Measuring and voicing waveguides v.s. flush mount tweeters

[dsrviola]

I'm about to begin my first measurements/crossover design of a system with a waveguide. I was wondering if people who've worked with waveguide designs do anything different when taking measurements and designing the crossover. Here's the measurements of the tweeter on a flat baffle and in the waveguide: (I tried to post just the graphs but the forum wouldn't let me post attachments without registering...and I'm already registered AND logged in!) http://techtalk.parts-express.com/sh...09#post1841309 As you can see there's quite a bit of "droop" in the off axis response for the WG compared to the flat baffle. I'm thinking if I approach this like a standard design the end result is going to sound dull/rolled off because of this. If I'm used to using on axis driver measurements for crossover design, should I use off axis measurements (let's say 30-40 degrees) for a project like this and plan on pointing the speakers more or less straight ahead? (I usually measure and listen about 5-10 degrees off axis to my non-waveguide designs.)

[BobEllis]

That droop is the benefit of the waveguide IMHO. If you match the directivity of the mid/woofer to the waveguide at the crossover frequency you will minimize abrupt changes in directivity and thus power response. I'd say measure and design for your usual listening position. EQ the tweeter for flat response on the listening axis and go from there. Of course, I may be completely off, having not designed a waveguide system.

[JonMarsh]

Bob has highlighted an important factor- issues around directivity and the fall off in level with off axis, and matching this to the next driver down (be it midrange or woofer.

However, let me say, IMO, generalizing about waveguides is just about as difficult as generalizing about DIY speaker designs - there are so many different animals out there with different characteristics and tradeoffs that the old saying, "No generalization is worth a damn, including this one" may apply.

Having designed several waveguide based systems, and evaluated a few, I have some observations and opinions, but I wouldn't presume to represent them as fact or hard guidelines. That might be ET's department, but then he doesn't have any truck with waveguides... :W

A basic we can all agree on is that a waveguide will increase the sensitivity in some portion of it's range, increase the output level, and reduce the amount of work the tweeter has to do for a given output- this in turn leads to a general lowering of distortion compared with the direct radiator operation. The range of this benefit and the degree of "amplification" of output depends on the waveguide design, including it's diameter, it's depth and angle, and the type of curve (oblate spheroid being preferred by many designers). These design factors seem to lead to other trade-offs in behavior that may or may not be considered desirable. More in a minute.

Another nominal benefit/claim/goal for a well designed waveguide (Hah! watch me slip these vague qualifiers in- we don't necessarily know how to create a well designed tweeter/waveguide combo, but we usually think we recognize it when we see it!) is that it will improve the off axis dispersion in the upper frequency range (above 8 kHz). In fact, one sign of this behavior is that it may well REDUCE the on axis sensitivity because some of the energy is sent elsewhere. So, a waveguide can also redirect where energy goes, in a desirable sense, to broaden the off axis response above 10 kHz.

Side effects to consider:

• Deep, large waveguide designs will have higher distortion reduction, but narrow total beam widths
• Deep, large waveguide designs may/usually show more fall off in overall level as you go to the off axis positions- so, while on axis curve is at nominal level (call it 0 dB), 45 degree curve may be 6-10 dB down or more.
• Interfacing the tweeter with the initial throat is a bit of an art, especially for hard domes, which may require additional diffusers or phase plugs for barely acceptable response smoothness and off axis behavior.

Now, for reference, I'm going to dredge up my last waveguide effort, used in two systems, one a two way, the other a three way. Contrast the measured characteristics against the provided link example. I'm NOT saying one or the other is better, just highlighting that they are different.


This is for a Jantzen or MCM type waveguide, tailored and setup with a Scanspeak D2608/9130 tweeter.

First, the on axis response to 40 kHz, in waveguide or flush mount.



Nice lift in the response and distortion reduction exists down to 3 kHz, falling off below that. I used these with a 2500 Hz crossover, and I think that was a good choice for this waveguide/driver combo. Good for a three way. I used the same setup in a two way down to 2000 Hz crossover, LR2, with a Seas ER18RNX. Still worked fairly well, as the response at 2 kHz was down acoustically by 6 dB. Would NOT have taken it lower. BTW, this combo provides better low frequency loading than the Seas DXT, so this is partly the basis for believing the DXT is a good choice more for three ways at a typical three way crossover frequency (2500-3000 Hz).

Axis Curves: 0-10-20-30-45 degrees



The 0-30 degrees curves show very tight spread and relatively low HF droop; overall droop in the curve at 45 degrees is decent, with good off axis response to 45 degrees at 15-16 kHz, which is pretty much impossible to get with a flush mount tweeter. And the tweeter to throat implementation worked out pretty well, minimal impact on passband smoothness and good performance at 20 kHz at 20 degrees off axis. In particular, the overall level for 4-8 kHz at 45 degrees is not that much lower than for on axis; many waveguides I've seen published have considerably more drop for the 45 degree curve than this, which I think is a byproduct of the extra gain and directivity they get from other design trade-offs.

I think that will impose some additional challenges in voicing, if you're trying to match up to a direct radiator in output curves below the crossover point- with the right selection of radiating driver diameter this can be managed at the crossover point. Below that, with lowering frequency, the angle of radiation will increase, and with it the off axis level. As long as that's a smooth transition, a down tilt in the power response is adjusted to easily, especially with a good axial response characteristic out to 20-30 degrees.

Still, my most recent project doesn't use a waveguide. My desire/goal was to target wide dispersion and make the far field power response more uniform; to that end, I sought to chose a physical design which minimized the baffle width for each driver set and pushed the shift from 2pi to 4pi radiation as high as possible. So far, I'm really liking how that is working out- the far field power has more energy in the upper range; and the system sounds better in adjacent rooms, for example. But it does place additional demands on the drivers, requiring careful selection for extended frequency response and low distortion with level.


Last, regarding these thoughts and observations, and anyone else's: In God we trust, all others we verify.