sensitivity or frequency response for cabs?

[Calaverasgrande]

As for the notion that a speaker for bass should be subject to a different standard than one for PA or HiFi or whatever, that's the same as saying you should rate the gas mileage for a car based on a 5280 foot mile, for a truck for a 4000 foot mile, and a motorcycle for an 8000 foot mile. It's just plain wrong and only serves to confuse and deceive the consumer, be it intentional or not.

Not a different standard, different evaluation criteria. After all sound reinforcement gave up on tube amps in the 60's whereas tube amps are still quite popular in bass playing circles. Bass cabinets are not just about bass reinforcement! Heck if all we were concerned about was fidelity we would all be using something other than magnet/coil pickups followed by tantalum capacitor eq!
Bass cabinets are not just playback devices, they are part of the tone of the players rig.
If you think you can get any tone with a nominally flat speaker, try replicating a marshall stack with a flat(ish) 3 way speaker and a JCM800. Yes some of us like some fudge in our tone. It's what makes it sound loud.

To use the vehicle metaphor: it is nice to know what kind of mileage different light pickups and station wagons (4x10's and 2x12's) get and then only test drive the ones that get at least 35 freeway (have high spl per watt). But also keep an eye out for ones that have abysmal city mileage (anemic low end) no matter how great the freeway.
Then pick for ride/handling (dispersion and tone).

 

[agedhorse]

Hmm, based on that chart I'd ball park you're looking at 96db up to almost 2khz. It's a little questionable in my opinion to average 2khz+ data with 50-2khz data, and say that 103b @ 2khz averaging with 95db @ 400hz makes a 98db average cabinet.

Look at the area under the curve... ie. integrate the amplitude over frequency. I also feel that for most bass players, a bandwidth up to about 5kHz (always subject to a player's personal tastes) is pretty representative of the overall spectrum that is of particular interest. That's why so much effort is placed on developing "user-friendly" voicing in the midrange. There's a lot of difference in tastes between manufacturers regarding this reigon and that gioves players a lot of very viable options from which to choose.

My thinking here is based on some anomalies in the specs though; if you average it out to 98db and then say the -3db point is where it hits 95db, your F3 is going to be really really high. If you say the driver's "real" sensitivity (pistonic) is 95.72db then look for the -3db point (92.72db) it's going to be another thing.

Where it sounds like the confusing is happening is you're saying the -3db point is @ -92.72db, but 98db is your average sensitivity. Kinda makes it confusing (hence my suggestion of a separate LF and HF sensitivity number).

There are two seperate things going on here, there's the low frequency response which for practical purposes is up to about 200Hz (again, this could be chosen to be 150Hz, 250Hz or whatever but it represents the region well beyond the effects of the box tuning) and the >200Hz region where the direct radiating effects of the driver alonmg with the response modifications that mid voicing affect. They are two seperate things that ultimately must agree where they meet.

In the low frequency model, the transition point to "flat" is <200Hz so at 200Hz that's the nominal sensitivity of the LF model and average this with whatever bump below (say 1dB) to the ~97dB is pretty close to 98dB. Take the 97dB datum and subtract 3dB and you have your -3db point which is 94dB.

In the >200Hz model (actual measurements provided), a simple bisection for equal areas under and above the curve gives the average SPL for this band which is >98dB.

Splicing these curves together and taking the equal areas above and below the curve also yield an average SPL of ~98dB.

A note about using two different approaches by frequency spectrum, this is similar to what we do when developing a driver and look at small signal versus large signal parameters (2 different mechanisms at play for what looks like the same specifications), or when designing amplifiers where we also look at small signal versus large signal versus overload signals. Each mechanism is different and each returns different data describing the particular process being studied. For amplifiers, we will also run burst tone tests, impulse response tests and look at the output results as they also tell us very important (in some cases far more critical) information about what is happening under the hood of a design. Also, while an amp's frequency response may be say 10Hz-80kHz +/-0.5dB, additional filters and compensation may be employed to limit the frequency response to say 20-20kHz to make the product more robust to faults, poor crossover designs (for example) and generally the frequency response that is reported is the wider bandwidth because that's what the real bandwidth is. An amplifier's performance is the sum of all of these results taken in the context of the particular tests and that they mean. You also have to trust that the designers who have a long history of successful products really do understand what they are doing.

 

[agedhorse]

rp, be sure to set the external series resistance in the simulator to something around 0.01 ohms, for some reason the default was 0.1 ohms which will affect the measurements. That's more typical of the internal wiring and connector. This may be where part of the small difference you noted comes from. Just a thought.

 

[fdeck]

I looked at Winisd and it does do a 1/2-space default and modeling with my data in Winsid 1/2-space shows 96.7dB/1W/1M for the low frequency component, and averaging the spectrum based on my data above 200Hz - 5kHz I get just about 99.1dB/1W/1M so again, it seems to me to be pretty darn close. I show an F3 of 46.8Hz and a -10dB point of 34Hz. Don't forget that we are using a cusom driver, taking advantage of optimizing the driver and box for the smaller size.

The transfer curve was very close between simulations too so that adds confidence to the results, or at least the math.

Note that simulations do not provide any meaningful data regarding information above say 200-300Hz, there are major variables that come into play that govern things above this reigon.

I think there's an issue with how WinISD computes the input voltage from the "nominal" input power. This would also result in an identical transfer curve with a constant offset in dB. I was able to reconcile my program with WinISD exactly when I accounted for this difference.

That's a reason why I would advocate an open source program, so that any expert could look in and critique the math.

 

[agedhorse]

I think there's an issue with how WinISD computes the input voltage from the "nominal" input power. This would also result in an identical transfer curve with a constant offset in dB. I was able to reconcile my program with WinISD exactly when I accounted for this difference.

That's a reason why I would advocate an open source program, so that any expert could look in and critique the math.

I understood that some corrections were made regarding this in the new version, but in the big picture it really is pretty darn close. My understanding is that there is a voltage correction function that normalizes power to voltage based on the driver's calculated impedance curve. I do not know how accurate this is.

 

[billfitzmaurice]

Not a different standard, different evaluation criteria.

Yes, a different standard. To sum it up G-B advertises performance that cannot be achieved according to the standard of loudspeaker testing that has existed for decades. Rather than admit the error of their ways they, or agedhorse as their representative, insists on using measurement protocols that don't adhere to the standard. When you do that there is no standard, it's everyman for himself.

 

[mulchor]

Yes, a different standard. To sum it up G-B advertises performance that cannot be achieved according to the standard of loudspeaker testing that has existed for decades. Rather than admit the error of their ways they, or agedhorse as their representative, insists on using measurement protocols that don't adhere to the standard. When you do that there is no standard, it's everyman for himself.

Well, he came and he explained his numbers. They seem at least reproducible. Definitely useful from a design point of view.

There's other companies. Carvin claims 40 lbs weight in its advertising, but 50-52 lbs in reality. Perhaps the advertised weight is without drivers. Or on Mars.

Acme claims +/-3 dB 41Hz to 22 kHz -6 dB at 31 Hz, for a 1x10. Is that reasonable?

This is an applied tech industry that's not about good empirical measurements, but rather about using the technology creatively. It's not like doing clinical trials where you submit all your calculations for review before you can sell your product.

Honestly, I think beating on agedhorse, when he's come and shared his methods & his thinking in good faith... it seems rather pissy.

Put another way, I've not seen *anyone* show numbers or measurements in a way that I would pass if I were on their thesis or dissertation committee. But that doesn't mean they don't deserve their degrees, eh.

 

[RickenBoogie]

Ok, I don't know squat about the science of loudspeaker design, but I do know that when I was testing, (with my ears) a bunch of 2x12 cabs for use with a Walkabout head, GB's Uber stood out by a long shot over all the competition. I don't think a bunch of numbers and charts, standardized or not, really do much good to the "average" bass player when shopping for speaker cabs. Plugging in and listening is still the only method most of us use, and the rest go by word of mouth, or shop by brand- as evidenced by all the threads asking "which cab for amp X". Just sayin'. Procede with your science.

 

[georgestrings]

, i believe everything andy says. they're having a dispute now, but andy is a very honest dude, and this dispute doesn't change that at all for me.

Same here...



- georgestrings

 

[jnewmark]

Ok, I don't know squat about the science of loudspeaker design, but I do know that when I was testing, (with my ears) a bunch of 2x12 cabs for use with a Walkabout head, GB's Uber stood out by a long shot over all the competition. I don't think a bunch of numbers and charts, standardized or not, really do much good to the "average" bass player when shopping for speaker cabs. Plugging in and listening is still the only method most of us use, and the rest go by word of mouth, or shop by brand- as evidenced by all the threads asking "which cab for amp X". Just sayin'. Procede with your science.

Probably the way it is for most of us, me included. OTOH, if you knew you needed a certain performance out of a cab, and checked out the specs and bought the cab based on those specs , then it failed on you because it could'nt deliver said specs, I think I would be pissed, and would be asking some pointed questions about false advertising. Just sayin' .

 

[fdeck]

I understood that some corrections were made regarding this in the new version, but in the big picture it really is pretty darn close. My understanding is that there is a voltage correction function that normalizes power to voltage based on the driver's calculated impedance curve. I do not know how accurate this is.

Aha. I've got an ancient version.

Confusion over the conversion from power to voltage is one reason why I prefer modeling speakers at a given voltage input rather than a given power input. It's also more faithful to how amplifiers behave.

As we say at my day job, "in your ample spare time," you're welcome to try my program. Its main particulars are that it's open source, and includes some progress towards modeling cone excursion for non-sinusoidal input signals.

 

[billfitzmaurice]

Carvin claims 40 lbs weight in its advertising, but 50-52 lbs in reality. Perhaps the advertised weight is without drivers. Or on Mars.

Good example. Maybe Carvin decided that 16 ounces to the pound didn't suit them and that 20 ounces to the pound fit their ad copy better? That would be a similar situation to averaging in above 200Hz results to a sensitivity spec. When the rules don't favor your game simply change the rules?

Acme claims +/-3 dB 41Hz to 22 kHz -6 dB at 31 Hz, for a 1x10. Is that reasonable?

It depends on sensitivity. With 90dB sensitivity it's entirely possible. With 95dB sensitivity it is not. You can think of the relation of the -6dB point and sensitivity as relatively constant. When sensitivity goes higher so does the -6dB frequency. When sensitivity goes lower so can the -6dB frequency.

Honestly, I think beating on agedhorse, when he's come and shared his methods & his thinking in good faith... it seems rather pissy..

I'm merely pointing out for the benefit of those who don't know how sensitivity is calculated that agedhorse's methods are his own, not those of the professional audio community, so his results will not be the same as those who do follow the rules. Therefore an objective comparison between the sensitivity and response claims made for G-B cabs and those of other manufacture cannot be made.

 

[agedhorse]

It should be pointed out that Bill has been ignoring driver efficiency in his comments regarding sensitivities. There is a wide range of raw driver efficiencies (efficiency is the conversion of electrical energy to mechanical motion) that can range from less than 1% to as high as about 8%. This efficiency, which also affcets average speaker sensitivity, is governed by variables such as wire (turns and cros-sectional area) residing in the gap, the size and shape of the magnetic field (especially at it's fringes), the useable field strength, the losses in the magnetic circuit, clearances, the effects of back EMF on the magnetic field, etc. Typically, "cheap" drivers have a lot more losses compared with more expensive drivers, so while the TS parameters governing response is important, so are other aspects of driver design in the overall performance. This information is (or should be) used when selecting a appropriate driver for the application and pricepoint. Unfortunately, the market often places cost as the number one criteria.

I came here to engage in a healthy discussion regarding which specifications are important to the average player. This is a bass player forum, not a pro audio forum nor a DIY forum. The products are intended for bass players who have their own particular set of needs that are distinctly different than those of the pro audio world. I have presented in depth how we develop our specifications and most importantly why we do it in a way that relates directly to this specific market. I have worked as an electrical engineer in this industry for almost 30 years, for different companies, designing products for different market segments including pro audio, bass, guitar, acoustic guitar, and cabling/interconnect. I design speaker products, amplifier products and signal processing products, so I have a deep understanding of the big picture by knowing exactly how each piece works in agonizing detail. I engage in technical discussions with other industry design professionals all the time, but without the somewhat disrespectful comments and accusations that have been made here. IMO, this discourages other designers and engineers from participating in forums such at this which is your collective loss. There are some very creative, skilled and well educated designers out there (some designing competative products to ours) who you guys would also benefit from if they were to participate in discussions on topics like this, but not without a little bit more respect and common courtesy. While accusations and pissing matches may be the norm in DIY forums and the DIY community, it is definately not the norm in the professional engineering community.

 

[agedhorse]

Aha. I've got an ancient version.

Confusion over the conversion from power to voltage is one reason why I prefer modeling speakers at a given voltage input rather than a given power input. It's also more faithful to how amplifiers behave.

As we say at my day job, "in your ample spare time," you're welcome to try my program. Its main particulars are that it's open source, and includes some progress towards modeling cone excursion for non-sinusoidal input signals.

Since an amplifier is (idealy) a constant voltage source, this does indeed relate more accurately in the modeling. While there would be an error between using a constant voltage versus power, I suspect that over the model bandwidth (certainly not at many individual frequency slices) would revert back to a reasonably close 1 watt though assuming the average impedance was indeed close to 8 ohms.

What would change is the frequency response chart which would reflect the conversion frequency by frequency. I will have to think about the benefits and drawbacks of each method more closely as one method may be better for one type of application, or give more pertenant data.

All this focusing on sinusoidal signals is also not representative of the real world, where asymmetry and distinctly non-sinusoidal signals rule the day. Of course the signals can be reprtesented as a series of sinusoidal fuctions, but it comes back to how much information do we really need and what it the benefit it provides.

I will take a look at your program when I get a chance... as you say, ample free time is never ample enough

 

[thudfromafar]

Fdeck, I haven't been able to use your program due to a run-time error

 

[billfitzmaurice]

It should be pointed out that Bill has been ignoring driver efficiency in his comments regarding sensitivities. There is a wide range of raw driver efficiencies (efficiency is the conversion of electrical energy to mechanical motion) that can range from less than 1% to as high as about 8%. .

Driver efficiency does encompass a very wide range. But as driver efficiency goes up low frequency response goes down. That's why drivers like JBLs and EVMs that have very high sensitivities also have weak low end response. There's never any such thing as a free lunch, and where driver efficiency is concerned the cost of higher efficiency is loss of bass response. Conversely when you sacrifice overall efficiency you can get better low end response. That's exactly the case with Acme cabs, as noted a post or two back. You can have response that's flat and goes deep with lower efficiency, or you can have response that's peaky and doesn't go deep with higher efficiency. You cannot have flat response that goes deep with higher efficiency at the same time. Claims that one can achieve greater sensitivity without a corresponding loss of bass response take us back into the realm of non-existent Magic Drivers.

Those of you who want to test this for yourself can do so with WinISD. Higher efficiency is accomplished with lower Qes, higher Bl and higher EBP. See for yourself what happens when you model similar drivers but with low Qes/highBl/high EBP versus high Qes/low Bl/low EBP. You'll see the former has higher sensitivity in the mid-bass and mids, but lower sensitivity in the sub-bass and bass.

 

[agedhorse]

Then why would the marketplace or a designer choose a $500 driver over a $50 driver?

 

[Kindness]

You're a class act, Andy.

In my opinion, choosing which specs to test and publish within the bass amp/cabinet realm are a judgment call. Bill clearly has an opinion that there is one set of official measurement techniques that are the only valid techniques - that's understandable and has clear inherent advantages. However, Andy has, to me, clearly demonstrated that there are logical and meaningful reasons to take a different path.

Since there is absolutely no possibility of validly comparing cabinet specs across manufacturers, I am 100% satisfied with a manufacturer making a judgment call and then taking the time to explain how and why he arrived at the numbers he has. That's already more than a couple of steps ahead of the game.

I demand the companies I support to be concerned about quality, value, ethics and customer service. By these metrics, I can't think of a better company than Genz-Benz.

 

[rpsands]

Then why would the marketplace or a designer choose a $500 driver over a $50 driver?

I think Bill is talking more about the realm of the top end of drivers. Obviously there are drivers out there that sacrifice everything for cheapness.

At the top end from what I can tell, you get what you pay for up to a point -- for 12" drivers, 150 bucks gets you about as good as it gets. Then you drop 300 on a boutique driver that just makes different compromises.

Why anyone would choose a 400 dollar Beyma driver over a 150 dollar eminence driver that meets or exceeds its specs, I don't know for sure. It almost surely comes down to those intangibles you talked about earlier; kinda like why some people still run those old heavy JBLs (E140s and so on). Different midrange character is obviously as you've said a big decision point in drivers.

This discussion hasn't really encompassed why some speakers just sound different and so on, and digging into that is where all the intangibles get started.

 

[JimmyM]

This discussion hasn't really encompassed why some speakers just sound different and so on, and digging into that is where all the intangibles get started.

and that to me is the big flaw in this argument that pretty well renders it moot. at least that's the way i see it as a non-engineer type looking in.