Open source Waveguides for CNC & 3D printing!

No I haven't. Are the diaphragm dimensions the same as the TW29RN?

Not sure if the dimensions is exact the same. The piston area on both tweeters is 9,6 cm^2. The normal holes for mounting the tweeter (with faceplate) is quite close.

Of course the two tweeters are not the same, but maybe a waveguide for the Satori-tweeter will work okay for the SB29.
 
Last edited:
Phase Shield

Early on in the thread at PE, we discussed this, and it would be fairly easy to design the phase plug as a standalone item with cylindrical keys at the end, with a matching depression in the waveguide wall which would be easily made with the end mill of the CNC. I "hope" with some optimization the phase plug won't be necessary.
I did some simulations with 3 different phase plugs. I think that the specific design of phaseplugs/phaseshields for a waveguide is a very good tool to achieve (as good as possible) constant directivity up to the high frequency range.

Reason is that a 1''-membrane already starts to beam >6kHz. This simulation (done using AxiDriver) shows a 25mm membrane in an infinite baffle (with SB26 TSPs):

SB26 in infinite baffle.png

and the respective directivity plot (normalized to 0°):

SB26 in infinite baffle directivity norm0.png

Now, where do 'phase-shields' (often seen as small, round, transparent plastics fixed at the back of protecting grids in front of metal domes) impact the directivity of dome tweeters?

This is a simulation of Brandons eliptical waveguide 'C' with a small phase-plug V1:

Waveguide Augerpro eliptic phase shield 1 image.png

The respective SPLs from 0-90°, 15°-steps, normalized to 0°:

2018-02-03 eliptic WG phaseshield 1 SPL 0-90deg 1 .png

The same for a phase-plug with higher diameter:

Waveguide Augerpro eliptic phase shield 2 image 1.png

SPL 0°-90° (as above):

2018-02-03 eliptic WG phaseshield 2 SPL 0-90deg 1.png

And a phase plug version 3, a ring with central hole:

Waveguide Augerpro eliptic phase shield 3 image 1.png

Again the directivity 0-90°:

2018-02-03 eliptic WG phaseshield 3 0-90deg 1.png

Where does the phase plugs impact directivity? As a comparison, the directivity 0°-90° of the waveguide without phase plug vs phase plug V2:

Waveguide eliptical without vs with phase shield 2 0-90Gr horiz 1a.png

For better comparison the horizontal directivity without phase plug:

2018-02-03 eliptic WG Brendon wo phaseshield SPL 0-90deg 4.png

The effects of the phase plugs appear around 13-20 kHz, where the WG-contour has no effect (may be beside bad match of tweeter membrane and WG-throat).

The suggested phase plugs are not meant to be the best possible solution here, they should just help to demonstrate the impact of phase-plug variations in this waveguide.

Before performing more simulations of different phaseplugs, I'd suggest to start with the WG-contour instead, if you're interested.

BW, Christoph
 
Christoph, your conclusions with regards to the phase plug are valid, based on experiments with optimization of horns by Scandinavian scientists. They found the impact of a phase plug is significant in the highest octave (albeit with some transmission losses).

I'm currently trying to replicate a similar setup but this will take a while (or even a long time ;- ).
Some optimization functions are available in Comsol, based on earlier work done by the Scandinavians. They have published more advanced methods recently, but:
- these involve a lot of scripting complex math in Matlab
- and might be overkill for this project.
 
Last edited:
Before performing more simulations of different phaseplugs, I'd suggest to start with the WG-contour instead, if you're interested.

BW, Christoph

I agree.

BTW I did some comparisons of phase shield shapes back in the PETT thread, but photobucket pulled the pics :( Anyway, I did four shapes: disc, cone (like Vifa/Scan), scalloped disc, and a ribbed plug similar to the ones Eton uses on their woofers. They were all terrible except the disc and scalloped disc. The scallops offered no advantage, so I further investigated the simple disc. I tried diameters of .5", .38", and .25". The .25" offered the best balance of properties.

I was curious about the ring type you simmed, since this is pretty common. I never really understood it, why only allow the parts of the membrane furthest from each other to radiate? And your sim reinforces my intuition. (maybe with a fabric dome this could work where the center of the doming is moving out of phase with the outside, but with the geometry involved puts the radiated sound back into phase at any typical listening axis).

I should be able to get measurements of the new waveguides taken tomorrow.
 
Interesting!
I was curious about the ring type you simmed, since this is pretty common. I never really understood it, why only allow the parts of the membrane furthest from each other to radiate? And your sim reinforces my intuition. (maybe with a fabric dome this could work where the center of the doming is moving out of phase with the outside, but with the geometry involved puts the radiated sound back into phase at any typical listening axis).
I did simulations to compare the impact of a simple 'disc' vs a 'ring-type disc' in the waveguide thread here. Diameter and placement of the 'disc' and 'ring' were identical.

Here, the ring had a positive impact in that it resulted in a more constant directivity between 15 and 20 kHz.

Simulations done using AxiDriver. Disc phase plug in front of SB26 dome:

ring phase plug image.png

Directivity -90° to 90°, normalized to 0°:

disc phase plug directivity norm0.png

Note the 'widening' aroud 17kHz.

Same simulation with 'ring' phase plug:

disc phase plug image.png

Directivity, normalized to 0°:

ring phase plug directivity norm0.png

The 'widening' smears over a larger frequency range and shifts to higher frequencies, resulting in a more constant directivity.

Note that AxiDriver (and ABEC) assumes an ideal membrane without break-up at higher frequencies. Simulation of fabric domes is therefore not really possible. Simulation of metal domes is (hopefully) more realistic though.
 
Version 1 shown above is the phase shield with a diameter of 0.25'' as you suggest.
Version 2 has a diameter of 0.56''.
Version 3 has an inner diameter (hole) of 0.318'' and a ring of 0.127'', i.e. a diameter of 0.572''.
May be I should have given the diameters of the simulated phase shields....
BW, Christoph
 
Note that AxiDriver (and ABEC) assumes an ideal membrane without break-up at higher frequencies. Simulation of fabric domes is therefore not really possible. Simulation of metal domes is (hopefully) more realistic though.

That is exactly what I have been wondering.
If you want to achieve highly accurate results, a precise sim of the dome tweeter becomes very important.
For instance: does ABEC (AxiDriver) take the aperture effect into account? > probably not, since it assumes an ideal membrane.
 
Last edited:
So I'm guessing modeling the TW29RN dimple dome isn't possible with ABEC? Would be a bummer since my measurements today are pretty wacky, and the iterative process might take awhile.

BTW I hope to finish up measurements tomorrow and start posting results. I think I have the SB26 nailed on the 5" waveguide (now version F).
 
Should be possible with a optimized/custom driver script.
This is one of the many features in ABEC, without images of the formula's and electrical diagrams:

The standard model of an electro-dynamic transducer models the electrical, mechanical and acoustical parts.

The electric network parameters are the voltage U and current I. The mechanic parameters are the force F and velocity v. The acoustic parameters are the pressure p and volume-velocity V, one pair for the front and one for the rear side of the diaphragm.

The electrical part consists of the Voice Coil Impedance Zvc and the electro-mechanic motor.

The mechanical part is modeling only one rigid body motion, which is the translation in axial direction. Rotations and eigen-vibration of the diaphragm are ignored.

Vacuum Model

The vacuum allows to define certain technical parameters:
- Resonance frequency
- Mechanical quality factor
- Electrical quality factor
- Total quality factor
- Mechanical compliance.


Looking at the Diaphragm Shapes and the Case Tree, it seems only regular cone & dome shapes are supported:
 

Attachments

  • ABEC Diaphragm Case Tree_200.png
    ABEC Diaphragm Case Tree_200.png
    407 KB · Views: 1,917
  • ABEC Outer Shapes.png
    ABEC Outer Shapes.png
    76.4 KB · Views: 1,945
Last edited:
So I'm guessing modeling the TW29RN dimple dome isn't possible with ABEC?
Christoff might shed a light on this.

Well, at least I try. To my understanding, the TW29RN is basically a 'ring-dome tweeter' like e.g. the Peerless XT25TG30-04 or Vifa XT 300 K/4 or the respective ScanSpeak models, just without a phase-plug in the center.

However, the membrane if the TW29RN is still fixed in the center, making it a 'ring dome tweeter'.

A good indication is the given Sd in the spec sheet: Sd=9.6cm2

The corresponding diameter of a circle with Sd 9.6cm2 is around 35mm. From the datasheet we can roughly take these dimensions:

Satori SB29 drawing 2a.png

The diameter of the dome is 29mm, corresponding to the voice coil diameter given in the datasheet.
There is a relatively wide suspension: Dia with suspension is around 43mm, i.e. the suspension is 7mm wide.

Usually, one would assume, that roughly the half of the suspension is radiating, i.e. contributing to Sd as given in the datasheet: Radiating surface = Dome + Suspension/2 = 36mm.

If the full area would contribute to radiation of sound, Sd would be around 10.2cm2.

The Sd-Difference is 10.2cm2 - 9.6cm2 = 0.6cm2. This area corresponds to a circle of 8.7mm (i.e. are of the dome not contributing to radiation because of fixed dome center. Though the datasheet indicates a diameter of 4mm, this might be still reasonable.

Note that is a bit more of the suspension would contribute to radiation, the non-radiating dome center would be less accordingly.

So if I would try to model the TW29RN with ABEC, I'd model it as a ring-dome tweeter and construct a dome (by 3D-CAD) with the dimensions given above and import it to ABEC.

I might give it a try, if you like...

BTW I hope to finish up measurements tomorrow and start posting results. I think I have the SB26 nailed on the 5" waveguide (now version F).
I'm very interested to see the measurements and would be happy to contribute with simulations.
 
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:

RQTEH7R.png

VbQ3f6q.png

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:

hJ3cuVx.jpg

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:

sfAqAsz.jpg

The SB19ST:

W12TVOL.png

e3DHIlH.png

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!
 
Just now actually looking at the results...I'm really happy with the way the lip smoothed the response of version F, but now the plots (say 10-40 degrees) are not as straight as version C in post 1. They have more arch. I didn't want that and not exactly sure I how I got it. The only other difference was the flare radius vertically: C is 1", F is 1.1", and E (below) was 1.25", which I didn't care for and should have even worse curving of the responses if that was the cause, and it doesn't.

K2uMH4I.png
 
Dagnabbit, I had a flawed phase shield that I *thought* I had corrected before this went to the printer, but had not. Rotating one direction vs the other produces a different response above about 8khz, with one direction being much straighter and the overall response lifted >8khz. The corrected phase shield *should* be an average of the two differing directions.

Now I have to print to verify. I can do other tweaks while I'm at it, like changing that lip depth, which impacts the very top of the range near the soundcard's cutoff. The dome's resonant spike is near there and some tweaking could get it to be nulled down.

Christoph> if you get a chance, work the sim more on the SB26ADC. Of interest: a shim at the throat to move the dome back, vertical wall radius, and to a lesser extent the horizontal wall radius - though I feel that is pretty right on.
 
Last edited:
Christoff, if it's not too much of a hassle, it would be interesting to see the effect of a small area of "low curvature" (or even flat) in a (rather deep) waveguide for a hemispherical dome tweeter.
Some space of 2-4mm around the surround "to breath" > might eliminate/cancel interferences.

Smiliar to this:

Waveguide.jpg
 
Last edited: