3D printed waveguides

[augerpro]

I keep eyeballing a Festool track saw so maybe I will build a small-ish baffle out of some 1/2" ply, something that can be atttached to my lazy susan measurement stand and put away behind a couch.

[augerpro]

I keep eyeballing a Festool track saw so maybe I will build a small-ish baffle out of some 1/2" ply, something that can be atttached to my lazy susan measurement stand and put away behind a couch.

[Face]

Festool's MFT tables also fold up nicely with the rail attached.

[TN Allen]

How large are you thinking for an effective baffle, could it be round?

[TN Allen]

Really interested to see what comes of this.
I think I have the distinction of the largest 3D printed DIY horns to date. Not that they are overly large, 270mm diameter mouth. There are mounted to an SB65 on a spiral enclosure that I also printed. I would look forward to trying to print whatever you all come up with.
I might also be able to help with design depending upon what is desired and finding some time...

How large is your printer, and which CAD program are you using?

[augerpro]

How large are you thinking for an effective baffle, could it be round?

@ 2'x3' but the final shape would be optimized for baffle diffraction. Why round? That would be the worst shape for diffraction.

[DanP]

Going back to our discussion earlier about flare profiles being almost locked in, I have attached a couple diagrams to make it easier to visualize. I the first pic, point A would be determined by the mouth diameter, and point B would be determined by the tweeter diameter and WG depth. If we now assume we want a curve that starts tangent to the baffle and is as consistent as possible, there is only one circle that fits these requirements as shown in the second pic. As stated before, you could deviate from tangent at the baffle and/or choose an inconsistent curve to purposely alter the shape of the flare, but these liberties would only deviate from optimum.

Dan

[TN Allen]

DanP,

Agreed, that is the same result I get creating an arc tangent to a line on the baffle plane at the lines termination at the mouth circle, that line would intersect the center of that circle, we're it not trimmed at the cricumference. The other end of the arc intersects the throat/tweeter circle at a point where the line would intersect the throat circle looking at both circles in a single plane, or the top view.

Rather than limiting the cross section to an arc, I have wanted to try a Tractrix curve cross section, but have not to date. I have milled a spiral cross section in the past, but have never tested it's efficacy.

If I have understood what you wrote regarding the relevance of the cross section curve to guide efficacy, I think you may well be right that it may not be as critical as one might think. I have suspected, listening to many of the guides I've milled that the contour isn't really all of that critical, although, I do attempt to make as smooth a transition as possible from the tweeter surround to the guide throat. On the other hand though, I know many guides are used over original face plates successfully, or at least considered a success based upon test results. I should add that all of the guides I've done and listened to lately combine midrange and tweeters, and are consequently, complex shapes rather than simple single driver contours. However, previous examples of single driver shapes have sounded quite good.


I wonder at times though how much testing is relied upon, rather than testing and listening critically with guides in place in a complete system.

[TN Allen]

@ 2'x3' but the final shape would be optimized for baffle diffraction. Why round? That would be the worst shape for diffraction.

A circular shape could be made in multiple pieces that can be help securly together with a piece of 1/8" Dacron line in a truckers hitch.

Could you explain why a cicle would be the worst shape. It seems to me that testing for the effect of an oval or round guide and driver in a round baffle would provide the most uniform effect. Of course, this would require the driver be tested in a similar flat baffle for comparison.

[augerpro]

Well, waveguide or regular driver, a round baffle is so terrible for diffraction that *I* would have a hard time putting any faith in my ability to mentally "subtract" its influence and see the real response of the DUT. With an optimized rectangular I should be able to get respectably smooth in the passband (also the area the waveguide has directivity control, and so further lessening diffraction) and the largest effect of diffraction should be below usable passband.

[augerpro]

Dan that is what I suspected the case would be. Pick 3 parameters and 4th is locked. Having said that, would it be too much trouble to design another for comparison where we relax the mouth termination just a bit and go with a larger radius? Doesn't have to be a huge change, I've found throat effects to change considerably with what seem to be subtle changes.

I know you said this would deviate from optimum in your experience, but here are my thoughts on this project: I believe this project could result in excellent and easy to build waveguides that the community will benefit from for years. But for that to happen we have to be thorough in *proving* what works best (objectively) by trying variations on both sides of what we think is optimum. This results in a anyone being able to see those differences side by side, measured under the same conditions, all in one thread. Given how trivial the work and money doing this will cost upfront versus the end payoff, it just seems a no-brainer to do this way, even if some of the work is redundant to those who have explored this before. I'm personally very excited on where this project could lead!

[augerpro]

DanP,


Rather than limiting the cross section to an arc, I have wanted to try a Tractrix curve cross section, but have not to date. I have milled a spiral cross section in the past, but have never tested it's efficacy.

That brings us back to something I mentioned before. From my understanding, the reason there has been so much work and speculation on horn flares over the decades is that they were designed to use compression drivers, where the assumption is that the CD produces a plane wave which must be transformed into a spherical wave. Dome tweeters definitely don't produce a plane wave, but consensus seems to suggest they produce a spherical already. Which means a straight cone is best as regards low HOMs (and ignoring mouth to baffle termination effects for the moment) and the problem is much simpler. So, it seem we could solve this theoretically assuming we already have a driver producing a spherical wave, OR if no one here has that expertise, do it the old fashioned way: try it and see! So if you build whatever flare, I have no problem testing it.

[DanP]

...would it be too much trouble to design another for comparison where we relax the mouth termination just a bit and go with a larger radius? Doesn't have to be a huge change, I've found throat effects to change considerably with what seem to be subtle changes.

I know you said this would deviate from optimum in your experience, but here are my thoughts on this project: I believe this project could result in excellent and easy to build waveguides...

This would not be too much trouble and if you think relaxing the mouth termination to go with a larger radius would make an easier to build waveguide, then I agree that that is a very worthwhile endeavor. I'm having trouble imagining how this minor change would make it easier to produce though.

At any rate, I have plans to build the 5" waveguide we discussed earlier, but have been busy at work and have a vacation coming up soon, so it won't be for a few weeks.

Dan

[DanP]

I wonder at times though how much testing is relied upon, rather than testing and listening critically with guides in place in a complete system.

I've never heard a speaker that sounded good that had a poor on axis frequency response. When an on axis FR curve has many peaks and dips, that means something is wrong. When an FR curve is flat, it means that on that axis, at that distance, everything you've accounted for is right. It doesn't necessarily say anything about other axes, distances, or unaccouted for variables. So with this in mind, I think listening is a very valuable final step to see if the things you didn't account for (because of lack of time, knowledge, equipment, desire, etc.) line up well or not. I don't personally spend the time on this step until I have a flat FR graph though. If the graph looks bad, it will sound bad. If the graph looks good, I have a fighting chance that it will sound good and begin my listening tests to verify. So I think testing these WGs is a great way to weed out duds. I am sure we will not throw away any gems that sound great despite their horrible response graphs.

Dan

[augerpro]

This would not be too much trouble and if you think relaxing the mouth termination to go with a larger radius would make an easier to build waveguide, then I agree that that is a very worthwhile endeavor. I'm having trouble imagining how this minor change would make it easier to produce though.

Dan

Why would it be any more difficult to print? If you mean in wood, that may be, you and TNA would know better than I.