I read somewhere that JBL had to use his life insurance after he passed away to save his company from debt and about 5 ~ 10% were left to his family. Now it is a traded company sold to the highest bidder (Samsung) - that needs to provide / maximize share holder profit every quarter.
Been in the business of small things already, seen enough customer trouble and disputes, suffering "managerism" plague and corporate nonsense. Can't do. I stay in my lane of gear, engineering, hobbies and such. Though if anyone is keen on followup, I will happily help anyone with building ones own kit. But first, the work. Still quite few weeks to go. Borrowing the 21DS115 etc.
You mostly need just a LDS and a USB DAQ.
Plus a simple high power amplifier (can be bought 2nd hand for very cheap).
The LDS I got for this was around 300 bucks delivered and has plenty of resolution.
Capable and affordable USB DAQ's will show up here and there on ebay.
I have already seen somebody else on a different forum having all the measurements done with exactly this system.
I have been actually working on a DAQ to specifically do all of this, and that price point won't be in your ballpark at all.
People most certainly don't need a scope.
True!
In other words - If I make something out of it, and confirm different methods give adequate results, and community be happy with new tools for measurements, you can feel free to build on that. I have no business intentions in this. We can make agreements that I can reap followup windfall findings/data with no rights reserved against me, and you can sell whatever you want. How about that. 🙂
That´s already grey territory, as that´s not how tempting offering works - to tell the customer "go find your own amp and if something doesn´t work out, too bad." That is usually customer nightmare.
In other words - If I make something out of it, and confirm different methods give adequate results, and community be happy with new tools for measurements, you can feel free to build on that. I have no business intentions in this. We can make agreements that I can reap followup windfall findings/data with no rights reserved against me, and you can sell whatever you want. How about that. 🙂
Enclosure requires skill, difficult to fully robotize (who will construct when the present generation retire - Gen Z ?) - difficult to build, execute and handle.
With unlimited power and improvement in Class D & H + DSP - Enclosure will play a back seat - High Power Drivers (looks like everyone has cracked the code - add 3 magnets and overpower / over engineer the drivers parts) + DSP & Hi-Power Amp - will be able to achieve any response - w.o. dependence on intricate enclosure examples -
HiFi -
https://soundstagesimplifi.com/index.php/equipment-reviews/191-kef-kc62-powered-subwoofer
10" x 10" x 10" , -3db @ 11hz , 105 db
Pro Audio -
https://www.bassboss.com/catalog/kraken-mk3-quad-21-powered-subwoofer
Highest SPL in smallest form factor.
I am very big fan of that approach myself, but hat only works for "rider class" gigs (in case of PA where you can afford to have serious requirements, and all is paid for). I had been "forced" to run the PA from one mains socket extended by 150ft far too many times. That is one part of why I am building my subs to run around its impedance peak. It might not reach the highest SPL peak, but low power compression and high efficiency will secure that four 18"s could be fed fairly from good single mains circuit.
There is market for almost any needs fortunately.
Who will build it? AI robots. 😂🤭
There is market for almost any needs fortunately.
Who will build it? AI robots. 😂🤭
Hello there, I happen to have a lot of experience developing and testing some underhung motor platforms. I use an in house klippel ka3 to validate and there are some things I'd like to share about your simulations there that I have learned from my journey along the same path.
In all of the simulated BL curves of my underhung designs, there seems to be a rearward bias in FEMM results that is actually not there in real life. I tried tapering the pole around .015" smaller in diameter from the top of the gap to the bottom, and this will let FEMM spit out a flat BL curve, but after testing a prototype this way that is not the case. If you flip the polarity of your magnet 180 in FEMM and change the current of your VC circuit to -1, it will bias the BL towards the outward stroke.
Also, the chamfers on the corner of the top plate won't alleviate the "cat ear" B plot; that does not translate to real world results. Undercutting the pole for the gap height helps here, and with a flared taper at the top of the pole ID instead of just a radius you do not need an extended pole. Flux is less likely to stray/modulate from steel the more saturated it is, so saturating your pole fully at the gap is a good way to control the fringe field.
Here is a cutaway of one of my early designs with the FEMM result overlaid transparent on one half to show the saturation at the pole that I am talking about. In this exact design, FEMM showed a rear bias on the BL curve but the klippel measured it perfectly flat and symmetric.
So most people don't understand what the red part of the graph is on a BL curve and I think you are misinterpreting it since your X axis marker is pointed at the start of the red section in the first graph.
When you set up a klippel LSI test, you have protections that you set in place. Basically with these protections, you are telling the klippel "I do not want you to move the driver past this protection amount". The protections offered are: voice coil temp delta above ambient, BL, CMS, and power. You can set these to whatever you want. Say I have a driver that could bottom out easily, I will set the BL protection to 80% so the klippel will not move the driver past 80% BL. Whichever protection you set that the driver meets first, the machine does not move the driver past it. So, say I set CMS protection at 50% and BL protection at 80%, once the BL reaches 80% it will not move the driver further even if the CMS has not resolved to 50%.
The legend on the BL graph denotes "Xprot" in red; this means that the red part of the graph is just how much the driver moves past the protection limit that you set at the beginning of the test. Obviously the machine is not 100% precise at stopping the driver from moving past the protection you set, but that's why you see the red part of the graph.
What has likely happened, is the BL protection on the LSI run for the 18 was set more aggressive down to say 50% BL, while the 21 could have been set more conservative at 70%. For those who don't know how to read klippel graphs, this can be confusing, but I'm here to clear this up for you. When you get a full klippel report generated, there is a section called "table states" where you can view exactly what the protection amounts for each variable are set at and the amount that the test actually resolved to.
As for the slight asymmetry on the 21 curve, that just looks like the coil on that particular driver that was tested wasn't centered exactly and was built just a bit higher than the gap center. If you take the graph and scoot it to the right a little, it will bring back its symmetry. This amount is denoted in the "bl asymmetry" field of the "nonlinear parameters" section of a generated klippel report.
Last edited:
Heya. Thank you for education, sincerely.
I did not make any assumptions on this and the red line crossing is "accidental". It was precisely picked for the reason that on the 18DS115 driver, at 13mm of excursion, the far end of the voicecoil exactly meets the pole piece thickness, while the 21DS115 has its coil out by 2mm already. There is no side intentions and in between the lines again, these assumptions are moot. I am glad you poke to it so we actually diagnose and eliminate mistakes, but "no".
This does not add up very well:
The Xprot for that driver was set to 18.1mm by the LSI protocol. Whatever more possible protections might be activated at 16mm, it allowed the driver to fly well past 20mm one way during the whole long test, and the driver touched above 25mm few times too. If you know that driver, you would know it really does not want to go that far at all, and if we interpolate Bl and Kms for these excursions, we would understand that this poor thing was getting 6,5-8kW peaks.
So I have very hard time believing it was very restricted, and this is the reasoning why Bl falloff at x 13mm shows same or higher than it does on 21DS115.
On the other hand, the depicted 21DS115, arguably, got some kind of restriction or protection set at 13mm, 3mm sooner than the 18DS115 had, therefore it should have beem showing lesser values if anything was to affect it, no larger. It goes exactly against the issue I am pointing out.
I am waiting for the original B&C Speakers 21DS115 LSI report to make more precise comparisons still.
Just gathering data, and I am going to try to make a Xmax sweep througout applicable frequency range (let´s say 15-60Hz) to monitor the DC offset across all frequencies. These presented graphs do not show that, and it is missing in my driver´s evaluation. That´s it, no biggie. Just wanna see further, to figure out if I needed and what the hell is going on in general.
I did not make any assumptions on this and the red line crossing is "accidental". It was precisely picked for the reason that on the 18DS115 driver, at 13mm of excursion, the far end of the voicecoil exactly meets the pole piece thickness, while the 21DS115 has its coil out by 2mm already. There is no side intentions and in between the lines again, these assumptions are moot. I am glad you poke to it so we actually diagnose and eliminate mistakes, but "no".
This does not add up very well:
On the 18DS115 graph, the red lining starts at ~16mm.
The Xprot for that driver was set to 18.1mm by the LSI protocol. Whatever more possible protections might be activated at 16mm, it allowed the driver to fly well past 20mm one way during the whole long test, and the driver touched above 25mm few times too. If you know that driver, you would know it really does not want to go that far at all, and if we interpolate Bl and Kms for these excursions, we would understand that this poor thing was getting 6,5-8kW peaks.
So I have very hard time believing it was very restricted, and this is the reasoning why Bl falloff at x 13mm shows same or higher than it does on 21DS115.
On the other hand, the depicted 21DS115, arguably, got some kind of restriction or protection set at 13mm, 3mm sooner than the 18DS115 had, therefore it should have beem showing lesser values if anything was to affect it, no larger. It goes exactly against the issue I am pointing out.
I am waiting for the original B&C Speakers 21DS115 LSI report to make more precise comparisons still.
I expect far for perfection, and this was still very close.
Just gathering data, and I am going to try to make a Xmax sweep througout applicable frequency range (let´s say 15-60Hz) to monitor the DC offset across all frequencies. These presented graphs do not show that, and it is missing in my driver´s evaluation. That´s it, no biggie. Just wanna see further, to figure out if I needed and what the hell is going on in general.
So, I've klippel tested a lot of drivers.. and that slight asymmetry on the first curve is still a very good result. The vast majority of drivers measure out a whole lot worse on symmetry.
Where are you seeing 13mm vs 16mm vs. 18.1mm protections? Klippel protections are not set by distance, they are set by a percentage.
Where are you seeing 13mm vs 16mm vs. 18.1mm protections? Klippel protections are not set by distance, they are set by a percentage.
That´s what I meant by my expression. I really will have to go by most basic wording, because obviously I can mingle it enough for others to get confused.
The 13mm is red lining starting as you mentioned on the post #128 of yours in this thread and #109 of mine , where I attached the 21DS115 Bl(x) image from AudioXpress. You assumed I picked that point of 13mm because of some kind of limit or whatever assumptions made. If any limit was imposed on the speaker, it was imposed sooner on the 21DS115 than it was imposed on 18DS115 by these graphical figures I attached in this post below again, therefore it should be 21DS115 pulling the shorter straw, with both sooner protection and its shorter voice coil. The 18DS115 has both longer coil and later limit in these measurements (graphs), and yet it almost loses in the Bl curve falloff amount. The 16mm "limit" - same thing, but on 18DS115 Bl(x) image in post #109, taken from the B&C´s LSI report. 18.1mm limit is taken from 18DS115 LSI measurement file as a value in the table.
Oh they are:
The limit values might be reasoned by percentages, but the actual action of protection system measured by laser distance probe is set as distance. It has been even described somewhere, unfortunately I do not remember where. Maybe one of the Benett Prescott´s vids. Of course this system is slow, it can probably only apply some kind of amp limiting or momentary gain control, that´s why the 18DS115 was allowed to fly past 20mm:
Sorry for not including full view, the file was sent to me personally, so I cannot just openly share it...
But you can see that on the LSI graph provided before, the red lining starts at about 16mm for 18DS115, while simultaneously Xprot is set to 18.1mm.
Possibly different kind of protection than Xprot is in action, but it still does not stop the driver from moving as it should in that x.
Last edited:
You specify the percentage of protection and the machine tries to figure out what that is in millimeters once it leaves enlargement mode and enters nonlinear mode. This changes with every result on the test. If you had the database file, you can scrub through the timeline and see this. This protection depends solely on what the test has resolved to at that point in time. Before the test resolves down to your protection percentages, and when it is in enlargement mode, it can be all over the place.
With all due respect, I have tested drivers weekly and sometimes daily for over a year, so I know how the machine works and how it gets its data. You are not going to tell me anything I don't already know about it. There is a lot more data to interpret in the database file than you can see on a generated report and trying to draw lines from the graphs it generates for the pdf file.
In the report you have, under the Table States section, what does it say in the BLmin% and Cmin% fields? It will have an entry that is formatted like: 70 (70) %. The first number is what the test has resolved to (unrelated to protection), and the number in parentheses is the protection percentage that you set at the beginning of the test.
With all due respect, I have tested drivers weekly and sometimes daily for over a year, so I know how the machine works and how it gets its data. You are not going to tell me anything I don't already know about it. There is a lot more data to interpret in the database file than you can see on a generated report and trying to draw lines from the graphs it generates for the pdf file.
In the report you have, under the Table States section, what does it say in the BLmin% and Cmin% fields? It will have an entry that is formatted like: 70 (70) %. The first number is what the test has resolved to (unrelated to protection), and the number in parentheses is the protection percentage that you set at the beginning of the test.
That's oversimplifying the issue.
There are plenty of companies with products that give a list of recommendations.
If it's gonna be community/DIY based, there will be also another list of recommendations.
(In fact, even Klippel has a list of recommendations in their documentation).
On top you can give a list of minimum requirements.
This is personally also the biggest critique I have at the Dayton system.
Why on earth do you want to compete with amplifier manufactures out there?
You will never ever reach that amount of quantities, therefor the price will always be less optimal.
Make no sense, plus it make the product unnecessarily expensive.
Additionally, it means that if that part breaks (which is far more likely compared to the other parts), the entire system won't work anymore.
Besides, I highly doubt about their quality as well.
Impossible to robotize.
Robotizing things only works when the prompter knows what he wants.
Guess what?
The very vast majority of my clients don't (exactly) know what they want.
They say A, mean B, we talk about C but they actually need Z.
After doing this for a while you can an additional sense for this.
Even more so for the grey areas in between.
Hi is the - Klippel earlier model - for sale 🙂
Kippel is actually tracking the driver movement with very high accuracy in real time to be able to control in a way to put the protections in place. Compare to this sort of tracking Dats DC offset model - though a gr8 idea seems to be far behind.
I am sure it can be all done diy at a fraction of the cost (with new types of sensors & processing), hope some one - gets a dream / thinks out of the box / figures it out.
Thanks for your inputs on Femm
Was this inferred from Femm
It's even percentage of like BL it self.
Which is very confusing, since excursion is the physical limit, not BL.
Don't understand what Klippel was thinking there.
Especially since they only show those BL(x) graphs on a absolute way, not a relative way (= percentage).
I always export the results to a spreadsheet to convert them to a relative graph.
That way it's much easier to see and compare different graphs with each other.
Another important note on the results.
Over the years, I have noticed that (some) manufacturers change mechanical properties.
This can also skew the results we're seeing here as well.
No they are not.
Those are calculated protection values, not the measured protection values.
Not the same thing.
Do you have dB-Lab installed?
Yeah, there is a lot of data to look at in the database to be able to determine if your results are good or if they are not. I have had some professionals tell me early on not to share klippel data unless necessary, because of the risk of it being misread or misinterpreted and for that to spread as if it were legit data. Not saying that's what is happening in this thread; it's just one of the lessons I learned early on.
This is two different BL results from the same test on a driver I tested two days ago; one is good data and one is bad data.. what data in the database do we look at and how do we tell which result is good or not? 🧐🧐🧐🧐
This is two different BL results from the same test on a driver I tested two days ago; one is good data and one is bad data.. what data in the database do we look at and how do we tell which result is good or not? 🧐🧐🧐🧐