March 2011

Building A Budget (Super?) Tweeter
Reaching the stratosphere!
Article By Jeff Poth

Difficulty Level

  As part of my Oblate Spheroid waveguides, I wound up with a system with a rolloff above 10 kHz. What is a guy to do when he needs a supertweeter to match an advanced mid/treble horn? BUILD IT! We don't mess around, here. We start with defining requirements. I'm hoping to fill in the very top octave. Would love to go higher still but that's less of an issue than just providing "Air" within the 10 kHz to 20 kHz range. We also want constant directivity, or as close to it as possible. For those unfamiliar with this term, it means that the off-axis performance is consistent relative to on-axis. The advantage is that the tonal balance stays constant when you move around. This can be achieved with pure minimalist open baffles, but more practical, typically, is horns.

Conical horns are generally CD, and some other styles are as well. In this case, we're going "other" because it seems to suit our needs the best. The 60x40 HM17-25 horn from Selenium has some impressive measurements when used with compression drivers. It sports a small size, at the cost of a high cutoff, about 2.5 kHz. This horn is an adaptation of the "Baby Buttcheek" horns from JBL, including the 2344 used in the groundbreaking 4430 monitor, and the huge 2360 theater horn. I'd previously tested the HM17-25 with JBL 2426 compression drivers and they performed exceedingly well within the 2 kHz to 12 kHz band. The JBL driver is well damped, and squeezing more high treble out of it is not advisable. Compression drivers with 1" throats have larger than 1" diaphragms, it's part of the way they're made. For extreme high treble, you want a small diaphragm. It just so happens that I had used a dome tweeter from Apex Jr in the past, and it's a pretty decent performer (see this link), and is highly efficient, and has a top octave rise, so I set about combining the two. Why would the top octave rise be desirable? Because CD horns usually need equalization of the top end to create a flat response. With the built-in rise of this tweeter, and it's intended application, I'm hoping to avoid that and keep the crossover network simple.

After cutting and trimming the horns, I did a test fit. The opening on the horn is smaller than that of the faceplate on the apex tweeter. Accordingly, a bevel is applied to the opening. There are a number of ways to do this, I used a utility knife and sanded it smooth. You can see here the full "throated" version, one with only the hacksaw operation applied, and one with the inner lip profiled.

Now, being surplus tweeters, there were some issues with the apex tweeters. Specifically, some dust particles had found their way into the ferrofluid with time (this is very common for ferrofluid tweeters). Ferrofluid is common in tweeters, and is a magnetic fluid intended to transmit heat from the voicecoil to the top plate. This greatly increases power handling, but also does a lot of damping, and thus throws away some energy in the process. It can also mask when there's a problem with the tweeter. I chose to remove it for this project, from tweeters with and without issue, prior to matching. We will be relying upon the airload of the horn to provide some mechanical damping at Fs. The other benefit of the horn is the increase in efficiency (within the coverage window). This means that the power handling requirements will be dramatically reduced, making the ferrofluid even less important. It also reduces distortion quite a bit.

After removing the ferrofluid, Fs dropped quite a bit to around 1100 Hz. This is a reasonably low Fs for a dome tweeter, though not as low as many of the more expensive models with chambers. Previously they were more in the 1500 Hz to 1800 Hz range. The high efficiency tends to mean higher Fs in tweeters, partially because the light coils and moving assemblies used to obtain efficiency mean a lighter mass-spring system. For our purposes, removing the ferrofluid to lower the Fs pushes the impedance spike further out of the passband, simplifying crossover design. Once the ferrofluid is removed, the tweeter can be reassembled, and the horn carefully glued to the faceplate of the tweeter. 

But What Does It All Mean?
The moment of truth- how does it perform? Quite predictably, as it turns out. This is a good thing, it means that the assumptions made when putting this together hold up in the real world. On-Axis we do better than +/-2dB from 1500 Hz to 20 kHz. Not too shabby! The top octave still beams, but because the rest of the band is directional, it's much less of an issue. Off axis we see fairly constant directivity from 2 kHz to 10 kHz or thereabouts. Below 2 kHz the loading of the horn falls apart and we see these frequencies as rough AND elevated relative to the directivity-controlled band. Not good! I wouldn't want to use this below 2000-2500, and that with at least a 12dB XO. That's filter function- remember, we're correcting for CD too, with a small series cap. 3.3uF is recommended for maximally flat operation, giving you about 98dB/2.83V/M sensitivity through the constant directivity passband. 

"Aren't horns really efficient?!" Yep! They sure are! But we're limited by the limited top octave power response, so we can't really do a ton more than the raw driver in that respect. So once we EQ the rest of the passband, we have a much higher impedance, so it takes a lot less current to achieve the efficiency. What was a nominally 4 Ohm driver now is a falling impedance profile (with the 3.3uF series cap), giving us approx 40 Ohms at 1 kHz, 20 at 2 kHz, 10 at 5 kHz, and falling down to the nominal 4 Ohms only in the 10 kHz to 20 kHz octave. This means a much easier load for some amps to drive in many apps (it won't present that high a load, as it will be in parallel with either a wideband mid or the 'normal' tweeter.

How Does It Sound?
I haven't been able to make this work in the context of the overall system. The OSWG has a top octave rolloff on paper, but it's really not a major issue. The additional complexity required to integrate the horn on this page just doesn't play out well. It is, however, an excellent tweeter solution for some other systems! The performance is there, it's just in need of a little TLC and matching to find a good application for it. Disappointing to be sure, but I'll hammer them into some other system at some point. I'm giving it a high iron rating accordingly- the actual work is exceedingly simple, but the filter design will take some playing. I strongly suggest that a builder measure them and spend some time with Jeff Bagby's Passive Crossover Designer software (free).