A 2-way Omni experiment

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
After reading Floyd Toole's book, I thought I'd make some back surround speakers for my setup and, for this, to try an omnidirectional design. Toole described some considerable research about audibility of resonances, specifically that high-Q resonances are relatively inaudible at least compared to low-Q resonances. Similar research pointed to general preference by most listeners to speakers that had a generally flat response with deep bass, but particularly those that had nearly the same response off-axis as on-axis. Since I also wanted the speakers to create rear channel ambience, it seemed like omnis would be the way to go.

The obvious choice for an easy to construct omni is Linkwitz' "Pluto", but what's the fun in EXACTLY duplicating someone else's design? I decided to go with something similar, but to try to have it omnidirectional over the whole spectrum (Pluto has a forward-only facing tweeter).

So this is another 'sewer pipe speaker'. I used a 7" Peerless driver mounted facing upward in a 6" diameter pipe, topped by a Peerles 1" 'full range' driver (in a short 2" diameter pipe) also facing upward but into a conic reflector. See attached photos. The idea was that, since the speaker would have internal amplifiers and an electronic crossover, I'd first find the arrangement that gave the most consistent spectral shape with mic variation in vertical orientation around the speaker. Flatness of response was not a criteria at this stage, as long as it gave enough energy for the crossover EQ to work with. And narrow response ripples or comb notches would be tolerated, if they couldn't be avoided.

I tried a lot of different reflector shapes and styles, curves, cones, balls, partially absorptive, small, large. I don't have the means (or patience) to try something like the Duevel, though that would probably be the best way to do it. The arrangement that worked best of the ones I tried was a small 3" diameter conic shape, with it's point cut off so that it could be positioned close to the tweeter's inverted radiator. That didn't give the flattest response (pretty terrible, actually) but the response it gave seemed the most consistent over vertical orientations. This was made using a sheet of black plastic (the cover of a spiral-bound notebook) cut out, wrapped, and heat-welded at its seam with a soldering iron tip. I mounted the cone on screws and standoffs that also mounted the tweeter to its pipe/enclosure. Aluminum rods were used to suspend this over the woofer. No name for these yet. Maybe "Saturns", because the reflector looks a little like a ring. Or maybe just "anonymous".

Like Linkwitz' Pluto, each speaker has a two-channel power amplifier inside with an op-amp based crossover/equalizer board. I ended up using a stereo TAS5630 based class-D amplifier with on-board SMPS power supply, made by Authlxl (who frequents the class D power amp forum here). This amp can produce 140W per channel into 4 ohm loads (both drivers are 4 ohm). 140W is probably seriously insane for a tweeter amplifier, but I wanted that kind of power for the woofer, and they came as a set, so that's the way it went.

The geometry of the arrangement makes omnidirectional characteristics horizontally around the speaker a given. Some plots are attached of the final responses at various vertical angles relative to the tweeter. The first shows raw curves, made quasi-anechoically using a 3msec window (and with non-anechoic response blended in at lower frequencies). The second plot has the same curves but with 1/3rd octave smoothing applied to show the general spectral balance through all the comb ripples. The yellow curve in each is the power average of all the curves, which should correspond to the response of room reflections from the speaker, assuming reflected response is the same as the one that radiates outward. That came out respectably flat.

I haven't had much of a chance to listen to these yet. I just got the EQ straightened out today (it looks like it still could use some tweaking around 2kHz), but can't be making much noise now because people here are napping. So I posted about it instead for now. I'll see what the pipes sound like as stereo main speakers, first, before moving them to surround duty. Assuming I don't blow the tweeters out before they get that far.

Cheers!
 

Attachments

  • Closeup 3.JPG
    Closeup 3.JPG
    25.5 KB · Views: 2,892
  • Vertical responses.PNG
    Vertical responses.PNG
    41.8 KB · Views: 1,044
  • Vertical responses (O3 Smoothed).PNG
    Vertical responses (O3 Smoothed).PNG
    23.7 KB · Views: 2,796
Well, after several hours of listening to these as mains, I'm impressed. I could certainly live with these instead of what I've been using (line source dipole planar magnetics). I'm even considering making another pair for the fronts now, though that would be difficult now as the tweeters aren't available any more (sold out from a Parts Express closeout). These things sound really good with everything I've tried so far. I'm set up with subs, though, so I can't say what their low bass would be like on their own. Pipe enclosures really do a good job of not sounding 'boxy', they seem as open as the dipoles I think. And the treble seems really excellent, which is surprising considering the raggedness of the direct response. And I don't think I ever heard a speaker that vanished so completely. It may be a while before I move these into back channel service.

Has anyone here heard Duevals or that Visaton 'Fontana' omni? What do they sound like? Is this an effect that people get tired of?

I've always been of the opinion that highly directional speakers are best, or else dipoles which at least suppress energy going out at the sides, maybe I was wrong. My mind is changing maybe - high directivity, unless so strong it can swamp out all reflections, maybe isn't the way to go after all.

Or maybe it's just because they're new and different and the 'builder's preference' effect is in play?
 
I think the sound would be more natural because it behaves more like a musical instrument.
Violins, guitars, saxophones, etc etc are all fairly omnidirectional. A speaker that is highly directional will never achieve the same sound.

I have considered building a budget version of the Visaton kit (one woofer being £70 means it's not a light decision - I have 4 £20 woofers that will do). The problem for me is that I'm in a bedroom, where things move around often. This means the acoustics are constantly changing, but altogether too congested to be able to fully utilise the idea.
 
correction -- that's a 2" (not 1" like I said) Peerless fullrange. There is a black coned version (830983) still available at Madisound and PartsExpress, so maybe I'll do another pair. I'm crossing it WAY below the freq range it was designed for, so overpowering it might not be an issue. Authlxl, I may be writing you for another pair of amp boards (maybe the Tripath version this time).

The woofer (835004 SDS) is the one that Parts Express is blowing out for $10.
 
Last edited:
I've always been of the opinion that highly directional speakers are best, or else dipoles which at least suppress energy going out at the sides, maybe I was wrong. My mind is changing maybe - high directivity, unless so strong it can swamp out all reflections, maybe isn't the way to go after all.

Or maybe it's just because they're new and different and the 'builder's preference' effect is in play?



If you have ever done any testing with real sources contemporaneously with a *radial* like you have here, then you would know that they are *more* correct in basic operation. (..they simply sound more life-like with respect to imaging and sound-staging.. provided you take some precautions with use.)

You can "get around" most of the worry of overly reflected sound by sitting closer to the speakers than they are to your listening-room walls. (..in other words move them closer to you.) On the other hand this aspect isn't a significant detriment when used as a side or rear channel.

You may even find that there is a particular distance between speakers that is more "correct" than others.

The problem with the design will be perceived dynamics and output (..there is only so much the drivers can do - even with some modest gain from the waveguide).
 
I doubt there is any gain from these waveguides since they splay energy over a wider range than it would cover otherwise. Not just the horizontal/radial pattern is uniform (not plotted), but the vertical pattern is substantially uniform, too, see the plots. So I think 'omni' is a better term in this case than 'radial', though that is correct also.
 
I doubt there is any gain from these waveguides since they splay energy over a wider range than it would cover otherwise. Not just the horizontal/radial pattern is uniform (not plotted), but the vertical pattern is substantially uniform, too, see the plots. So I think 'omni' is a better term in this case than 'radial', though that is correct also.

I think you might be surprised just how much gain the waveguide is providing in the top 2 octaves (..5 kHz up - with the higher the freq. the greater the gain). If you are really interested you can always measure it without the waveguide. ;)
 
I have never heard or even touched these speakers, but they look interesting.

I have heard the Plutos and personally favor 3-way dipoles for both front mains and rear surrounds. The Plutos gave me a "well that's a new sound" rush, but when I put on a multi-channel SACD or DVD a full range dipole both created a soundstage, while still maintaining the room's acoustic signature. "They are here" for better or worst in your untreated living room.
 
Thanks very much, Bill, for another very thought provoking contribution to the audio world. To maintain what's left of my sanity I'm going to read past your comments regarding dipole line sources because I've been working way too long on a dipole project and can't look back at this point. Nonetheless, ever since hearing some Walsh drivers I've thought that omni- or quasi-omnidirectional speakers have some very interesting properties. I'm intrigued that you've gotten such satisfying sound out of conventional drivers used a bit unconventionally. If I can ever finish the dipoles I'll have to consider omnis for my next project. Maybe they'd work well for nearfield computer-based listening.

Few
 
Um, I'm a measurements guy, own a measurements company! Trust me, this one does not provide gain. And the response had to be EQ'd with a boost centered at 11kHz (among other things).


That's unusual considering the diameter of the driver which should have a significant loss in pressure above 10 kHz off-axis. (..normally anything displacing/directing the primary axis +/- 30 degrees *should* usually add something, even if it's with in a "broad" (vertical dispersion) radial pattern.)

But if you've tested it vs. 0 deg., then sure - I'll take your word. Besides, its all academic.. you've achieved a satisfactory result. :)

You might also try looking for Dan Neubecker's (dlneubec) designs on this forum and others.. his work isn't the same, but there are some similarities. I believe el Ol also has some radial designs here as well (..and particularly those using wave guides).

Anyway, nice project!
 
After reading Floyd Toole's book, I thought I'd make some back surround speakers for my setup and, for this, to try an omnidirectional design.

very nice, perhaps You can read and join the thread: http://www.diyaudio.com/forums/multi-way/121385-loudspeakers-room-system.html

omni or quasi-omni in a listening room is one of main topics of it

BTW - can You try Beveridge placement with those pipes in Your room? It is the placement described in the first post of the thread.

best regards,
graaf
 
Last edited:
To maintain what's left of my sanity I'm going to read past your comments regarding dipole line sources because I've been working way too long on a dipole project and can't look back at this point.

Few, I don't have any bad things to say about linesource dipoles, so I'd say full speed ahead. I think I'm going to stay with my dipoles in the front channels because the room is most often set up for two listeners in recliners with a space between. I'm able to cross the dipoles' forward pattern so we each sit partially in the null of the closer channel which helps broaden the sweet spot a lot. Doing that with the omnis doesn't work, the closer one is always louder. Can't toe-in with omnis!
 
Scott G, thanks for the tips on related posts, also.

I'm pretty interested in waveguides, a friend just sent me a plot of the off-axis response he got using a cheap conic waveguide and driver, pretty impressive. What would be neat would be an axial waveguide arrangement that controlled the vertical pattern for uniform response spectrum. I don't know a lot about waveguides so I don't know if that's even possible.
 
Toed-in omnis, what a thought. Kind of like the sound of one hand clapping.

I hadn't thought about the "toe-in-ability" of the dipoles. Since my goal is to adjust my ESL panels' segmentation to minimize the variation in their response as a function of angle, and then equalize the response to flatten it (sound familiar?) I think that approach would be ideal. Thanks for the idea. I'm having flashbacks to my earliest encounters with Ted Jordan's designs way back in the day. He was a proponent of crossing the axes in front of the listener as well.

Thanks again.
Few
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.