DrWho

Btw, here are some STC for Plexiglas: http://www.eplastics.com/Plastic/plastics_library/Plexiglass-Noise-Reduction But note the caveat below the charts: For free-standing plexi, you can subtract a good 10 to 20dB from these charts. 4" may have been an exaggeration, but the point is you can't get it thick enough to pull off numbers equivalent to a good cardoid without other implications.

The PZM pattern is the same as an omni-directional with the bottom half cut off - just a single hemisphere. In fact, the construction of a PZM is really just an omni-directional capsule placed as close to the boundary as possible. This means comb-filtering above the 1/4 wavelength distance - and that's where the special sauce happens. Your rejection pattern (the side opposite the hemisphere) is limited by the attenuation properties of your boundary. A thin piece of plexi doesn't offer consistent attenuation and gets less as you go lower in frequency. And diffraction is the reason the panel needs to be large - again, something that gets worse as you go lower in frequency. Guess what - polar control of the speaker gets worse as you go lower in frequency too. And cardoid mics don't sound awful. When are you going to start recommending magic microphone cables and special power cords?

The polar pattern tells you the relative pickup of the microphone in different directions. Directly in front is normalized to 0dB. As you rotate around the polar pattern, the sound pressure turned into voltage decreases as the polar bubble moves towards the center. The steps in these charts are generally 5dB. When you place a microphone in a system, the GBF benefit of the microphone is the on-axis response subtracted from the off-axis response. The on-axis response is defined by the angle of the mic relative to the desired source (the person talking). The off-axis response is defined by the angle of the mic relative to the undesired source (the speakers or reflection points in the room). If you want to be hardcore, then you need to sum the energy from all of the undesired sources. We don't need to run real numbers though - just point the rejection lobe of the microphone towards the source of feedback. Here are pictures illustrating the concept. The PZM on-axis and off-axis response is the same - so the GBF benefit is 0. With a cardoid, you could be as high as 20dB, but in practice it's usually more like 12dB (because there is usually more than one feedback path). Using plexiglas, there is a lot of complex frequency dependent behavior happening - even if you position the PZM so that the 180 degree pickup is pointed away from the speaker. It's because the plexiglas isn't infinitely large. The plexiglas also lets sound travel through it, as well as vibrates at certain frequencies (it becomes it's own reradiator). At the end of the day, the cardoid is already providing a 180 degree pickup in the forward lobe, and provides better attenuation in the rearward lobe. Now if you wanted to use some 4" thick plexi in a 4x4 ft sheet, then maybe you'd get enough blocking - but why stop there? Why not put everyone inside a fishbowl?

Hah! Nice!

The difference between a choir and a theatrical performance is the number of people speaking (singing) at the same time. The distance from the acoustic source to the microphone is offset by the number of acoustic sources. A single person talking would need about 10dB more gain versus 8 people covered by the same mic. Ever heard a solo from the middle of a choir that wasn't using a separate mic for the solo? The other problem is the amplitude variation as the talent moves around. That effect gets masked a bit when multi-mic'ing a choir. It's much more noticeable with a single voice. Perhaps it's because I'm a professional, but I simply won't "attempt" something that won't confidently meet a minimum quality standard. It's also been my experience that inconsistent sound quality is a lot more frustrating than consistently bad. It's also been my experience that a consistent bad scenario is the best way to drum up the resources to actually fix the problem correctly. When you go in trying to make a less than ideal setup work, then you waste all this time distinguishing between user error and the actual demerits of the approach. But hey, you armchair quarterbacks won't be on the receiving end of the scowls generated by these crazy ideas. It has been my experience that doing something poorly is much worse than doing nothing. When doing nothing sounds bad, then you can say "hey, here's the proper way to solve this problem". If every parent of these kids donated $200 like this one parent wants to, then the camp is going to be set for a very long time with a proper solution. This burden shouldn't be carried by the one dad with enough experience to know that it's a solvable problem.

This isn't my opinion - physics tells us that it's not ideal for GBF. You are very keen to point out that you've had success in the past - I won't argue with that perception at all (it would be insane actually). You're an honest dude and I fully believe your experience. You had enough GBF in your application. However.... You had enough GBF despite your approach - not because of it. The people in your example were were simply projecting enough to not require much gain from the system. The thing about GBF is either you have enough or you don't. If you have enough, then it doesn't matter if one approach gives you more - once you don't have feedback, then it's not like you can have less than no feedback. This doesn't mean your approach is the best approach - it just means it worked for your application. And that point is important because your approach doesn't work for the other 90%. If you were giving advice, would you recommend something that worked 10% of the time, or 90% of the time? And if you saw someone offering up crap advice, then wouldn't you speak up?

GBF = Gain Before Feedback.

I forgot to mention one thing....with any microphone solution, you're going to need to EQ the feedback frequencies. These frequencies are unique frequencies that wrap around the speaker or microphone due to their construction. There are often 3 to 6 of these frequencies that exist - while all the others tend to be well behaved. Be careful though because these frequencies slide around with position. EQ'ing out these frequencies is basically flattening the off-axis response of the system, which lets you turn the microphone channel up higher. There are feedback reducers that do this automatically for you, but they don't work better than an educated operator. All you need is a high Q parametric EQ. Don't get crazy with this though - notch the first few frequencies and then don't try to add more. You'll hear when it's enough because the feedback will move from one tone to a series of multiple tones. This is why hyper-cardoid mics are bad....they have big spikes in the off-axis response. The goal is to have smooth rejection patterns. For both the speaker and the mic. The thing is, the good microphones and speakers that do this aren't cheap.

It's more of a liability to be honest. You don't get a rejection pattern with a PZM - this reduces your GBF. The Heresy absolutely would not work. That small cabinet with a single 12" driver has no control over the polar response....that translates to lots of bleed back onto the stage, which also reduces GBF. Most of the Klipsch pro line suffers from the same problem. You need to get into the bigger horns before the pattern control starts to get controlled adequately. There is some Klipsch pro that accomplishes this. Line arrays control the vertical polars extremely well - but they often don't control the horizontal very well. Whether or not that matters depends on a lot of factors. The amount of horizontal control varies quite a bit too. The thing about these devices is that they are tools. Anyone getting on a philosophical pillar about which method works best doesn't understand the intricacies of the tools. Just as a carpenter may have multiple saws and hammers, so does the sound engineer. The factors for sound quality in the live sound world are totally different than any audiophile pursuits that might be discussed here. If you're looking for a good pro audio resource, then check out this website: http://www.prosoundweb.com/ They have forums too.

For $200 you're simply not going to accomplish your goals. Floor mic'ing is not a viable solution to this problem. I could explain why, but my experience tells me you're going to try it anyway. So to get the most out of floor mic'ing, the first step is to get the right polar pattern. Don't use highly directional mics (no hyper-cardoid). Use a standard small diaphragm condenser microphone with a cardoid pickup pattern. Position the mics in such a way that the rejection null of the microphone is pointed towards the source of your feedback (the speakers). And if you have control over the speaker position, then work hard to minimize the amount of sound from the speaker that ends up on the stage. Your goal here is to maximize the "gain before feedback". This is a boring physical reality that can be quite readily calculated. There are ton of articles on the subject, but it all amounts to minimizing how much sound from the speaker ends up in the microphone. The reality is you will be limited by the gain before feedback. The other thing that limits the gain before feedback is the number of open channels. You can buy yourself another 2 to 4 dB by riding the faders and only having one mic turned up all the way at a time. The unused mics don't need to be attenuated by more than 10dB for this to work. There are automixers that do this for you (Shure SCM810 or SCM820), and some digital consoles have this functionality built-in. (Yamaha TF and QL series). I'm a little biased towards the SCM820 since that was one of my projects at Shure. ======================== That said, if you really want to solve this problem as dictated by the laws of physics, then you need to individually mic each kid with a lapel or headset mic. Even then you're going to be limited by the gain before feedback - especially if they're not projecting. In the professional world, this is solved by using several wireless microphones. Ya, I work for a wireless mic company, but that is not biasing my recommendation here. You might be able to rent enough wireless gear for $500 to $1000 for this musical depending on how many channels you need. Figure $50 to $100 per mic per day.

Yikes, that's a horrible idea....but I think it'd require a lot of battery engineering experience to explain why. The Model S is a crap car though, so maybe the exploding batteries might do the world a favor by getting rid of these obnoxious cars and their owners, hah. I'm a big fan of swapping batteries, but standardizing on Lego pieces doesn't work. The real solutions for batteries won't be ready until 2025'ish based on the battery conferences I attend. Btw, autonomous doesn't require electric.

20log(Sd/St) Where Sd = driver area St = throat area Conceptually, the driver moves one unit volume of air for a given excursion. When that volume of air moves through a smaller opening, then the velocity increases proportionally (because the same total volume gets displaced). The job of the horn flare that comes afterwards is to maintain as much of that higher velocity all the way to the mouth - where the surface area is even greater, and you're displacing even more air. In order for this to work, the horn needs to be long enough to allow the pressure to build up into a proper wave. If the horn flares too fast, or is too short, then the air "rushes around the corner" of the mouth / horn wall and yields a large pressure change. That pressure change creates an opposite polarity wave that travels in all directions, including back into the horn. That's why the quarter wavelength horn length is such a magic number. This is a horrible explanation for the concept of impedance matching, but I think it's helpful from an empirical perspective. The compression ratio basically sets your starting point.

Hornresp is fairly accurate if you maintain the assumptions behind its equations. I use it as a best case first order sanity check. It's usually not too hard to calculate by hand when you're deviating from the ideal assumptions - I'll often manipulate the inputs to the tool to see the results of those deviations. You kind of have to put the full story together in your head after the fact....or just build it and measure. Hornresp is modelling the performance of a circular cross-section with the defined area expansion. Once you deviate from that, then hornresp is totally invalid. It's also assuming the wavefront follows the shape assumed by the exponential/conical/tractrix equations, which isn't reality either - even for circular horns. So from that perspective it's probably better to say that it's inaccurate for most systems. FEA (or BEM) is the way to go, but you're not going to get a nice interface like winISD or anything like that....and you can really muck things up if you don't get your meshes right. That or it will take weeks to solve the model.

There is a big difference between comparing designs and comparing a working circuit against a circuit that was damaged or built improperly. Circuits behave the way they're built, not how they're designed.

The reason it digs low is because the compression ratio is low. When the compression ratio is 1:1, then there is no acoustic gain. The driver excursion for a given SPL is the same. If the horn flare is narrow, then there is DI gain, but the total power is the same. This 12" driver has what, an 81 sq in Sd, and that's firing into a 49 sq in throat? That's a 2.2 dB max increase in power for an infinitely long horn. When the horn is shorter than infinity, then there is less gain at low frequencies. The shorter the horn, the higher that low frequency corner gets. What all that amounts to is a short flare horn like this is not much different than a direct radiator configuration, especially at the lower frequencies. This can be seen in hornresp if you setup a maxspl plot with very limited cone excursion and high power handling. 2.2dB at the top half of the passband isn't nothing, but it's a lot less than the 5.6dB gain over the full bandwidth of the Jubilee. That's why the Jubilee is so much longer. It starts with a smaller throat and loads to a much lower frequency. This isn't to detract from the design, but to point out why a small "horn" has a lower than expected low frequency corner. It's basically the direct radiator response of the drivers you're using.

I have grown to believe the circuit topology has more influence than the type of active device employed. That said, the best topology for SS is likely different than the best topology for tubes. So it'll never be apples to apples. And SS is such a wide field....we talking BJT, MOSFET, JFET? They all behave very very differently and have their own design considerations. Then there's what, triode, tentrode, and pentode tubes? I feel like it doesn't make a ton of sense for audiophiles to have these discussions because there are both good and bad SS and Tube designs. How do you know what you're hearing is due to the active element, or certain design decisions? The point is that there are both good and bad implementations of each. Sure, there are in fact inherent limitations or flavors to the different active elements that do impart a common character to the sound. I think it would make more sense to try and understand those flavors - and find ways to describe what we're experiencing. The different characters detract differently from different types of source material and listening habits. But to really compare SS and tubes, then you're going to get into a very in-depth discussion that ultimately revolves around circuit design....and we're simply not equipped to cover those nuances on a public forum. Even the best audio circuit designers in the world are still learning more about how these devices behave. All that to say, SS is the best. Tubes should die along with vinyl and wax cartridges. They are archaic relics of the past that serve only nostalgic purposes. Haha (said totally in jest btw)

Do you have a plot of the frequency response for each driver overlayed with the full speaker response? It would look a lot like the screenshots of your filter settings. This will give a good indication for how much summing you're getting at the xover frequency. One thing about phase is that it will change dramatically if you move the microphone around. Are you driving all three woofers in parallel? Didn't they do a tapered array in the passive xover? Are you sure you're not seeing polar lobing at 250Hz? Or is the mic within 14 inches of a reflective boundary? That could also be true of the speakers too. Is there a 28" path-length difference between the bottom and top woofer?

Btw, have you seen this website? http://www.data-bass.com/data?page=systems&col=2&type=1&sort=desc&mfr=-1 That's a direct link to sorting by CEA2010 distortion numbers over a 10-63Hz bandwidth.

There is an assumption that "horn loaded systems" have less driver excursion than direct radiator systems. It's a nice thought, but that simple remedy is simply not true across the board. You see this in both simulation and measurement world. This becomes the reality when the horns are undersized, which is the case for very low frequency devices. Also, if we're talking about "impedance matching" - which is one way to talk about horns, then you can't talk about "impedance matching" without first understanding the load you're trying to drive. The load in this case is the listening room - and specifically the wavefront that arrives at the listening position. The source in both cases is the driver. If you crank through the math, then you will find that a "classic horn" is not the best way to couple the wavefront to the listener at the listening position for a "typical" home environment, especially if you're dealing with an "undersized horn" in an acoustically "small room". A "classic horn" assumes radiation into free space. You get more power per displacement using other methods in a "small room". Not that this is an example of quality reproduction, but there's a reason you don't see horns in the car audio subwoofer SPL drag races. It's the same physics at play - just that the room is even smaller. I'm putting quotes around these terms because they are loosely defined concepts that can have an effect on these conclusions depending on the specific situation. Anyone claiming that a horn automatically has less modulation distortion is drinking kool-aid. Can it have less? Yes. Can it have more? Also yes. It depends on the design and implementation. The thing is - you're going to get more bang for the buck avoiding the classical horn route for very low frequencies. It's just how the physics play out. Anytime you make the horn bigger to give it more advantage, then you can always add more drivers and end up with a smaller total footprint that meets the same SPL requirements. And even if you make the horn bigger, it's the compression ratio that determines your best-case excursion reduction. Make the compression ratio higher and the horn has to get much much longer for the same pass-band ripple. It's more volume-efficient to use multiple drivers to achieve the same excursion per SPL at the lower frequencies (and you have a perfectly flat passband). This is why pro-sound touring companies use direct radiators for their subwoofers. At high frequencies, horns are always better. Our rooms are acoustically large at high frequencies (so the free space assumptions are valid) and we're getting very large compression ratios because there is enough space for the horn to be appropriately sized and couple well. At low frequencies, we have small rooms and the horn isn't big enough to get the energy out of the horn. The frequency response ripple in an undersized horn is the result of "room modes" inside the horn. Designs that are undersized but appear to have less ripple are converting energy to heat inside the horn (which is still energy not getting out). On top of that principal, inside a small room, the direct radiator gets coupling benefits from the room itself. That "room gain" provides more gain to the direct radiator than the horn firing into the same location. A lot of that impedance matching work of the horn is duplicated by the room - the thing is, you can't get a double benefit from impedance matching. Anyways, that's not to say that you can't get a good sounding horn loaded subwoofer. It's simply going to be a larger cabinet than a comparably performing direct radiating setup. And by comparable, I mean same bandwidth, spl, and excursion requirements.

Just to clarify, I was recommending to purchase the IB drivers, but don't install them in the walls or ceiling. You can put them in a classic speaker cabinet of your choice. Start with the drivers and then figure out which cabinet alignment meets your current needs best.

There is no science in the world backing up this aversion to direct radiating bass...especially for a narrow bandwidth subwoofer. There are measurements however that disagree with the claims presented thus far, which also lines up with the science. I'm not going to debate it though because these guys have drunk too much kool-aid. I have always been a proponent of audio purchases aligning with long-term goals. If you want to go the IB route in the future, then perhaps you might consider investing in the IB drivers now, and installing them in a slightly different cabinet alignment. That can be hard with the IB drivers, but would be something worth investigating I think. Plenty of guys on the AVS forum to help hash out the details. What kind of amplifier will you be running for the sub?

LTSpice is free btw - and also my simulator of choice: http://www.linear.com/designtools/software/#LTspice The tool is fairly straightforward, but adding parts is a major pain.

Interesting - where are they located? Hidden behind that grill going down the sides?

The JBL 2226H uses a 4" voice coil and extends beyond 1kHz... https://www.jblpro.com/pages/pub/components/2226.pdf That's what, a 25 year old design? I'm sure there are plenty more examples - I hope it's not poor form to mention a JBL driver here. Do they count as "low mass" drivers?

I was just thinking - if we're really using different words to describe the same thing, then perhaps we can throw some numbers on the table? When I hear low mass 15" driver in a sealed cabinet of that size, then I'm expecting an acoustic F3 somewhere between 50Hz and 70Hz, which is basically lascala territory. It's interesting that PWK noted 50Hz in that article @dwilawyer posted, so I'm probably not too far off. Is it too early to release numbers like that? Is it possible you've achieved better THD numbers at the expense of increased frequency modulation distortion? PWK had a lot to say on that subject...