa!!!

I don't mean to be rude, but why aren't you doing this on a CNC instead? You could even use aluminum if you'd wanted to!

CAD & CAM was done on Fusion 360.

Chuck

[augerpro]

Mostly because at the time it seemed just as easy and cheap, and could integrate the phase shield. Now I know better and another user I was working with was going to CNC it in wood, and I was just going to have the phase shield printed in resin by a commercial outfit, then glued to the wood. I haven't heard from him lately though.

At this point, it's just prototyping, so whatever means a willing partner has is just fine. As long as the quality is good enough to not impact the response.

[NyxOne]

Mostly because at the time it seemed just as easy and cheap, and could integrate the phase shield. Now I know better and another user I was working with was going to CNC it in wood, and I was just going to have the phase shield printed in resin by a commercial outfit, then glued to the wood. I haven't heard from him lately though.

At this point, it's just prototyping, so whatever means a willing partner has is just fine. As long as the quality is good enough to not impact the response.

Does your model require two sided machining? AKA you need a special mounting plate on the back of the waveguide? I've never tried two sided machining although I plan on trying it in a few days for a waveguide I'll be milling for a pair of Scanspeak D2604/8320-02.

I don't usually offer my help because I rarely have free time but my wife is leaving me the house for a few days so I might be able to machine one of your waveguide if want to know what kind of finish you can get on a CNC. I cannot assure you that it'll work as I'm fairly new with F360 and am a super at machining.

Chuck.


Envoyé de mon iPhone en utilisant Tapatalk

[augerpro]

So the idea was that the test baffle would have a 9" (slightly oversize actually) round cutout, 1/2" deep. Send me your email and I'll send you the designs in .f3d, and look for the big mounting flange (intended for use with 3D printed wg's). All the smaller prints were going to mount to that, then that would mount to my test baffle. For CNCing in wood, you would have to glue up a stack of mdf, so the plan was to make the front layer of 1/2" mdf so it would be the right thickness, and 9" diameter. the other layers would be smaller diameter, just enough to machine, so that would leave a flange around about 1/2" wide (at least). Again, look at the mounting flange for 3D prints and you'll get the idea.

If you do make one, could you drill holes with the same dimension of the mounting flange, that way I can use that as a pilot to drill the holes in teh test baffle and everything I print or CNC after will have holes that will all work perfectly with each other.

Also, just turn off the phase shield, I'll have those done professionally in resin so they will be high quality and attach them later.

BTW don't laugh when you look at my workflow history, it's pretty inelegant, I'm working to use some better technique in the future.

[augerpro]

Chuck what machine do you have? I'm looking pretty hard at the Shapeoko 3.

[scottvalentin]

This is a good process to go through Brandon, and Chuck, I look forward with great interest to your implementation of the waveguide with the Scan Discovery tweeter! Sounds like a Scanspeak Modula XT!

Just so happens I've purchased a pair of SB26 STACs from a PE user. Might have to see about using a waveguide!

[NyxOne]

Waveguides for 3D printers and CNC's

I only did one waveguide so far for a pair of SB29 and easily got 5 dB of boost on 5 inch waveguide X 0.75 inch deep. Shallow waveguide don't screw up the FR above 10khz and allows you to have time aligned design.

I hope I'll be able to help. It would sure be nice to be able to contribute this time around.

Chuck

[wkhanna]

So why don't you just continue with the ABS when you lay the base layer?

the PVA is brittle & the bond is weaker allowing the final print to be removed from the acetate.
the ABS bonds tightly to the acetate.

How thick is the PVA layer on the bottom after this?

one layer

I know when I had one printed the guy also did a base (roughly 1-2mm) that he said he needed and that would have to be removed. I didn't find a way to do this without having an uneven gouged face for the tweeter. If I understand what you are doing, your method should help that? Good base layer, followed by a thin PVA layer, then the actual ABS for the part, which should result in a flat, smoothish bottom after the PVA is removed?

the PVA is just a removable interface between the acetate & ABS
otherwise it is used to support building up overhung layers of ABS

[wkhanna]

i do not have a CNC yet.
in about six months that will change.
we just ordered a Hass 20 x 16.

if the waveguides were to be CNC'd, i would likely make it in two pieces, then fasten it together by way of mechanical or adhesive

the current print will need smoothed on the critical surface, but this process is not difficult.
acetone can be used to 'melt' the surface and provide the required, non-disruptive finish.

the solvent tank:

[augerpro]

Need advice from the CNC guys. I'm planning on having the phase shield printed for use with wood guides made on a CNC. I'm thinking about how best to get this lined up correctly when attaching the PS to the guide. More so for an end user who just wants to glue and go. I had planned on leaving a round detent so the round arms would just drop in, but this would only be good for height, it could be rotated. So then I thought about making the arms V-shaped with matching V detent. This would force perfect alignment, but I don't think a CNC could make this shape, at least without a small V bit and the ability to rotate the mill. Any ideas?

[csmielke]

There was a guy here on HT named Joni from Finland who partnered with some other DIYers and bought a cylinder of POM plastic and made a number of pairs of wave guides which he used to build a floor standing version of Jon's Modula MT XE. I bought a pair from him that were designed for the SS D2608/9130 tweeter. Perhaps he could provide some help on how they accomplished it. The finished product was beautiful but did not include the phase shield.

[wkhanna]

Brandon,

you could put a square form just at the ends of the round cross section phase plug support.
then machine in the mating square interface on the WG.

this method would index the alignment.

[ergo]

Something like this can be done - ie coming down with the flat end milling tool and just milling a small dent into the WG. Acoustically this should still remain small enough and not have an impact.

[ergo]

I finally got around to trying to mill Brandon's model on a CNC as well.

After about a day worth of learning the CAM part of Fusion 360 and today messing about with trying to make the F360's G-code to be accepted by my machine software I did get some progress.
There are minor issues still, like small lines visible from flat end mill roughing stage. Probably need to keep bit more material before the ballnose mill makes the fine run.

All in all something like this can be milled in bit less than hour of machine time - assuming all works out fluently. First time with this F360 -> machine flow took of course more like 3/4 of a day due to various learnings and experiments etc.

For CNC process it would maybe be better to make a model where the curve of the WG continues closer to center and then the actual throat hole is cut later with flat end mill following the circle specified. Today I could not convince hte F360 to continue the fine milling path really down to the throat. Not sure if it's me or do they have some sort of a bug there. I did also change the backside compared to model as making a solid piece is much easier.

[wkhanna]

that looks V promising!
also like the simple plunge using a center cutting end mill for the phase plug mount.
the ends of the phase plug could easily be made to fit while indexing its position :T

would be nice if we could find someone with a CNC lathe......

[augerpro]

Awesome Ergo!

Perfect solution to the PS too. I think making the matching key on the PS arm might take a bit to figure out though. But looks like the right way to do it. If you mess with the idea in F360, doesn't even have to be on my waveguide design, send me the f3d so I can see your history.

BTW I'm going to email you updated files momentarily. Caught a few small mistakes and cleaned up my constraints. If you already starting milling another, not a big deal, the mistakes were pretty small.

[NyxOne]

Well, I'll have to give up for now! I spent the last two days battling between Fusion 360 and my machine and no matter what I do my parts always end up being between 0.02" and 0.03" too small. That wouldn't be a problem If it were for milling a baffle but for waveguide some of the feature need higher precision which I'm unable to provide right now.

Sorry for letting you down guys! :cry:

I'm not giving up but I won't have time to work on this until mid February.

Chuck

[Steve Manning]

Well, I'll have to give up for now! I spent the last two days battling between Fusion 360 and my machine and no matter what I do my parts always end up being between 0.02" and 0.03" too small. That wouldn't be a problem If it were for milling a baffle but for waveguide some of the feature need higher precision which I'm unable to provide right now.

Sorry for letting you down guys! :cry:

I'm not giving up but I won't have time to work on this until mid February.

Chuck

Hey Chuck, did you try giving a yell to online support. I found the Fusion crew to be very helpful.

[NyxOne]

I did and got a reply but it didn't solve my issue although it helped me get better resolution but i'm still too far off. I don't know why but it seems to work well when my gcode is generated by CamBam but is always off when using Fusion360 (they generate very different toolpath). One thing is sure, I'm not moving back to Cambam for this sort of machining, Fusion 360 is so much more powerful so I only need to find out how to fix this!

Alos, I'm not sure but I'm might have a small misalignment on my slave axis, that or Mach 3 - 2010 Screenset is not homing my slave X properly. Either way there seems to be something weird going on...

I spent 6 hours the first day for CAD/CAM setup + Two sided machining + Tool changing (had never done the last two before yesterday)
Yesterday I spent 14 hours trying to fix this issue and 4 more this morning.

Like I said, I'm not giving up, it's only postponed! I'm just exhausted trying to fix this issue and I'm running out of time for my other project (Solen's contest).

Chuck

[Steve Manning]

I did and got a reply but it didn't solve my issue although it helped me get better resolution but i'm still too far off. I don't know why but it seems to work well when my gcode is generated by CamBam but is always off when using Fusion360 (they generate very different toolpath). One thing is sure, I'm not moving back to Cambam for this sort of machining, Fusion 360 is so much more powerful so I only need to find out how to fix this!

Alos, I'm not sure but I'm might have a small misalignment on my slave axis, that or Mach 3 - 2010 Screenset is not homing my slave X properly. Either way there seems to be something weird going on...

I spent 6 hours the first day for CAD/CAM setup + Two sided machining + Tool changing (had never done the last two before yesterday)
Yesterday I spent 14 hours trying to fix this issue and 4 more this morning.

Like I said, I'm not giving up, it's only postponed! I'm just exhausted trying to fix this issue and I'm running out of time for my other project (Solen's contest).

Chuck

I hear you with the Fusion 360 stuff, now that I'm starting to get the hang of using it, it's very evident that it's a very capable package. I have not messed with the CAM stuff at all yet since I don't have a machine of my own yet.

Let us know what you find out.