+1. Quite a well-spoken criticism of the "religious beliefs" (and shibboleths) of the local wannabes-engineer sim-friendly crowd. Umm, good thing you didn't get into more alienation by diss'ing "tapped horns" - criticism of TH is not tolerated at DIYaudio.
But given sumaudioguy's and Planet 10's critique, the most sensible reaction would be to abandon unjustified faith in trick "higher-order" boxes that depend on assumptions about assumptions to work right?
While the elements of the Olson/Beranek model still determine the sound of a given driver, there is only one design decision for sealed boxes (and comparably for open boxes). That is, how big a box do I need for that driver to keep from shifting my resonance beyond what I can live with?
Overall, a great misplacement of effort to fuss about boxes - and really laughable when folks dream of miraculous benefits from sixth-order complex constructions. As far as I'm concerned (and since starting to research it long ago), if you really care about quality bass sound (and know which is the hot end of a soldering iron), the path forward is motional feedback.
Question for sumaudioguy: what causes the divergence of the two markers of resonance. Shouldn't they be the same (at least in free-air), like we always thought?
Ben
But given sumaudioguy's and Planet 10's critique, the most sensible reaction would be to abandon unjustified faith in trick "higher-order" boxes that depend on assumptions about assumptions to work right?
While the elements of the Olson/Beranek model still determine the sound of a given driver, there is only one design decision for sealed boxes (and comparably for open boxes). That is, how big a box do I need for that driver to keep from shifting my resonance beyond what I can live with?
Overall, a great misplacement of effort to fuss about boxes - and really laughable when folks dream of miraculous benefits from sixth-order complex constructions. As far as I'm concerned (and since starting to research it long ago), if you really care about quality bass sound (and know which is the hot end of a soldering iron), the path forward is motional feedback.
Question for sumaudioguy: what causes the divergence of the two markers of resonance. Shouldn't they be the same (at least in free-air), like we always thought?
Ben
Last edited:
if you measure T/S in free air sure i'll take a bite from this apple... but useless ?
but what is the reality? DIYers often take the T/S data from what the seller claims without understanding tolerances at all. If you have incoming inspection QA measure in a sealed box ~1/3 to 1/2 Vas you should be golden either way. also since frequencies rise in boxes , accuracy of instruments should get better. IDK we could talk about inaccuracies of old Vernier dialed instruments had in the 10's of hertz scales?
Last edited:
You poo-poo T/S theory and then you bring up alignments? If this is so easily provable, then prove it. After all, you seem to think you are a genius... My guess is you will prove nothing but that B4 is maximally flat....and isn't that just an amazing result, seeing that is what it is called? 😉
Qt=0.3827... only arises from an oversimplified version of the vented box model from 1961. A model with damping terms on box, port and leakage cannot even be represented by a butterworth polynomial.
By derisively/condescendingly stating my name multiple times you are not proving anything...
Criticism of everything (including tapped horns) is very welcome. But you have to know what you are talking about. Since you don't understand tapped horns and have never even heard one your criticisms are not particularly welcome as they are almost universally incorrect. Besides, your only criticism is your dogmatic belief that resonances are evil.
Sumaudioguy's critique is mostly incorrect. Planet10's comments have to do with a mysterious thermally induced shifting of t/s parameters that sounds exactly like mild power compression but apparently is not power compression. If you take a look at compression measurements like the ones done by data-bass you will see that there is no dramatic (3+ db) shifting of frequency response curves until you approach the rated power handling of a driver. And this is not a problem for ported boxes alone, even sealed boxes exhibit the same frequency response shifting behavior at high power.
High order designs, even massively undersized ones work just fine and all you need to design them is a simple simulator and a bit of common sense. In fact your beloved Klipshhorn is an example of a pretty bad horn by modern standards. Many of the horns (including tapped horns) being designed on a regular basis by members of this forum are much better than your antique Klipshhorn with it's ragged resonance riddled response.
That is a uniquely simple minded approach to sealed box design. There are a multitude of considerations even for a simple sealed box design. What do you want the frequency response to look like? Will the frequency response work in the environment it will be used in? How much space do you have available? What will the thermal considerations look like (how much power to reach xmax)? I could go on (and on and on) because there's a lot of things to consider that you apparently know nothing about.
And the fact that you don't understand the simplest things like qtc (based on your "whomp up" comment) means you can't even begin to make reasonable design decisions.
As I told you a few times now, motional feedback is the expensive and not particularly effective route. Instead, buy quality drivers that don't have high distortion to begin with. Use distortion reducing mounting techniques (like push pull) to reduce distortion even further if desired. Use double (or more) the amount of drivers, since you can afford them because you aren't using expensive motional feedback garbage. You will have dramatically measurably lower distortion at any given spl level with this approach compared to motional feedback. And this approach will get you a system that has much higher max spl as well.
Your proposed "ideal" subwoofer is useless garbage. You advocate a sealed box with a driver with the lowest fs possible (like your antique AR-1) so you don't ever have to use it at or below fs. And motional feedback. This will result in a very expensive low sensitivity, high distortion design and unless you have several of them the spl will never be as high and the distortion will never be as low as a single simple inexpensive well designed ported box.
Last edited:
I don't think Dave said "thermal," though I may have missed it. It's pretty easy to measure the TS parameters at (say) 0.1, 1, and 10W and see if they are different.
Sy's comment is well taken, as usuall. But real question is what lowest level.
A lot of attention is devoted to speaker performance maxima and peaks*. But what is ignored is "where you live" in terms of performance. While challenging to measure let alone to standardize, I think we need to know how primitive Rice-Kellogg drivers work at say, 40 dB below regular home listening levels, just as want to know about amp residual noise levels.
Drivers pass through the small-signal range on EVERY cycle but that's just a tiny part of their sound on loud sounds. But most of the time, and esp. for subs, they are barely moving at all. That's where cleanliness matters**.
At those levels, I bet it turns out Rice-Kellogg drivers have some kinds of unlinear suspension flaws akin to stiction or tin-can-top (I forget the name) or accordination (I made that up) that isn't in the T/S model.
Ben
*There is one particularly noxious character who is forever fretting about what alignments produce max power, as if anybody wanted 110dB in their homes. Poor Guy, probably lives in his parents basement and can only play loud when they are out.
**And it may turn out that the folks who like smaller "faster" subs are right all along because small subs spend less time in smaller motions, Doppler distortion aside.
Last edited:
No he did not mention thermal at all but he did say it would show an effect on dynamics and when the volume knob is turned up. This would seem to imply a thermal effect.
T/s will definitely measure different at levels between 0.1 and 10 watts, but as it has been pointed out (most recently that I've seen by Jeff Bagby) at these moderate power levels the t/s shift in complimentary ways that show the same response curve in a given box despite the differing t/s. It's not until you get into thermal compression that you start to get a dramatically shifting (+3 db) difference in frequency response.
This is the subwoofer forum and Dave likes the SDX7 if I recall correctly. At 10 watt levels there's no way that this driver should be drifting in tuning or having any significant response curve differences compared to milliwatt power levels in any type of enclosure including ported boxes.
Dave brought up a lot of terms and phrases that I am simply not familiar with outside of the concept of power compression. "Tightly tuned", "the box moves in and out of tune", "horizontal t/s curves". You can't get a much more widely tuned port than a horn which is resonant over as much as 3 octaves, and yet horns experience this same type of t/s parameter shift at high power and the frequency response curve shifts just as much as a ported box does as t/s shift. This is because the wide tuning of a horn is simply due to MORE impedance peaks, not wider impedance peaks. But the response curve doesn't shift much (if at all) until you get into thermal power compression territory.
And like I've mentioned a few times now, this is not unique to ported boxes, if the t/s shift enough to affect the frequency response curve it doesn't matter what type of box the driver is in, even a sealed box will show a difference in the response curve. This is all seen pretty clearly in compression sweeps, data-bass has a large collection of them (although not many ported examples but there are a couple), there are a large collection of sealed and a few horn examples.
Last edited:
Once you are out of the millivolt range, you aren't measuring T/S any more...the entire definition is based on small signal. T/S also ignores inductance, which makes your impedance peak asymmetric and introduces measurement errors. One way to minimize is to use sqrt(Fl*Fh) instead of Fs in the Qms calculation...
Of course there are some flaws that the t/s parameters don't reflect. That's what Klippel type testing (and Dumax and high power compression sweeps, etc) are for.
This stuff is all accounted for in various types of testing and distortion measurements.
Put your antique AR-1 subwoofer (which you consider to be near the pinnacle of subwoofer engineering) up against a design like the Keystone or Xoc1 18 tapped horn in head to head distortion measurements. Your AR-1 should be the easy winner right? The Keystone and Xoc1 design are both tapped horns with resonances all through their passbands and they both use 18 inch drivers so they must be garbage, right? At any given spl level the tapped horns will easily dominate in low distortion tests compared to your beloved AR-1, and the tapped horns will keep going at high power levels long after your AR-1 has choked at 100 percent distortion and thermally failed.
You can continue to talk at length about nonsense like "primitive" moving coil drivers (although you use them yourself), the evils of resonances (even though you loved the Klipshhorn which is a particularly fantastic examples of ragged resonant response) and use made up terms (like accordination, whomp up, bonkers) that have no meaning in an attempt to describe things that don't exist in an attempt to "prove" your incorrect theories but simple distortion measurements and BLIND listening tests will show that all your nonsense is completely unfounded.
To address this part specifically, max spl DOES matter. It has implications that apparently elude you.
Take your OB for example. You have an antique woofer with maybe 2 mm xmax on an acoustically small (when used down to 20 hz) baffle.
You refuse to provide any info on the driver like t/s parameters or baffle dimensions so it can be analyzed in a meaningful way except to say that it will NECESSARILY be a very low power, high distortion design because it makes no sense at all. Just because you apparently listen at very low spl levels and are happy with this type of ridiculous design doesn't mean you know things that the rest of us don't.
Let's also look at the measurement you just posted, which implies a low max spl esl panel with incredibly high distortion. Yes, max spl does matter, it implies things about other aspects of the system in addition to pointing out the performance brackets of the system.
FWIW I do not live in my parent's basement and my nearest neighbor is at least 1/2 mile away so I have no restriction on max spl at any time of day or night.
It is a bit amusing though that you constantly talk ABOUT me (noxious, poor guy, sad fellow, probably lives in his parent's basement) to the point of dedicating your signature line to me; but you don't have the technical skills to actually have a discussion with me and you seem afraid to actually address me personally except through these taunts and adding +1 to other people's posts.
Last edited:
so what some guy measures some anomalies on drivers and damns the whole modeling principles. I say bring on the data and lets remeasure it another way. I don't doubt his data just his interpretation /conclusions.
FWIW measuring stuff at lower frequencies isn't easy, for example Peerless recommends baffle mounted measurements on some of their older acoustic suspension stuff, so they gave two values of Fs.
FWIW measuring stuff at lower frequencies isn't easy, for example Peerless recommends baffle mounted measurements on some of their older acoustic suspension stuff, so they gave two values of Fs.
Last edited:
Dave did not say thermal. Here is more including a measured example:
http://www.diyaudio.com/forums/soft...rements-book-win-mac-linux-4.html#post4657538
dave
Have any of the gents discussing Thiele/Small parameters stopped to look at the original papers?
In them you will find that the signals and parameters measured with them are only true for small signal measurements.
Even back in the Flintstones age the change in a driver's functional parameters relating to impedance change were well documented.
You can even go as far back as some of the General Electric research papers from the 30 's and see evidence that in the stone age they had a grasp on what was happening. Just not a worked out set of equations that could be used as well as the Thiele and Small contribution.
And do I mean contribution. As in something we should be grateful for.
Without their work we would be having quite a bit of fun trying to figure out what would be happening when designing loudspeakers.
Olsen and Benson were getting close. But not quite close enough.
In them you will find that the signals and parameters measured with them are only true for small signal measurements.
Even back in the Flintstones age the change in a driver's functional parameters relating to impedance change were well documented.
You can even go as far back as some of the General Electric research papers from the 30 's and see evidence that in the stone age they had a grasp on what was happening. Just not a worked out set of equations that could be used as well as the Thiele and Small contribution.
And do I mean contribution. As in something we should be grateful for.
Without their work we would be having quite a bit of fun trying to figure out what would be happening when designing loudspeakers.
Olsen and Benson were getting close. But not quite close enough.
Or down.
SDX7 is not really a subwoofer, but a nice midBass. We use it in sealed boxes.
dave
^this
the same data used to calculate Q (-3dB points) of the measured response, then take the geometric mean for Fs.
The observed peak isn't always the natural frequency!
I'd like to see a couple of sets of t/s from a single driver to show this shift in parameters. i'd like to sim both sets of parameters in a box and see how different the frequency response is. It won't be much different at all.
I would also like to know what you mean by "our boxes typically use bass tuning that is tolerant to changes in T/S", because if the t/s shift it will affect any box design.
Have you stopped to look at compression measurements ranging from very low power level to ragged edge of destruction high power level?
If you had you would see that the small signal model is quite effective at predicting the response curve right up to very high power levels. It isn't until you approach and exceed the power rating of the driver that you start to get significant (+3 db) changes in the frequency response curve.
Because it's not what I asked for.
You have Mark's measurement (Linear X), Factory (I assume that means published spec), and "My ave measure" (I assume that means the average of all the drivers you tested).
First of all, these 3 categories are all from different drivers (in the case of the averaged category from as many as hundreds of different drivers).
Second, you can't just average each individual parameter and expect the end result to be representative of an average driver. You might end up with specs that aren't even possible (conflict with other parameters) if you average each individual parameter and use these averages to sim the driver.
What I am looking for is two sets of measurements taken in quick succession from ONE SINGLE DRIVER that show this shift in t/s parameters at low to moderate power levels that you are talking about.
I'm quite certain I would get different results if I used the parameters in these three categories in your link, they come from completely different drivers, perhaps hundreds of different drivers in your ave category, use averaging of individual specs and were measured with different equipment, probably with different procedures, maybe at different drive levels, at different times with different barometric pressures, different temperatures, maybe even different elevation. While a lot of these considerations won't make a huge difference, using completely different drivers (which have a normal variance in consistency as you mentioned), averaging individual parameters and stuff like that make for a completely meaningless comparison between different drivers, not a meaningful comparison of ONE driver's shifting t/s parameters.
Last edited:
My average was for a pair of very closely matched drivers (so drivers have essentially the same measured parameters), Mark's are for 1 of the same pair, factory is for reference, i don't know how many they use to create that, you can ignore that row.
So your objections are moot.
dave
- Status
- Not open for further replies.