More ScanSpeak tweeters with waveguide
Copyright 2012 Troels Gravesen

The article here is not meant to be an exhaustive study on waveguides for ScanSpeak tweeters, rather some practical guidelines for using the waveguide developed for Audax TW034 as it needs an adaptor and modification to fit the ScanSpeak tweeters as can be seen from drawings and images below.
If you want to learn more about waveguides, there's an excellent article here. Generally waveguides allow better coupling of tweeter diaphragm to the surrounding air and they reduce the impact of baffle edge diffraction. When properly constructed they may enhance dispersion in upper registers and reduce overall distortion.
More interesting reading here.

The ScanSpeak drivers seen below features the same plastic mold holding the diaphragm, thus can all be used with same waveguide when removing the faceplate. "Can be used" doesn't imply all tweeters are equally well suited for such arrangement. The trouble zone is the 10-20 kHz range, where we may experience a rather ragged response from some of the tweeters. Generally high-frequency is trouble! Wavelengths are short and all sorts of things can happen to the frequency response profiles when adding different faceplates or waveguides. Based on measurements the ring-radiator looks well suited for this application, but despite a somewhat uneven response from the other tweeters, I wouldn't exclude any for waveguide application. I would have liked to include the R2904/700000 tweeter also, but I don't have a pair for the time being.

Producing domes having a consistent response in the upper octave is not easy. Coating is critical and I have experienced domes having different dispersion characteristics due to coatings or damping rings being unevenly applied to the thin fabric dome. Soft domes do not operate pistonically, thus various parts of the dome will load the waveguide differently and not even the beryllium dome can be considered operating pistonically as it features a wide suspension made from fabric and I'm sure part of the fabric surround is adding to the radiating area as well as the dome itself. I haven't yet tried inverted domes (Accuton) with waveguides but will in the future - if I can remove the grille without damaging the tweeter.

I've now used the 7100 + waveguide extensively in the DTQWT construction and it's going to stay, offering enhanced resolution to the upper mid and treble range. More details and better transparency - and it can be used without changes to the crossover, which is quite a coincidence.
There will not be a commercially available one-piece waveguide for these ScanSpeak tweeters. If you want to explore what a waveguide does you have to make it yourself. Sorry, but I would never sell 100 pieces of a specialty product like this (minimum order).
Post honeymoon (3 months): I inserted the TW034 in the DTQWTs again - and kind of felt being back to where I initially wanted to be. Maybe it's my several years with this dome that has become my reference standard. The TW034 makes a more coherent soundstage with the 8008 midbass, where it seems that whenever we optimise on certain parametres it's at the expense of others. At the end of the day: A matter of taste.

Click images to view large (actual 8320 here is Peerless XT25TG and 8300 is Vifa DX25TG - same drivers with the new ScanSpeak names)

ScanSpeak tweeters, click to download data file: D3004/664000, D2904/710003, D3004/660000, R2604/832000, D2604/830000

Tweeters without face plates: From left: 6640, 7100, 6600, 8320, 8300.

Waveguide adaptor dimensions

Making the waveguide adaptor the the TW034 waveguide.

This adaptor could ideally be made on a lathe and from plastic or aluminum, but it can also be made from a piece of HDF. Yes, high-density fiberboard. MDF is too soft and doesn't work. Thanks to Jesper who offered me a slant of HDF from some kitchen interior, hence the yellow colour. This piece of HDF is 15 mm thick and we really only need 10 mm but I didn't want to buy a whole board for two small ring adaptors.
Start by fastening the HDF to a board of MDF. I used a nail air gun as can be seen below. It's important to fasten the routing center block rock-solid, hence a couple of nails close to the center hole.  

Left: Start by routing two holes of 38.3 mm (+/- 0.1 mm) diameter. Use an 8 mm router bit. Right: Due to the 15 mm HDF I started routing to a depth of 5 mm and to a diameter of some 80 mm (a little more than the final diameter of the adaptor). Next I used a V-groove router bit to make the conical shaped ring fitting into the tweeter diaphragm mounting plate. Remove residual material around ring to make a final panel thickness of 7 mm (check drawing above).

Now we need to chamfer the hole from the other side to fit the modified waveguide. Use a rounding over bit (I think the name is) with a radius of 12 mm and lower it slowly to make an outer diameter of 43 mm. This will provide a curvature suitable for the waveguide. Mount the board again upside down and finish the adaptor to a final outer diameter of 66.4 mm (not critical). Sounds easy, doesn't it? Well, may take a few test samples before perfect.

Check out adaptor on tweeter and press firmly against tweeter to compress gasket and see the dome doesn't move. If so, it should be OK. Right image: As can be seen we need to trim the waveguide and remove some 2 mm material from the rear side.

Making solid supports is important before routing the waveguide.

Left: Go 2 mm deep as shown on drawing. Right: We need a flange for mounting the tweeter. Goes like this:

For mounting the tweeter I used three M4 screws and made treads directly into the POM plastic waveguide.

The finished tweeter/waveguide


Measurements were done on a 55 x 75 cm cm baffle with tweeter placed 30 cm from top. Measuring distance 0.5 meter at 0, 10, 20 and 30 deg. Input normalised for 2.8V/1m. 1/12 octave smoothing applied to all graphs.


Left: 6600 as-is, flush-mounted. Right: 6600 with waveguide. 
I'm a bit puzzled by the suck-out at ~650 Hz and as this goes for both the 6600 and 6640 tweeters having the same magnet system, I think I have a leak somewhere and will come back with new measurements when I find out what is going on. However, it doesn't impact performance above 1 kHz. 

6600 comments: On-axis the 6600 dispay a dip around 13-15 kHz fairly consistent with measuring angle. Dispersion with waveguide remarkably uniform over a 60 deg listening angle. Despite the high frequency dip I would not exclude the tweeter as a candidate for an application including waveguide. Suitable for 90-92 dB systems.

D3004/664000 Beryllium

6640 comments: Absolutely a candidate for waveguide and we can even see an improved dispersion in the 10-20 kHz range compared to the "naked" tweeter.
The 6640 was tried out with the same waveguide by Scott/US here. I've tried the felt ring and get identical measurements.


7100 comments: Already in use with the waveguide - and I like what I hear! Note an unusually  even power response in upper octave, almost the same SPL over +/- 30 deg.

R2604/832000 - XT25TG-30-04

8320 comments: On paper the driver having the best response in upper octave, but dispersion really doesn't differ from the naked tweeter. Suitable for 90-92 dB systems. The double magnet 8330 may be used even higher. It seems ring-radiators are well suited for waveguides, possibly due to an almost flat diaphragm loading the waveguide more evenly.

D2604/830000 - DX25TG-05-04

8300 comments: No objection to this tweeter as well. The 8330 (double magnet) may be interesting also and suitable for high-efficiency systems as well.