Tweeter front plate design

[ctrl]

Have you ever wondered why the tweeter front plate design is somehow the same and yet different and if these small differences matter at all? Some tweeter front plate are just flat, some have something like tiny little wave-guides or "lenses" integrated,...

I would like to try, to show here with a few simple simulations, what certain changes of the tweeter front plate cause. For this purpose we simulate an ideal 1'' tweeter with about 700Hz resonant frequency fs.

1) Flat front plate
Let's start with the tweeter dome on an infinite baffle. So the tweeter front plate is simply a flat plate.
First the sketch (lateral cutaway drawing of the tweeter dome), then the frequency responses from 0° to 80° (in 10° steps) on infinite baffle (IB) of the dome on a flat front plate and then those normalized to on-axis frequency response:



2) Front plate with 3mm step
Next we introduce a 3mm high step (I use the term "step" because it looks like a step in the lateral cutaway drawing of the tweeter dome), one could also say the tweeter has 3mm off-set/recess to the front plate. It is a tiny wave-guide which on-axis contributes to an increase of the SPL of the high frequencies and at the same time makes the radiation less wide.



3) Front plate with 2mm + 1mm steps
Instead of just a big 3mm step, we round the corner a bit. The diameter of the tiny WG remains unchanged.
The radiation becomes minimally narrower, but more uniform above 15kHz.

Something like this is what most tweeters have.


4) Front plate with 2mm + 1mm + 2mm steps
Let's add another step, creating a small wave-guide with 40mm diameter. In the 2-9kHz the on-axis SPL has increased. The radiation becomes a bit narrower.



5) Front plate with 2mm + 1mm + 2mm + 3mm steps
The next step with a height of 3mm and a diameter of 70mm provides a wave-guide that most would recognize as such.


If we now compare the normalized frequency responses of the tweeter with flat front plate with our 70mm WG, for example, at 80° radiation angle the frequency responses at 4kHz are offset by 2.5dB SPL.

The less wide dispersion in the frequency range 2-4kHz usually makes the transition to the woofer more even. However, there are other effects such as edge diffraction that are not discussed here.

6) Front plate with 2mm + 1mm + 2mm + 3mm steps and reflector
If you now want to straighten the on-axis frequency response up to almost 20kHz, you can add a reflector at the expense of radiation - this would look like a washer when viewed from above.


If we now compare the frequency responses (deg0-80 in 10° steps) of the tweeter with flat front plate and the tweeter with the 70mm wave-guide with just 8mm height and reflector on IB, the differences are dramatic (same color, same angle):




To see the impact of each change in direct comparison, here is the FR as a series:

1) Flat front plate
2) Front plate with 3mm step
3) Front plate with 2mm + 1mm steps
4) Front plate with 2mm + 1mm + 2mm steps
5) Front plate with 2mm + 1mm + 2mm + 3mm steps
6) Front plate with 2mm + 1mm + 2mm + 3mm steps and reflector


7) Front plate with "small lip" around the tweeter dome
See post#12 for more details.

8) Tweeter flush mounted but with gap around face plate?
See post#19 for more details.

 

[ctrl]

Hope it could be shed some light on why manufacturers use different front plate designs, even though the size of the tweeter dome is identical.
By the design of the front plate, the sound character of a tweeter can be influenced significantly - see below. Every little edge, every protruding screw has an impact.

Here are a few summary comparisons of the effects.

First, let's look at the effect on the on-axis frequency response:

Next, we compare the 60° frequency response when normalized to on-axis frequency response to see how the radiation changes (using the 60° FR as an example):

What happens if the tweeter is not flush mounted with the baffle?
The typical tweeter has a front plate diameter of 104mm and front plate thickness of 3-5mm.

Let's compare a simple tweeter with two small steps, one flush mounted and one not - plate thickness only 3mm:


Small difference, big impact, just like when you run your hand over the rotating flush mounted saw blade or whether it protrudes 3mm

 

[ctrl]

Updated post#2. If you prefer more exotic front plate designs models, see the simulations made for Focal Solo6 ST6 monitor - its free

 

[René - Acculution.com]

Very nice. I can run a shape optimisation on a similar setup to find the best curvature for some given targets.

 

[ctrl]

Very nice. I can run a shape optimisation on a similar setup to find the best curvature for some given targets.

That sounds very exciting. So you would try to find a "short waveguide" for, say a 1'' tweeter, which corresponds to a given radiation pattern with shape optimization? What parameters would you need for this?

 

[617]

This is amazing. The focal study is especially interesting.

Curious what a simulation of the wide dished SS revelator flange would look like.

 

[ctrl]

Curious what a simulation of the wide dished SS revelator flange would look like.

If you mean this SS tweeter, don't expect too much.

D2905/990000 – Scan-Speak A/S

www.scan-speak.dk

Have used the tweeter in the process of the Directiva r2 design simulations and the results were not satisfactory in terms of even radiation.

For a uniform radiation the tweeter should radiate narrowly in the range of the typical edge diffraction (to counteract the radiation widening by edge diffraction), which is partly fulfilled in the range 2-4kHz, but above that the radiation widens a bit and then becomes very narrow.
Manufacturer measurement:


Directiva r2 simulation with SS D2905/990000


Simulation result up to 10kHz first FR deg0-180, then normalized to on-axis FR and normalized contour plot:


The results are not very bad, but not SOTA either. Because it is a treated textile dome (very early damped breakup versus ideal tweeter in simulation), the real radiation might look a bit better though.
Anyway the radiation of the Seas DXT tweeter is more uniform, therefore it was also used in the r2.

Scan Speak should have hired @René - Acculution.com to shape optimize the heck out of it

 

[Penelinfi]

What about these kinds of "waveguides"?

 

[mv8]

Have used the tweeter in the process of the Directiva r2 design simulations and the results were not satisfactory in terms of even radiation.

I own a Seas DXT as well as a Scan Speak d2905-990000 and in my opinion the DXT is a much worse tweeter than the SS. The DXT has a much higher level of distortion and a narrower usable frequency response. These speakers sound completely different. DXT with a proper front end in terms of diffraction performs very well in measurements, but its sound will still be worse than SS.

I did measurements of the SS in a wide baffle, with a large radius rounded edge, and they came out very well. The SS narrows the radiation in at around 10 kHz, but I don't consider this a big problem.

Personally, I would choose an SS in a non-optimal enclosure for diffraction than a DXT in an excellent enclosure for diffraction without a second thought.

 

[neRok]

What happens if the tweeter is not flush mounted with the baffle?
The typical tweeter has a front plate diameter of 104mm and front plate thickness of 3-5mm.

Let's compare a simple tweeter with two small steps, one flush mounted and one not - plate thickness only 3mm:

What happens if the tweeter is flush mounted, but the recess it sits in isn't perfect. Eg there is a 1mm gap all around? Or the recess has a chamfer? If the gap was a problem, would filling it in with silicone or something be a solution?

 

[René - Acculution.com]

That sounds very exciting. So you would try to find a "short waveguide" for, say a 1'' tweeter, which corresponds to a given radiation pattern with shape optimization? What parameters would you need for this?

Really just some starting geometry. I would have to decide whether or not the depth the tweeter sits at compared to the baffle should vary also, but the targets are actually the most difficult part as they are easy to express in words but not so mathematically. I will keep this in mind, currently to most going on.

 

[ctrl]

Front plate with "small lip" around the tweeter dome

What about these kinds of "waveguides"?

Nerdy question
Let's take a look at what the manufacturer had in mind when designing the tweeter with a front plate like a lip augmentation from a plastic surgeon.


Let's start with the third tweeter front plate design from the initial post for comparison. Again a 1'' dome tweeter with fs=700Hz.
The parameters for the simulation (resolution, distribution of the simulation points,...) I tried to choose identically to the initial post. But I didn't write them down, so small deviations can occur.

1) Front plate with 2mm + 1mm steps

First the sketch (lateral cutaway drawing of the tweeter dome with surround and front plate),
then the frequency responses from 0° to 90° (in 10° steps) on infinite baffle (IB)
and then those normalized to on-axis frequency response
The same applies to all other simulations:

As we already know, this small measure raises the on-axis FR significantly above 3kHz and leads to a flat response.

2) Front plate with 3mm + 1mm steps
In the next step we increase the first step by 1mm to 3mm.

This results in a slight on-axis hump between 4-10kHz and a slightly narrowed radiation above 5kHz.

3) Front plate with 3mm + 1mm - 1mm steps
Now we do our "lip surgeon" and form a small lip around the dome.

Wow, this again results in a flat on-axis FR and much wider radiation above 2-3kHz - even wider in radiation than the front plate simulation 1).
The only downer is the resonance at 15kHz. The simulation often exaggerates the consequences a bit with pointy/tapered shapes - nevertheless, it's not pretty.

4) Front plate with 3mm + 1mm + 1mm - 1mm steps
Our "lip surgery" went wrong in 3) somehow and cause a resonance, it seems no one likes pointed lips. So we make our lip around the tweeter a little rounder:

Problem solved. Now we have all the advantages of 3) without the resonance at 15kHz.

Now one could get the idea and ask, what advantages did all the effort bring us, if version 1) is already almost perfect?
To answer this, we compare deg0 and deg60 FR from both versions 1) and 4) - normal FR and normalized FR:

Our "lip surgery" gave us a little more SPL above 5kHz on axis and a little wider radiation in the 3-12kHz range.

 

[ctrl]

I did measurements of the SS in a wide baffle, with a large radius rounded edge, and they came out very well. The SS narrows the radiation in at around 10 kHz, but I don't consider this a big problem.

Personally, I would choose an SS in a non-optimal enclosure for diffraction than a DXT in an excellent enclosure for diffraction without a second thought.

Because of the low harmonic distortion, high power handling and possible low crossover frequency, the D2905/990000 was considered.

Since the goal with the Directivia r2 is to achieve SOTA radiation at a low overall price, preference was given to the Seas DXT over the D2905/990000, which is 2.5 times as expensive, but with slightly inferior radiation. The directivity of the Seas DXT front plate ("short WG", Seas marketing calls it "DXT lens") is more uniform.

Directiva r2 tweeter full hor and ver measured responses, CTA-2034A and hor normalized contour plot:


The simulations for the D2905/990000 tweeter have shown an unsteady directivity above 5-6kHz, as your measurements also show.
Simulation (only to 10kHz) Directiva r2 with D2905/990000 full hor and ver responses, CTA-2034A and hor normalized contour plot:


No problem if you came to a different conclusion

 

[PKAudio]

I own a Seas DXT as well as a Scan Speak d2905-990000 and in my opinion the DXT is a much worse tweeter than the SS. The DXT has a much higher level of distortion and a narrower usable frequency response. These speakers sound completely different. DXT with a proper front end in terms of diffraction performs very well in measurements, but its sound will still be worse than SS.

I did measurements of the SS in a wide baffle, with a large radius rounded edge, and they came out very well. The SS narrows the radiation in at around 10 kHz, but I don't consider this a big problem.

Personally, I would choose an SS in a non-optimal enclosure for diffraction than a DXT in an excellent enclosure for diffraction without a second thought.

Exactly my experience from listening and working with a lot of tweeters (really many, from cheap tweeters to Bliesma Be, Mundorf AMT). DXT measures great in optimized faceted baffle and the reasons it was chosen for Directiva are obvious.

 

[PKAudio]

CTRL, very nice study, thank you.

 

[holdingpants01]

extremely interesting thread with zero use for me and I love it, thank you

 

[mv8]

Some time ago I built small monitors, in which I paid special attention to offaxis characteristics. The tweeter is a cheap Dayton from WG.
In my opinion, it looks very good, but unfortunately it in no way equals the sound of Scan speak in an ordinary shoebox-shaped enclosure.

I pay very close attention to the offaxis characteristics in every project I do, but I don't think their absolute accuracy makes a very big difference in sound. I still think that the most important thing in a design is to use high-quality speakers, because they have the biggest impact on the sound.

If someone is looking for exactly the sound that Scan Speak offers, they won't find it in the DXT in any enclosure, even an ideal one, unfortunately

 

[617]

If you mean this SS tweeter, don't expect too much.

D2905/990000 – Scan-Speak A/S

www.scan-speak.dk

Have used the tweeter in the process of the Directiva r2 design simulations and the results were not satisfactory in terms of even radiation.

For a uniform radiation the tweeter should radiate narrowly in the range of the typical edge diffraction (to counteract the radiation widening by edge diffraction), which is partly fulfilled in the range 2-4kHz, but above that the radiation widens a bit and then becomes very narrow.
Manufacturer measurement:
View attachment 304960

Directiva r2 simulation with SS D2905/990000
View attachment 304965 View attachment 304966

Simulation result up to 10kHz first FR deg0-180, then normalized to on-axis FR and normalized contour plot:
View attachment 304968 View attachment 304969 View attachment 304970

The results are not very bad, but not SOTA either. Because it is a treated textile dome (very early damped breakup versus ideal tweeter in simulation), the real radiation might look a bit better though.
Anyway the radiation of the Seas DXT tweeter is more uniform, therefore it was also used in the r2.

Scan Speak should have hired @René - Acculution.com to shape optimize the heck out of it

Rene was probably 12 years old when that tweeter was designed!
In general I agree, it's curious how few tweeters with decent sized waveguides are available. Your study shows that even a moderate waveguide in a ~150mm faceplate can make a big impact in a conventional design.

 

[ctrl]

Tweeter flush mounted but with gap around face plate?

What happens if the tweeter is flush mounted, but the recess it sits in isn't perfect. Eg there is a 1mm gap all around? Or the recess has a chamfer? If the gap was a problem, would filling it in with silicone or something be a solution?

If gaps are present and not too narrow, I always stuff them with black foam or black sealing tape (or black cardboard) - this allows the tweeter to be removed again without any problems.

What happens if the tweeter is flush mounted or mounted with a 1mm or 2mm gap?

As always first the sketch of all three cases:
flush mounted no gap,
a gap of 1mm and
a gap of 2mm around the tweeter.

then the 0°-90° FR (10° steps)

and last but not least the comparison of the 0°, 30° and 60° FR of the tweeter with 1mm, 2 mm and without gap


The simulation software usually depicts the effects a little more dramatically than they are in reality (it is rare that a gap is exactly 1mm wide all around, the corners are perfectly right-angled all around,...).
Therefore a gap of 1mm or less is IMO still acceptable (perfectionists also fill 1mm gaps ), but with 2mm and more you should probably fill the gap if possible (this could already have minimal tonal effects).

Update: These are of course very academic considerations, the easiest way is to seal the gaps with easily removable tape, take an on-axis measurement and then compare it to the one without tape.

 

[badspeakerdesigner]

Some time ago I built small monitors, in which I paid special attention to offaxis characteristics. The tweeter is a cheap Dayton from WG.
In my opinion, it looks very good, but unfortunately it in no way equals the sound of Scan speak in an ordinary shoebox-shaped enclosure.

I pay very close attention to the offaxis characteristics in every project I do, but I don't think their absolute accuracy makes a very big difference in sound. I still think that the most important thing in a design is to use high-quality speakers, because they have the biggest impact on the sound.

If someone is looking for exactly the sound that Scan Speak offers, they won't find it in the DXT in any enclosure, even an ideal one, unfortunately

Well now, I was about to build something just like this, almost a cheaper DXT mon if you will, but it's good to hear some feedback on the idea. Guess I will pass.

I'd be curious to see what sims say about a pretty deep loaded tweeter like the Peerless D27. That things dome is below the faceplate and the sides are basically vertical walls with a very small round over.