Open source Waveguides for CNC & 3D printing!

Ask winbocn03 on Skype, they want 65$ per waveguide plus shipping.
 

Attachments

  • IMG_9687.JPG
    IMG_9687.JPG
    523.7 KB · Views: 1,522
  • IMG_9686.JPG
    IMG_9686.JPG
    343.1 KB · Views: 1,508
Update! I'll post results over the next few days as I progress. This post will be all SB Acoustics: SB26ADC, SB21SDC, and SB29SDAC. I also will have results for the CSS LD22, Dayton RST28 A & F, and SB19ST over the next few days.

Couple reminders, all of these have elliptical mouths. Measurement conditions are MLS with 7ms of reflection free signal, NO smoothing. The spikes you see around 24khz are artifacts of the sound card cutoff causing some crazy FFT math. This isn't my old monster baffle, so baffle effects start to show around 3khz and below. The most obvious sign is the waviness of response and ripples and dips that flip around according to angle. Anechoic response would be smoother and SPL would drop as you move off axis.

First up, version G for the SB26:
zd1kTeC.png ozyO0pC.png

I would say we're done here, but for the heck of it I designed one with an oval throat, vertically aligned. I don't have measurements yet of that.

Unless you can design a phase plug similar to a compression driver, the key to a smooth top octave is to make the throat small, the closer you can get to an inch or so the better. That is why I was excited to see SB's new 21mm tweeter with fabric dome (no phase shield needed) and copper shorting ring! Results were very satisfying:

noc1atP.jpg emRdPeM.png Ad1aN7m.png SurE12q.png

I have one last waveguide to test with SB19ST but with the above results and the fact that the SB21 has a shorting ring, makes me think spending more time on the SB19ST is unneeded.

Next is the big fabric dome SB29SADC. I reused the 8" waveguide from the TW29 tests I did. The SB29 looks a lot better, but still that big throat causes some issue. Still the directivity would probably match an 8" woofer well 1-1.5khz and you would have a smooth upper midrange and lower treble.

VEzfO1x.png
 
Last edited:
Today's update. First up the CSS LD22. When I was given the throat diameter I knew there would be top octave trouble, and this is exactly what we see. The good news is now that I have this in hand, I can tighten up the throat quite a bit, at least to 1.31" used in version G of the SB26 guide which is looking quite good now.

ruMKGlR.png


Next is the new Dayton RST28A and F. Lot's more to experiment with: using the standard faceplate the throat is huge, and I have my doubts about the phase shield being so far away from the diaphragm. Improvements in both areas in future versions.

HKaVkx3.png

yoKLCo0.png


Up next is the SB26 version G, but I made the throat slightly oval. Imagine it being about a 3/32" longer in the vertical dimension. Forgot to do no phase shield measurements.

AZAMMiv.png


Earlier in my experiments I played with phase shield shape and size. Eventually settled on a round .25" diameter phase shield. But since those waveguides were so different -all were circular mouths- I thought I'd revisit phase shields to see if the earlier results held with these current, more optimized, designs.

fmWk5lk.png


And last, the SB26 version G with no phase shield (and with and without a shim at the throat):

CLiKI4O.png

BLwuwCi.png


6" and 8" guides for SB26 and RST28 coming soon.
 
Last edited:
Thanks for the update and for posting all these measurements - very interesting!

I just realized, that some images I posted earlier are barely readable after resizing (e.g. post #80). As you are still evaluating the impact of adding a 'step' or 'shim' at the WG throat, I add one image showing simulations of a WG (V2, VarC-related) with steps (or shims) from 0mm to 5mm.


attachment.php
 

Attachments

  • WG 0-5mm step.jpg
    WG 0-5mm step.jpg
    58.3 KB · Views: 2,022
As you are still evaluating the impact of adding a 'step' or 'shim' at the WG throat, I add one image showing simulations of a WG (V2, VarC-related) with steps (or shims) from 0mm to 5mm.

Yep, I've noted when I use shims in the plots. Most of the designs include about .015" to get a good seal, IIRC. So the shims noted would add to that.
 
Last update! Moving now to the 8" waveguides. (note: 8" is the width, as these are all elliptical mouths based on the golden ellipse, the height will be .618 x width).

First up is the SB26ADC. W/ and w/o phase shield, and w/ and w/o throat shim:

DZsQXaF.png 3ZoKlmn.png 5GcKmew.png HmH6FUE.png

RST28F on 8" waveguide:
wrvJNhW.png

RST28A on 8" waveguide:
dARRxSB.png

Can't weight to redesign with the smaller throat and no faceplate...

Next are the 6" waveguides. Starting to get enough variation to identify trends accross size. SB26 first, in .75" and 1" depth:

mgE6mvV.png 9qN4IQ9.png pYeaHlR.png mVr9WnX.png
 
To get an idea of what these could look like after simple crossover/eq is implemented, I've taken the SB26 on the 8" waveguide C above and applied a HP and high shelf. Shown is on axis, and 40 degrees off-axis which I've found is a good proxy for the "early reflections" as defined by Olive @ Harman:
 

Attachments

  • zrA2oiA.png
    zrA2oiA.png
    56.5 KB · Views: 985
I want to take a moment to demonstrate how different measurement techniques impact the results. I use a 7ms gate that is FREE of reflections, and no smoothing. The gate itself actually has a smoothing effect once you see what is going on in the math. For example, at 7ms, the cutoff frequency is 143hz. This is also the first valid data point. The next valid data point is double that: 286hz. Each following data point is 143hz from the last and everything in between is interpolated. So if the valid data points are wider apart, with everything in between interpolated, you can see that effectively you lose resolution, or "smooth" the response. How would the data points be further apart? A smaller gate. For example at 3ms the first valid data point is 333hz, and all of the next data points are simply every 333hz after that.

The reason I bring this up is that I want people to understand that my plots show as close to reality as practical. Because of that, some people may say, oh look at that dip, or that peak, I don't want to use that driver, speaker, etc. I'll use that one over their that is oh so smooth. But the smooth plot is probably not realistic. If it looks like it was drawn by a child with a crayon, I have my doubts on the accuracy :) Below is a plot where I play with gate and smoothing to give you an idea of the effects. This is why knowing the measurement conditions is critical when evaluating, and especially comparing, different plots.
 

Attachments

  • 78DXZIb.png
    78DXZIb.png
    40.7 KB · Views: 966
Last edited:
When I first started this project, I didn't want to make generically sized waveguides, but ones driven by application. The first application was: crossing to a 12-15cm woofer at 2.5khz. So let's pick that apart. First, we want the directivity to be about the same at the crossover point. Looking at some manufacturers' plots of the 0 and 60 degree responses, they were typically 2.5-5dB apart at 2.5khz, with a sweet spot of 3.5dB. So this is the first criteria: 0/60 degree plots 3.5dB apart at 2.5khz. Second, I wanted to match the acoustic center of the woofer, typically about .75" - 1" deep for 12 - 15cm woofers. I picked .75" and moved on. I should mention though that now that I have some good designs, I need to verify this assumption. Some plots to illustrate:
1Quj3rV.png Nn6cro3.png A1ndXH2.png YLOE25t.png

Regarding directivity, I think this design hits its target, in fact it might even be able to be a bit smaller. Acoustic center is an assumption to be verified, but should be pretty close.

So what are some other applications? I think matching to a 18 - 22cm is an obvious one and one reason why I jumped from 5" to 8" waveguides. I still have to define directivity @ XO, and acoustic center targets. So work continues there. Any thoughts here appreciated.

Now most of the time people are wanting to use a waveguide to push lower, because you now have control of dispersion. But I can also see value in using it to push higher. Think a small 3" midrange where the designer wants it to carry most of the vocal range, so no crossover until 3-4khz. Like this Scan Disco 10F:

GqVSy9I.png

The waveguide allows us to push higher because now, even though we have some roll off of the 3" driver response, we can now match that response for the tweeter. So the off axis response will be falling, but still smooth. So we could define that application as -3.5dB at 3.5khz, and maybe .5" - .75" deep to match the acoustic center.

Anyway, that's how I'm approaching this. What typical apps would you guys like to see?
 
Last edited:
Anyway, that's how I'm approaching this. What typical apps would you guys like to see?
If I can drum up enough interest in a group project we will be considering a waveguide for the tweeter of a 3 way tower (1" tweeter, 4-5" midrange and 2 x 8" woofers). The midrange is likely to be 5.25" but more like the Discovery 15M rather than 15W because of the sensitivity. The drivers will follow from the design and so a 4" is possible but a 6.5" unlikely. The tweeter will need to be from a standard range in order to fit the budget but otherwise whatever performs best in the waveguide. Suggestions?
 
..I jumped from 5" to 8" waveguides. I still have to define directivity @ XO, and acoustic center targets. So work continues there. Any thoughts here appreciated.

Now most of the time people are wanting to use a waveguide to push lower, because you now have control of dispersion. But I can also see value in using it to push higher.
Don't forget inter-driver spacing vertically in relation to about an octave above the crossover freq. +. (..at least relative to the type of crossover).

One thing (besides a steeper filter) that you can do without much penalty is effectively "cutting into" (or designing "into") the waveguide to get the vertical midrange/midwoofer closer to the tweeter in the waveguide. While it might show a slightly "rougher" result (even done well), it's often an overall audible improvement. (..and typically the smaller the midrange, the better the result.)

I think this is an area where designing it yourself pays dividends (..as opposed to cutting into a pre-made waveguide).

Also,
-the deeper the waveguide, the greater the effect of cancellation/reduction in freq.s in the upper treble on (and near)-axis. (..it also tends to sound less realistic as well where the rest of the speaker is effectively omni.) In this respect then I wouldn't push much past 1.25" in depth.

-results look good so far! :up:
 
If I can drum up enough interest in a group project we will be considering a waveguide for the tweeter of a 3 way tower (1" tweeter, 4-5" midrange and 2 x 8" woofers). The midrange is likely to be 5.25" but more like the Discovery 15M rather than 15W because of the sensitivity. The drivers will follow from the design and so a 4" is possible but a 6.5" unlikely. The tweeter will need to be from a standard range in order to fit the budget but otherwise whatever performs best in the waveguide. Suggestions?

Well, so far the SB26ADC or SB21SDC on a 5" wg are easy choices. But I'm not sure how the sensitivity matches up as the SB26ADC is somewhat low. You'll need to figure that out. Then you'll need to define the crossover point and likely slope. Then match the directivity of the mid used at that point. That Disco 15M is super cool.