Spin-o-rama for synthetic Bipole and Dipole speakers?

[Blumlein 88]

So we don't have many spins for dipoles and bipoles. Panels more or less don't fit in the Klippel. So I am wondering if some of you folks can take the data from Amir's tests and show the resulting directional plots and response for in room results along with what Harman score such a thing would achieve. I can visualize pretty well what those look like, but haven't learned to create such a thing in the software. What I have in mind is if you can take the Klippel results for a regular speaker and create what the result would be if you had two speakers back to back let us say within 2 inches (5 cm) of each other. For each such a pair they can be in phase for bipoles or out of phase for dipoles obviously.

If you want suggestions I might suggest back to back F208s, LSR 308s or M106s since I have used those. Maybe some lesser scoring speakers would be nice to show as well.

I have seen @MZKM, @napilopez and a few others show plots that makes me think they can do this. So if some of you can forgive my laziness and provide those results I would be interested. Maybe it would be a topic of interest to others. Maybe I'm asking for too much free effort from those who could do this. Thanks in advance for anyone who can do this for this thread. Even one for a good speaker would be nice.

Or if any of you know how to make use of the data from Amir's testing, point me to where I can learn and do this for myself. I'm just not up to speed on it.

 

[Duke]

So we don't have many spins for dipoles and bipoles. Panels more or less don't fit in the Klippel. So I am wondering if some of you folks can take the data from Amir's tests and show the resulting directional plots and response for in room results along with what Harman score such a thing would achieve. I can visualize pretty well what those look like, but haven't learned to create such a thing in the software. What I have in mind is if you can take the Klippel results for a regular speaker and create what the result would be if you had two speakers back to back let us say within 2 inches (5 cm) of each other. For each such a pair they can be in phase for bipoles or out of phase for dipoles obviously.

If you want suggestions I might suggest back to back F208s, LSR 308s or M106s since I have used those. Maybe some lesser scoring speakers would be nice to show as well.

I have seen @MZKM and a few others show plots that makes me think they can do this. So if some of you can forgive my laziness and provide those results I would be interested. Maybe it would be a topic of interest to others. Maybe I'm asking for too much free effort from those who could do this. Thanks in advance for anyone who can do this for this thread. Even one for a good speaker would be nice.

If anyone decides to undertake this, if you're modelling a bipolar loudspeaker, note that below the baffle-step frequency there will be significant interaction between the energy from the front-firing woofer and the rear-firing woofer. The wrap-around energy from the rear woofer will partially cancel the front-firing energy at frequencies where it's arriving out-of-phase (or approximately so) due to the delay imposed by the wrap-around path length. And at lower frequencies where the wrap-around path length is a small enough fraction of a wavelength, the wrap-around energy will partially reinforce the front-firing energy.

The Mirage M1 that @Floyd Toole mentions in his book has a wide-and-shallow configuration to mitigate this wrap-around dip. To be fair the wrap-around dip is not very audible, as it's a comb-filter effect so the energy is still there in the reflection field, but it shows up in my simulations (which are not Klippel-based). Below the dip frequency region the wrap-around energy from the rear woofer adds to the energy from the front woofer. I used to manufacture full-spectrum bipolar speakers and used a similar wide-and-shallow configuration.

There are other bipolar formats which do not have this issue, but they are not going to be replicated by the simulation Blumlein88 is describing.

I would be interested to see what happens to the predicted preference score for a simulated bipolar version of a speaker that has been measured.

For a simulated dipole speaker, I would expect the preference score to take a nose-dive because of the response peak where the wrap-around path length is equal to one-half wavelength, followed by cancellation resulting in a 6 dB per octave rolloff below that frequency.

There is a possibility that I will have a bipolar speaker Klippled one of these days, but it probably will not be a full-spectrum bipole.

 

[Blumlein 88]

If anyone decides to undertake this, if you're modelling a bipolar loudspeaker, note that below the baffle-step frequency there will be significant interaction between the energy from the front-firing woofer and the rear-firing woofer. The wrap-around energy from the rear woofer will partially cancel the front-firing energy at frequencies where it's arriving out-of-phase (or approximately so) due to the delay imposed by the wrap-around path length. And at lower frequencies where the wrap-around path length is a small enough fraction of a wavelength, the wrap-around energy will partially reinforce the front-firing energy.

The Mirage M1 that @Floyd Toole mentions in his book has a wide-and-shallow configuration to mitigate this wrap-around dip. To be fair the wrap-around dip is not very audible, as it's a comb-filter effect so the energy is still there in the reflection field, but it shows up in my simulations (which are not Klippel-based). Below the dip frequency region the wrap-around energy from the rear woofer adds to the energy from the front woofer. I used to manufacture full-spectrum bipolar speakers and used a similar wide-and-shallow configuration.

There are other bipolar formats which do not have this issue, but they are not going to be replicated by the simulation Blumlein88 is describing.

I would be interested to see what happens to the predicted preference score for a simulated bipolar version of a speaker that has been measured.

For a simulated dipole speaker, I would expect the preference score to take a nose-dive because of the response peak where the wrap-around path length is equal to one-half wavelength, followed by cancellation resulting in a 6 dB per octave rolloff below that frequency.

There is a possibility that I will have a bipolar speaker Klippled one of these days, but it probably will not be a full-spectrum bipole.

Yes, a friend of mine owned those Mirage M3i speakers for years. Quite familiar with them. And I've owned more panels than any other kind myself. I know it was a bit of a fad for while when people like Thiel were telling people to buy pairs of their speakers and run them back to back as bipoles. Heard a few of those too. I know of course when you have the cabinet between box speakers back to back it isn't going to be a replication of panels.

Part of my thinking on this is having worked with microphones where you combine different patterns for different reasons. While in theory you'd want them to be in one plane or point they often work better than expected if they are just close. Yet box speakers will be further apart than mikes. Unless a dedicated design like the wide and rather thin Mirage.

 

[NTK]

I am a complete noob with AFMG EASE Focus (free download, link below), but it seems it can do it if you have access to the loudspeaker's GLL model.

EASE Focus 3 - Free Download | Ahnert Feistel Media Group

www.afmg.eu

Below is my screen dump of 2 back-to-back Genelec 8351s (one with inverted polarity). Again, I am a complete noob of this software so it will take me significant learning to go further.

[Edit] Here is a presentation that should give a pretty good idea what the software is for and what it can do.

Focus Your Sub Arrays! | Ahnert Feistel Media Group

www.afmg.eu

 

[Duke]

I know it was a bit of a fad for while when people like Thiel were telling people to buy pairs of their speakers and run them back to back as bipoles.

Really? I had no idea! Never heard of that recommendation coming from a speaker manufacturer before. And Jim Thiel, no less! Do you remember roughly when that was?

Hmmm. You might be able to mitigate the SBIR dip by having the two woofers at significantly different distances from the front wall.

Heard a few of those too.

So... what did you think?

Did you ever compare single-speaker versus back-to-back pair in mono? Imo it's rather interesting. Of course you hear some problems because it isn't optimized, but you can also arguably hear the potential in the bipolar format.

 

[Blumlein 88]

Really? I had no idea! Never heard of that recommendation coming from a speaker manufacturer before. And Jim Thiel, no less! Do you remember roughly when that was?

Hmmm. You might be able to mitigate the SBIR dip by having the two woofers at significantly different distances from the front wall.



So... what did you think?

Did you ever compare single-speaker versus back-to-back pair in mono? Imo it's rather interesting. Of course you hear some problems because it isn't optimized, but you can also arguably hear the potential in the bipolar format.

This would have been just after the introduction of the CS5 and when the CS3.6 were out. 1990-95 or so. Thiel had in mind a flat power response which many thought an optimum design choice prior to Toole's work. So that is close to flat on axis in room which is too bright. I always thought Thiel's were bright. They used 1st order crossovers with off axis lobing which made them picky about placement and aiming. So back to back it warmed them up, and was a better balance it also altered them in the room I'd think so that some of the backside sound likely bounced around filling in off axis dips and since some of that sound from the back would be absorbed more in well damped rooms it helped with its bright signature. So it did sound better to me. But the issue was even a CS3.6 is not a cheap speaker if you essentially double the cost you had better choices. They needed pretty powerful amps so you needed another pair of powerful amps. Not surprised it never caught on. Did not every hear them this way on mono material.

 

[Elkerton]

Section 9.4 of Toole's Sound Reproduction goes into 'boundary-friendly speakers' like the Allison Ones and the AR-9s. Figure 9.14 shows an Infinity Beta ES250 and sweeps on axis, +or- 30°, +or- 10° vertically, and sound power in its bipole configuration. The curves are very good.

 

[staticV3]

So I am wondering if some of you folks can take the data from Amir's tests and show the resulting directional plots and response for in room results along with what Harman score such a thing would achieve.

You can find graphs and preferences scores for a few panel-type speakers here: https://www.spinorama.org/?sort=date&reverse=false&shape=panel

 

[Blumlein 88]

You can find graphs and preferences scores for a few panel-type speakers here: https://www.spinorama.org/?sort=date&reverse=false&shape=panel

Had not seen these. You can see lots of issues with these. That was one of things I wondered about if you used a couple of good testing speakers to create a dipole. Panels are still a very appealing sound to many. Is it partly the dipolar nature despite other issues? If so would a pair of controlled directivity speakers making dipolar sound be an improvement? Opposite direction to some of the cardioid pattern speakers made by Kii or D&D.

 

[Floyd Toole]

I just dipped into this thread and have a request: please, please stop putting any reliance on the calculated "scores". Learn to interpret the spinorama curves. That will have to do until we have an "educated" AI version of sound quality prediction. The ratings that were calculated by the Harman research group were done to prove a scientific point, and that done, they ceased to be used even by the people who created them. We rely on visual interpretations of the family of curves.

 

[Blumlein 88]

I just dipped into this thread and have a request: please, please stop putting any reliance on the calculated "scores". Learn to interpret the spinorama curves. That will have to do until we have an "educated" AI version of sound quality prediction. The ratings that were calculated by the Harman research group were done to prove a scientific point, and that done, they ceased to be used even by the people who created them. We rely on visual interpretations of the family of curves.

Well interpreting the curves with what understanding I have of them seems more useful than the scores. A score is listed for speakers Amir measures here and sometimes it looks okay, and sometimes I'm wondering about it. Do you have any further insight into how the curves can most usefully be applied? If more of us know that we won't care about the "score".

 

[olieb]

Learn to interpret the spinorama curves.

Sure, but this interpretation will be subject to all kinds of sighted bias and self trickery again.
While I would not trust the calculated score too much, it has the advantage of being an objective number that - in theory - could be tested and falsified. It would be nice to have ever improving prediction models of course.
And for the question in this thread it is not so clear how to interpret the curves for a dipole kind of speaker at all.

 

[Floyd Toole]

I assume you have read my book, because I think the info you need is in it. I am currently working on a 4th edition that I think might spell out the process more explicitly. In any event, recognizing a neutral loudspeaker - which is the goal - is primarily to identify if it exhibits audible resonances. This is done by scanning down the family of spinorama curves looking for repeated "bumps". A bump on axis is not enough evidence, but if the pattern is repeated in the lower curves it is definitely a resonance. This is why we like smooth curves on and off axis. If the loudspeaker passes this test, the other issues are spectral balance (broadband "tone control" kinds of trends) and frequency dependent directivity.
A smooth flattish on-axis and listening window curves are essential beginnings. Irregularities in off-axis responses (early reflections and sound power) are unwanted, especially when they also are signaled in the DI curves. Directivity, per se, is not a major "sound quality" issue, but something that interacts with one's impression of stereo listening, where there seems to be endless effort to find a loudspeaker that can make directionally and spatially deprivesd stereo deliver something it is incapable of. Spinoramas will reveal problematic bipoles and dipoles.
Low frequency extension is very important, because bass accounts for roughly 30% of an overall sound quality rating. A bookshelf speaker will always score lower than a floor stander with more extended bass. However, add a properly integrated subwoofer, or multiple subs, to a bookshelf speaker and it is a different game entirely.

 

[Ron Texas]

I just dipped into this thread and have a request: please, please stop putting any reliance on the calculated "scores". Learn to interpret the spinorama curves. That will have to do until we have an "educated" AI version of sound quality prediction. The ratings that were calculated by the Harman research group were done to prove a scientific point, and that done, they ceased to be used even by the people who created them. We rely on visual interpretations of the family of curves.

I have noticed some members here getting tied up in their shorts over preference scores to the point that they are obsessing over a .2 difference. I always suspected this was the case. Now all we need is a lesson from one of the experts here on how to read the curves.

Thank you Dr. Floyd.

 

[Neuro]

An important development of the spinorama technique could be, based on objective measurements, to extrapolate how in a given room (the customer's listening room) calculate the arrival time and loudness of dominant reflexes in the listening position relative to the direct sound. Based on this, it is possible to roughly determine whether a given speaker and a given room will provide optimal colouring of the direct sound or not.
Concerning dipole and bipole speakers, such a calculation should give a more objective picture of the mix of direct sound and reflected sound in rooms other than Harman's spinorama listening room.

There is some consensus on how optimal reflexes should be delayed and attenuated in the listening position. Brannon, Toole/Olive, Genelec, .... suggest approx. 8 dB attenuation and approx. 20 ms delay relative to the direct sound of reflexes with a frequency distribution similar to the direct sound in the listening position.

Is this difficult to achieve?

 

[kemmler3D]

An important development of the spinorama technique could be, based on objective measurements, to extrapolate how in a given room (the customer's listening room) calculate the arrival time and loudness of dominant reflexes in the listening position relative to the direct sound. Based on this, it is possible to roughly determine whether a given speaker and a given room will provide optimal colouring of the direct sound or not.
Concerning dipole and bipole speakers, such a calculation should give a more objective picture of the mix of direct sound and reflected sound in rooms other than Harman's spinorama listening room.

There is some consensus on how optimal reflexes should be delayed and attenuated in the listening position. Brannon, Toole/Olive, Genelec, .... suggest approx. 8 dB attenuation and approx. 20 ms delay relative to the direct sound of reflexes with a frequency distribution similar to the direct sound in the listening position.

Is this difficult to achieve?

I think this type of simulator is very doable - raytracing reverb simulators were available at the "consumer" level at least a decade ago. Upload the floor plan of your room, pick out where the speakers go, run the raytracer, generate the IR and then you can listen to the simulation on your headphones.

I think the tricky part would be getting accurate absorption / scattering behavior in there. I think reflective surfaces are easy to simulate this way, but I am not sure how you'd approach the couch, rugs, curtains, and bookshelves full of crap, let alone purpose-built acoustic treatments.

 

[somebodyelse]

I think the tricky part would be getting accurate absorption / scattering behavior in there. I think reflective surfaces are easy to simulate this way, but I am not sure how you'd approach the couch, rugs, curtains, and bookshelves full of crap, let alone purpose-built acoustic treatments.

Windows, doors and different types of wall too. Non-cuboid rooms are more of a user interface problem than a calculation problem.

 

[kemmler3D]

Windows, doors and different types of wall too. Non-cuboid rooms are more of a user interface problem than a calculation problem.

With present-day AI tools, it feels like it should be possible to put together a very accurate layout of a room based on a floorplan drawing and some photos or a video. I guess there's still the issue of whether the simulator will handle all types of surface properly though. I am sure it can be done but maybe we're now talking more about a pro-level app than a small consumer-grade project. Anyway, I am not a software developer so don't ask me about the details

 

[benanders]

From a consistency standpoint, I worry interpretation of pref scores or (perhaps less so) graph curvatures could be inaccurate - depending on a given viewer’s level of experience and other factors.
Standardization through the step of interpretation/inferences is key for this sort of thing, so it’s a game not everyone will be fit to play. I for one do not pretend I’m a meaningful contender, but I could just as easily be missing something (yes, 3rd edition of aforementioned book has been read).
The level of bias imposed on different persons’ considerations by this strategy (and vice versa) could be, well, considerable.
Stuff like this, that’s far better than nothing, is still prone to misuse and/or abuse whether purposeful or inadvertent. Proceed with caution?