Neo3 PDRW and Neo8 PDR radiating area orientation

[twest820]

BG's whitepapers for these drivers mention PDR reducing the effective radiating area as frequency increases but don't specify how the area varies. The Neo3 PDR paper talks about the area narrowing to a slit which is apparently parallel to the long axis of the driver, suggesting the long axis should be kept vertical. There doesn't seem to be an equivalent paper for the updated PDRW version and I'm not finding any details of how BG adjusted the behavior between the PDR and PDRW versions. The Neo8 PDR paper is quite vauge as to the Neo8's behavior. I'm not finding the PDR patent, but the one speaker I'm finding which uses the Neo8 PDR apparently also keeps the long axis vertical.

Has anybody tested any of these drivers to see how off axis behavior varies as a function of driver orientation? What I'm aiming for here is getting raw driver data which I can use to reason about the consequences of placing the crossover between the Neo3 and Neo8 above the 300Hz-3kHz band where the ear is most sensitive.

[twest820]

(Reserved for summary of conclusions reached---if need be I'll buy the Neo3 PDRW, Neo8 non-PDR, and Neo8 PDR and measure them, but I'm hoping someone else has already done that. If not perhaps we can arrange for some drivers to be shipped around and measurements made. I see lots of folks mention they've had various Neos lying around for a year or more...)

[JonMarsh]

BG's whitepapers for these drivers mention PDR reducing the effective radiating area as frequency increases but don't specify how the area varies. The Neo3 PDR paper talks about the area narrowing to a slit which is apparently parallel to the long axis of the driver, suggesting the long axis should be kept vertical. There doesn't seem to be an equivalent paper for the updated PDRW version and I'm not finding any details of how BG adjusted the behavior between the PDR and PDRW versions. The Neo8 PDR paper is quite vauge as to the Neo8's behavior. I'm not finding the PDR patent, but the one speaker I'm finding which uses the Neo8 PDR apparently also keeps the long axis vertical.

Has anybody tested any of these drivers to see how off axis behavior varies as a function of driver orientation? What I'm aiming for here is getting raw driver data which I can use to reason about the consequences of placing the crossover between the Neo3 and Neo8 above the 300Hz-3kHz band where the ear is most sensitive.

You run the danger of using an axiom to chose your crossover frequency rather than measurements in a specific configuration, and choosing a higher frequency than you will get good physical blending and performance based on driver behavior.

the vertical listening axis for a neo8 is vertical- don't plan on laying them down and getting good lateral dispersion. I have measured and posted them previously, but am in an airport heading out and business travel and don't have time to look for the results. Also, the Neo3 has some real limits about how low it can go which need to be considered in any crossover configuration. I'd suggest measurements over speculation, and using distortion measurements to inform the crossover choices as well. A well executed crossover can sit right in the middle of the 300 Hz to 3 kHz band and be inaudible and difficult to detect by measurement. A crossover above 3 kHz is generally much easier to spot by ear and measurement, unless coincident drivers are used- phase differences due to the wavelengths causing off axis cancellations.

[twest820]

Thanks for the crossover comments. We seem to be in agreement; as a gentle reminder the question I'm asking is whether I need to get the data if someone else has already done so. The only thread I'm finding where Jon's posted Neo8 data is this one, which has Neo8 data for 0, 15, and 30 degree rotations about the long axis. It doesn't have data for rotation about the short axis or for the Neo8 PDR---am I missing something? Jon, are you saying the radiating surface of the Neo8 PDR narrows along the driver's short axis but continues to run the full length of the long axis?

The long axis of the Neo8's radiating area is around 160mm, which I personally wouldn't expect to disperse well crossed above 2kHz, but the BG Radia R-LCR is a WMTMW with Neo8 mids that have their long axis horizontal and cross at 2.4kHz. Martin Logan's Mosaic uses a similarly oriented Neo8 type driver crossed at 3kHz, though I've no information on what modifications Martin Logan may have requested of BG. Perhaps these manufacturers know something about the driver's behavior which I don't. Perhaps ML is using a PDR configuration which reduces the effective length; if so, that might be the same configuration as the standard Neo8 PDR. In which case, crossing at 3-3.5kHz to relieve the Neo3 might be viable.

To be clear I'm not thinking of crossing above 4kHz and the default driver orientation I'd use the same as the latest version of the Minilith (which, incidentally, crosses at 3.2kHz). However, as the Neo8 has some upside frequency potential one possibility I'm considering is a Duelund type crossover in the vicinity of 2kHz. Those are second order on the mid and fourth on the tweeter. The Neo8 wouldn't be all that far down in the 4-5kHz range and so having some sense of what its radiation pattern is like might save some time cutting baffles and taking measurements.

[villastrangiato]

If you think about it, the off axis response has to directly correlate with the width of the radiating area. The pdr versions simply have more padding extended towards the center of the driver from the sides. If a vibrating membrane excites all frequencies uniformly across its surface, it is a forgone conclusion that frequencies whose wavelength is shorter than the radiating surface width will suffer from combing cancellations. At 20khz, sound pressure waves are roughly three quarters of an inch. This explains why on axis behavior at that frequency might be fine but forty degrees off axis, a 20khz wave launched from one side of the diaphragm will most definitely cancel a similar wave launched from the opposite side. This is why it's not always a good idea to run a high quality 7 or 8 inch driver up to 3khz - even if it's clean up there - the beaming you'll get off axis could get pretty darn annoying. :E

[twest820]

Good data, thanks; I was hoping BG might have changed the membrane etching or magnet configuration rather than applying damping. Does the padding extend from the sides along the short axis of the membrane, the long axis, or both? From the pictures it looks like the drivers are pressed together, so I assume cracking them open to adjust the padding would be quite a hassle.

Yep, I'm familiar with self interference of acoustically large sources. Electrically large sources too, but I digress.

[villastrangiato]

I spoke to Igor - the chief engineer/designer of the Radia drivers at BG- briefly about this distinction, and he confirmed that it's simply more padding extending from the edges to the center across the shorter length (horizontal axis). As Jon indicated, the Radia drivers are primarily intended and used with the longer axis positioned vertically - this provides for the intended design criteria of restricted vertical dispersion (desirable for line sources) while maximizing the horizontal dispersion. You can see how increased driver width impacts off axis (horizontal) response when going from the NEO 8 to NEO 10 - the bandwidth specifications show a very dramatic change. The difference between the NEO 3 and NEO 8 are not quite as pronounced as these drivers possess a radiating vertical band of comparable width. While the drivers are intended primarily for maximal horizontal dispersion and reduced vertical dispersion (javier huerta can attest to limited vertical dispersion 8O ) some BG products have been produced as HT center channels with drivers "on their sides" to maximize vertical dispersion. But that has not been the focus of BG's products and seems more like an afterthought than an original design goal. A number of people have complained that these drivers are "difficult to work with". To me, it's a matter of understanding the design's original intent - hence its strengths - and then putting those strengths to work in a design that calls for them.

[villastrangiato]

The long axis of the Neo8's radiating area is around 160mm, which I personally wouldn't expect to disperse well crossed above 2kHz, but the BG Radia R-LCR is a WMTMW with Neo8 mids that have their long axis horizontal and cross at 2.4kHz. Martin Logan's Mosaic uses a similarly oriented Neo8 type driver crossed at 3kHz, though I've no information on what modifications Martin Logan may have requested of BG. Perhaps these manufacturers know something about the driver's behavior which I don't. Perhaps ML is using a PDR configuration which reduces the effective length; if so, that might be the same configuration as the standard Neo8 PDR. In which case, crossing at 3-3.5kHz to relieve the Neo3 might be viable.

To be clear I'm not thinking of crossing above 4kHz and the default driver orientation I'd use the same as the latest version of the Minilith (which, incidentally, crosses at 3.2kHz). However, as the Neo8 has some upside frequency potential one possibility I'm considering is a Duelund type crossover in the vicinity of 2kHz. Those are second order on the mid and fourth on the tweeter. The Neo8 wouldn't be all that far down in the 4-5kHz range and so having some sense of what its radiation pattern is like might save some time cutting baffles and taking measurements.

If you are very concerned about off axis performance, then the clear limitation on the NEO 8 would be about 6khz. Running it considerably higher would present a noticeable dispersion issue. As for lobing - the limited vertical dispersion is going to play a major role in the end result - particularly if multiple NEO 8's are being used in a line source configuration - this is a good example of how one designer's perceived weakness or liability can become another designer's strength - again - BG's corp's focus has been and thankfully continues to be line source transducers. If you're frame of reference is point source - you really should be looking at other manufacturer's products.
For me, off axis performance is not as critical as tonal balance, transient response, and the ability to resolve dynamic extremes. Most people who are thinking of working with large format drivers for the first time would do very well to take a stroll down to their local Magnolia dealer and listen to the Martin Logan Source. Compare it to the top of the line Vienna Acoustics speaker - also a nice sounding design. The differences, design goals, and "strengths" of each markedly different design will be immediately apparent.

[twest820]

Tradeoffs, tradeoffs. My application is dipole and the rear radiation path of small woofers is quite obstructed by the spider and magnet. AMTs don't fit in the budget, ribbons don't go low enough, and back to back driver mounting gets awkward below tweeter frequencies. For the point source design I have in mind it would be better if the Neo8 were square, but it's still compelling for its reasonable cost, low distortion, and dipole properties. My current system has a 330mm radiating height crossed LR4 at 2kHz. I've lived with it for years and mostly been fine with the vertical dispersion. Latest thinking is to cross from the Neo3 to Neo8 LR6 at 2kHz, lower if I can get away with it, so vertical dispersion of the non-PDR version with the long axis vertical shouldn't be an issue.

[villastrangiato]

I'm not sure what the advantage is to running Neo 3's down below 3 or 4k if both drivers are set at approximately listening height and your polar response is the product of a L-R crossover. If you simply apply Pythagorus' theorem you'll find under those circumstances that the sound wave path length differences for the crossover frequencies spread across the two drivers vertically mounted one over the other won't amount to a hill of beans. Now if you stand up and move around the room - all bets are off :f>

But that is part of the design compromise of just about every large format point source implementation you can think of - the Martin Logan Source is a good example. Most people listen critically to movie soundtracks and cd's from a stationary listening position. The horizontal plane is not a problem so you shouldn't experience any significant degradation moving for example from one end of the couch to another - provided you're 8 to 10 feet away from speakers that are toed into your general location.

If you're hell bent on running the NEO 3's down to 1100 to 1500hz, that's not a problem - you can take two of them mounted one over the other and set them up in an MTM configuration (third order cross) with the backing cups removed either in a shallow stuffed enclosure or dipole. Running them in series will bump up the impedance and lower sensitivity a tad but will reduce the stresses associated with producing lower frequencies without distortion at higher power levels versus trying to run one driver down that low. But again, why? The NEO 8 doesn't have any horizontal dispersion issues until you get well past 4khz. If you are set on doing a two way cross from Neo 3's to a decent 7 or 8 inch woofer at 1100 hz or so - scrapping the Neo 8 altogether, this might make some sense - but otherwise - why go this route?

[twest820]

Below 2kHz means perhaps 1800Hz; see the other discussions of the Neo3 here and at DIYAudio. As to crossing below 3kHz, see Jon's comments above.

The Source doesn't seem particularly relevant as its panel is much larger than a Neo8.

[villastrangiato]

Below 2kHz means perhaps 1800Hz; see the other discussions of the Neo3 here and at DIYAudio. As to crossing below 3kHz, see Jon's comments above.

The Source doesn't seem particularly relevant as its panel is much larger than a Neo8.

I am very familiar with the Radia line of drivers. If you want a strong indication of what the Neo 3 is capable of in terms of bottom end - see Zaph Audio's section (Battle of the Non - Domes). Pay particular attention to the waterfall plot and distortion graphs provided - they show second and third harmonic distortion levels at 1khz that rival traditional tweeters costing nearly ten times that of the Neo 3. Granted, Krutke's measurements also show roll off beginning at 2khz - also backed up by BG's data. But if you pair up two Neo 3's, according to the manufacturer, you will get constructive interference at the bottom end which levels response out to 1khz. For the money, there are very, very few tweeters that come anywhere near this level of performance (approx. .05% 2nd and 3rd harmonic @ -10db or -60 to -65 db down at 1khz according to Krutke)

As for the Martin Logan Source - I disagree entirely. It shares a very significant feature with the large format Radia transducers - and by definition, they are indeed "large format" when you compare the radiating area of most tweeters and dome midranges. The feature in question is highly directional radiation patterns. In the Martin Logan Source case, if you stand up from the "sweet spot", it's as if someone turned off the tweeter/upper midrange switch. The same can be said for both Neos- move significantly in the vertical plane in close proximity to the driver and the driver's output practically disappears - not just high frequencies - all frequencies are severely attenuated. If you own Neo 3's and 8's - I know you know this from experience. In this very significant way - the Logan is strikingly similar.

[twest820]

I'm unsure what point you're trying to make. I believe it's well established the Neo3 is preferable to the Neo8 as a tweeter and that the PDR Neo3 is preferable to the non-PDR Neo3 due to the Neo3 PDR's greater dispersion. What's unclear is the specific off axis behavior of the units. General discussion about radiation patterns and analogies to much larger electrostat panels aren't going to provide the data needed to design with the two Neo drivers. Hence my question about measurements at the start of this thread.

[villastrangiato]

BG's whitepapers for these drivers mention PDR reducing the effective radiating area as frequency increases but don't specify how the area varies. The Neo3 PDR paper talks about the area narrowing to a slit which is apparently parallel to the long axis of the driver, suggesting the long axis should be kept vertical. There doesn't seem to be an equivalent paper for the updated PDRW version and I'm not finding any details of how BG adjusted the behavior between the PDR and PDRW versions. The Neo8 PDR paper is quite vauge as to the Neo8's behavior. I'm not finding the PDR patent, but the one speaker I'm finding which uses the Neo8 PDR apparently also keeps the long axis vertical.

Has anybody tested any of these drivers to see how off axis behavior varies as a function of driver orientation? What I'm aiming for here is getting raw driver data which I can use to reason about the consequences of placing the crossover between the Neo3 and Neo8 above the 300Hz-3kHz band where the ear is most sensitive.

Yes, getting back to your point of what you were seeking in the first place. I think the answers given were pretty much on the mark - the drivers are highly directional having very little vertical dispersion (see javier huerta's minilith build comments), the off axis horizontal response is well documented in John Krutke's full analysis online, and I quoted you directly from conversations I recently had with the guy who actually designed these things what was done to achieve wider horizontal dispersion in both drive units. I also answered your question about the crossover at 3 - 5k to the Neo 8 as being inconsequential in terms of polar response and crossover design (see huerta's build and the serious consequences of moving from sitting to standing position with these types of implementations). Once again, these speakers have very limited vertical dispersion as they were primarily intended to be used in line source designs (BG's product line should serve as a slight hint there :drool: ) Frankly, I don't think anyone has gone off topic or engaged in a :rant: . On the contrary, a wealth of information has been cited for your benefit. But as always, no good deed seems to go unpunished....... :roll:

[twest820]

I do appreciate the effort, but I'm afraid some of is in a direction which is not helpful to me. The information on the padding is quite useful in ruling out certain driver and crossover combinations---I think at this point I can exclude the Neo8 PDR entirely. Your mention of the Neo8-S on the elliptical thread is a most interesting heads up as well.

It occurs to me the Neos are trivial to model in Edge; don't have time tonight but I'll take a look and post the results. That should provide some data for discussion; I'm still a few weeks out from ordering and testing Neo3s and will push any Neo8 work out until I can evaluate the Neo8-S and Neo10.

Oh yeah, I'm aware of Javier's build but I guess my takeaways from the thread differ from yours.

[Dennis H]

the PDR Neo3 is preferable to the non-PDR Neo3 due to the Neo3 PDR's greater dispersion.

Not necessarily. Someone over at DIY found that the polars measured better on a dipole baffle with the non-PDR version and he got several dB more sensitivity on the low end as a bonus. Keep in mind that you're balancing the controlled directivity of the dipole at low frequencies and the controlled directivity of the membrane width at high frequencies.

You can bench race this to death but at some point you just have to buy some drivers, take some measurements and decide which works best for your particular application. That's the Y in DIY. You do have measurement gear, right?

[villastrangiato]

Not necessarily. Someone over at DIY found that the polars measured better on a dipole baffle with the non-PDR version and he got several dB more sensitivity on the low end as a bonus. Keep in mind that you're balancing the controlled directivity of the dipole at low frequencies and the controlled directivity of the membrane width at high frequencies.

You can bench race this to death but at some point you just have to buy some drivers and take some measurements and decide which works best for your particular application. That's the Y in DIY. You do have measurement gear, right?

I was gonna mention that but I didn't want to beat this one to death.....but.....since you've picked up the ugly stick......... :B I'm in total agreement. To put it plainly and simply, the pdr versions give up sensitivity, maximum spl, and low frequency bandwidth for some gains in lateral dispersion. Is that for everyone? Clearly no. And certainly not for me anyway. It depends mostly on how you listen. I gave up a long time ago trying to maintain great imaging and uniform response from a system while walking around a room. From a physics standpoint, I'm convinced it's folly - simply unachievable to any meaningful degree. I generally don't listen that way anyway. Typically, I'm sitting in one place - not an unusual scenario for a lot of people. So from this perspective, a highly direction (typically line source) solution makes perfect sense. The soundfield is narrowed considerably leaving less energy reflected off the walls to compete with the first arrival wavefronts and confuse the image your trying to assemble. Hence, it should be pretty obvious I'm not a big dipole, bipole, or Bose-pole fan. This is why when someone starts talking about line source speakers or drivers - I have an immediate inclination to find out what the person is looking for - what are they trying to accomplish - do they really understand the myriad of differences between the point source and line source approach to sound reproduction in a home environment. I'm not saying you can't use line source ribbons or planars in a speaker that is intended for point source radiation. But in a way, it's like taking a top fuel dragster to the Brickyard in Indianapolis to go racing. Yeah, it'll blow everybody's doors off until that first turn..... 8O

[twest820]

I've a measurement rig with mic preamp issues and the replacement interface is in the post. In the near term (by which I mean the next several months) what I'd like to accomplish using the BG drivers in a point source design is to get ears on with magnetostats, dipoles, and open baffles in a prototype design. For doing that the price point of a pair of Neo3s is tough to beat.

Screen captures from Edge for the raw drivers are below. Curves from top to bottom are 0, 15, 30, 45, and 60 degrees off axis. In particular the 30 degree off axis response about the Neo8's long axis stops tracking the on axis response around 1.5kHz. That's comparable to where Edge predicts a roll off for a 6.5 inch woofer. Should be thoroughly unsurprising result as the woofer diameter and the long axis of the Neo8 are about the same size. The short axis data from Edge is in reasonable agreement with BG's published data and Edge's woofer model agrees passably with the handful of published woofer measurements I checked.

Rolloff on the Neos is steeper than with a circular driver as, in relative terms, more of their surface area is farther away from the acoustic center. In theory I don't necessarily see a problem using the Neo8 with its long axis horizontal so long as it's acoustically crossed high order at 2kHz or below. But that orientation wouldn't be my first choice. I'm not finding measured off axis curves for the non-PDR Neo3 but the short axis results from Edge show rolloff similar some of the more directional soft dome tweeters like the rather popular Seas Millennium as well as hard dome tweeters without phase shields.

I have a high frequency hearing loss that means I don't hear much above 8kHz. So I think I'll start with the Neo3 non-PDR and see how that goes.