Dipole phase question

That's a really good question! Definitely it makes sense that the two mirror each other. It's hard to believe Paradigm doesn't recognize that. Otherwise it seems like that would be like having one of your main speakers reverse phase. As for which one, plus or minus faces the mains is more difficult to answer. Especially considering this is an omni speaker from 300 Hz down, it may not be that critical. You would probably have to just experiment and see what seems best. Maybe put in switches so you can switch them on the fly during listening test. The thing is that wavelengths are relatively short at 300 Hz up compared to distances between speakers. No matter which way you put them, the actual relative phase to the mains is going to vary all over the place with frequency and listening location. So my guess is that it doesn't make much difference which way you place them but definitely I agree with making the two mirror each other. I would have rewired them too!

Monte

This has been discussed quite a bit here in the past. There was a long thread regarding it. Do a search, see if you can find it.

Flipping the speaker wire at the terminals would have been all you needed to do instead of rewiring the speaker itself. Or maybe that is what you meant?

Not necessarily. Depends upon how the drivers in the speaker are all wired relative to one another. With some dipoles, only the tweeters are wired out of phase. With others, only the woofers. And with others, both drivers, in which case it would work. If it would work, though, as the OP asks, which one would you wire in reverse?

As has already been mentioned, this was discussed at length in the past. I always heard that the drivers facing the front speakers should be in phase with the fronts. But there were others that argued the opposite. I figured that it wouldn't be a problem looking up the right answer, but I was surprised that I couldn't find any real information. I even tried to look up the original patent (was it Tomlinson Holman?) but that didn't pan out either. Anyway, another thing to do is see what the polarity is on other brands that make mirror image dipoles. I believe that Triads are consistent with what I said.

Anyway, I'd argue that you'd want stable imaging between the fronts and surrounds and so that indicates the facing drivers be in phase.

Ed

I don't buy that for a second. Paradigm used to make mirror image pairs. I believe that it's a cost cutting measure. They can sell them individually and not have 2 sets of inventory.

Ed

Well you'd never expect anyone to admit that if it were true.

I'll admit that probably the difference between mirror image pairs and non-image pairs is very minimal. Further, the speaker industry is very competitive and the industry probably is not very lucrative these days. So I can understand that it's a good business decision to make non-mirror image dipoles.

But I'm a purist and it bothers me that they now do this.

Ed

I have the ADP 590's. I have sat on the left side of my couch and the right side of my couch and there is no difference in my opinion.

I am a big fan of dipoles, especially in a room like mine where the back wall
is to close to my couch to get a good monopole speaker.

Some manufacturers make it easy to turn one of the dipoles upside-down. That achieves the proper configuration. You might want to see whether that works for the Paradigms - maybe with small modification. Of course you'd have to figure out which one needs to be turned upside-down. Or open up the speaker and change the wiring on the drivers.

Ed

I don't think that left/right side couch is the way to go about checking this. As I said before, the effect is probably subtle. I think you'd have to compare with mirror image pairs.

Ed

You guys are right, my bad. I don't much attention to these dipole surrounds and forgot that most on the market aren't really dipoles through their full passband. You don't want the woofers out of phase of course, but this just means one of the tweeter/mids is wired out phase with it's own woofer. If you want a mirror image of sound this obviously won't work. Then again the whole idea is pretty unsound, which why there is no truly correct answer for which polarity should point to the front or back. Wired to be bipoles is probably better if you want the "ambiance", with no weird nulls or power response. Otherwise monopoles are probably the most technically correct-assuming the recording used these channels properly.

Well, I wouldn't give up on Paradigm just because of this. What you SHOULD give up is surround specific speakers. Keep that in mind if you switch to B&W.

I'm thinking you need to not worry about which is left and which is right and here is why...

In the case of 2nd order crossovers, the relative phase between woofer and tweeter require the tweeter to be reverse wired. This does not place them in phase but only partially in phase. That's why the output at the crossover frequency of each is -3 dB (71%). By wiring each tweeter with the opposite mid-range, they took care of the tweeter phase reversal.

Some other crossover topologies work differently. A 4th order Linkwitz for example has the woofer and tweeter in phase at the crossover frequency. That's why these crossovers have the output of each -6 dB. (50%). In each case the two add up to 100%. The two add up in the case of the Linkwitz because 50% + 50% at zero phase shift equals 100%. The two add up in the case of the 2nd order like I suspect your speakers have because 71% + 71% at 90 deg also add up to 100%.

Anyway I'm just realizing I'm out of time so I will try to continue this explanation later but where it's going is that there a lot of phase things going on much more complex than just plus or minus. I agree they should be mirrored but because both these and your mains have different phase at different frequencies it simply becomes too complex to decide which should be left or right. Which part of the frequency band do you want to be in or out of phase with the mains? Speakers are not linear devices so both the mains and the surrounds go through phase shifts across frequency.

Again, I have to run, I'm late for an appointment so I will try to comment more later.

Monte

Isn't all this irrelavent because you are describing what is happening in the vicinity of the crossover frequency? What is important here, is the polarity of the tweeter asymptotically above the crossover frequency.

Ed

If you must stay with a dipole setup (as opposed to wiring it to be bipole) then the only sensible thing to me is to keep the same relative polarity of the dipole woofers with the main speakers because the wavelengths are so long. The high frequencies have such small wavelengths that there really is no way for them to be perfectly in phase with the main speakers. They will go from in phase to out of phase depending on the frequency, room reflections, listening position, etc.

Still, if you have dipoles with the woofers in phase, than these would have to Butterworth filters (probably 3rd order) for them to still maintain a mirror image. But Butterworth filters don't form proper dipole radiation due to their nature of being in phase quadrature (90 degree phase seperation). If they are using Linkwitz-Riley (whether 2nd or 4th order) one side will have a null at the crossover. Neither of these options can be considered technically correct from any POV. I probably missed it but what slopes are they using-and not what the electrical filter is, the actual acoustic slope?

True but apparently in this case it is a 2nd order crossover with the tweeter reverse wired so above the crossover they are opposite the woofers. Also, depending on the impedance curve, there will be some phase shift across frequency in the pass band. If the main speaker has a different driver arrangement it's crossover may be in a different place. It's tweeter may or may not reverse phase. As for low frequency, the drivers resonance, the resonance of any tuned ports in either the surround or main will add additional complexity to the problem.

So, my argument remains that it is too complex to discuss as a simple plus or minus issue. I believe the manufacturer is correct, there is no right or wrong way to choose which speaker is right and which is left. That's not to say there is no difference. There might be but I think it would be very difficult to calculate which way is best. The only way will be through extensive listening test.

I do however agree with the OP's suggestion that it is important the two be mirrored. What ever one is doing, the other should be the same. I know I would do it that way. Actually I'm thinking I'm going to build bi-poles for my so called, "all dipole theatre" surrounds. Everything else will still be dipole however.

Monte

Even that description is unclear. Assuming they are in fact talking about acoustic slopes, and assuming the the mid/woofer is LR2, than the mid should be wired out phase relative to the woofer. Knowing that the woofers are wired the same, but the mids are opposite means one of the mids is wired in phase with it's woofer, which will cause a null around 300hz, made worse by the fact that it is 2nd order and will cover a very broad area, at least 150-600hz. Not to mention that you have bipole radiation becoming dipole radiation through this same area and this seems like a big mess.

The problem is there is no good science behind the idea, so it is impossible to say what is correct.

EDIT: I see there is only one woofer which is "shared" by the opposing TM's. If the TM's are out phase relative to each other then you will definitely have the problem I described above. This is just a poor design. I like the +/-2dB in the "reverberant soundfield" spec, that's just BS to cover that fact that they measure poorly, but if you put in their special reflective test room with an RT60 of 10 seconds and use a long measurement time it looks really smooth!

Yes, "a big mess" is a good description. It's true that it doesn't make much sense to combine omni with dipole for more reasons than one. The fact one of the midranges is in phase with the woofer and the other is not is the least of the problems in such a design. The bigger problem has to do with the directivity factor. The DF is 1.0 for the omni and 3.0 for the dipole. Further, at lower frequencies where it is omni, the room becomes a 2pi environment whereas it will become closer to 4pi at higher frequencies. This really messes up the power response. (Read this...http://www.musicanddesign.com/PowerMatching.html)

So yeah if you want to get technical about it, it's a really dumb design but that doesn't mean it can't still sound ok. I've pointed this out before, I don't hear too many Martin Logan owners complaining about the sound quality of their speakers and those have the exact same design flaw. I'm not sure why they do this. I don't know if they just don't know any better or if they have an arguable reason but ideally a monopole/dipole hybrid should be dipole at low frequency and monopole at high frequency not the other way around. This isn't perfect either but it's way better than the reverse at least in terms of a balanced power response.

So I go back and say that the ultimate answer here will be to mirror the speakers and flip a coin to determine which will be left and which will be right. Another option would be to make the speakers bipole. While my system continues to change and evolve as I slowly get it built my current plan is to go with omni/bipole surrounds. It may well be similar in design to the system in question here with the exception of having much of the woofers enclosure volume in the attic to maintain a more compact visible package. The rest of my system, sub-woofers, mains, center etc will be 100% dipole. Also not a perfect solution for balanced power response but I think about as good as it's going to get.

Monte