...which makes them LESS accurate than sim programs that do. I don't know why that is so difficult for you to grasp. Good grief - It's like trying to say that programs that choose a tire size for your car without using a value for rim size will be more accurate than those that do use that value.
And no, those programs don't use "other parameters". The programs you refer to all use Vas, Fs, Qts, Qes, and Qms and Hornresp uses BL, Cms, Rms, Sd and Mmd, but double-click on the Sd box on the Hornresp "Enter Parameters" page and see what comes up. Looks familiar? It should, because Vas, Fs, Qts, Qes and Qms can be DERIVED from BL, Cms, Rms and Sd, and vice-versa. Le however CANNOT be derived from any of those parameters, and any programs that don't consider Le in a loudspeaker sim are by definition less accurate than those that do.
The 'lossy Le' feature in Hornresp is simply an adjustment to compensate for the difference between sim and actual measurements that's been observed when drivers with long high-inductance coils are used. You can choose to it, or choose to ignore it, and no, it's not ONLY a Hornresp "problem". The problem will show up in any sim program that uses the basic t/s parameters.
On the important part, the low end, it does not. And that's what's really needed, isn't it?
I already thought the same several times about you. But I got hope again, you stopped claiming 'lossy Le' is a problem for WinISD.
Well, that's a pretty accurate description. Except you have to swap your candidates. What use is it if you can simulate the upper part half-way correctly if you can't rely on the low end?
Yes, that's absolutely correct. But I already said it, you get much better, precise results if some of these values are not accurate. I.e. the Qms and Rms are already calculated, not actually measured (you can't measure Rms directly i.e.) and you additionally get another TWO rounding errors. On all of them. Especally on Qms you'll get a huge variety of different specs for the same driver, each time someone measures the parameters. And the Le value is also dependent on the frequency. Yes, that shifts with the frequency. And a lot folks who measure the Le unfortunately have very big measurement tolerances.
Le can't be derived, that's ofcourse true. But for the low end it does not matter. I don't care about the upper end, that's often not that precise on hornresp either because the driver response becomes more and more important. And remember the 4dB gain at the upper end at your simulation? You couldn't explain that despite constantly claiming it would be correct. And for the accuracy: Well, did you finally compare the simulations? You obviously did not.
You didn't compare simulations. You aren't even able to grasp it can't be a problem if it isn't something the simulation relys on. Please, please, please, simulate the drivers in your linked thread with hornresp and with WinISD or BassCAD and compare them to the measurements. You'll notice they match perfectly on the lower end without any estimated inserted value. I never claimed WinISD is perfect, I admitted the downside myself. If the only tool is a hammer, everything looks like a nail. Please try a different tool and you'll find that screws actually exist and screwdrivers are actually working.
You can't claim it's less accurate because it doesn't use it and at the same time argue it would have the same problem. If I don't use the spare tire, how can it be a problem being flat?
Okay, so I tested. WinISD Pro, this driver: Data-Bass
Compared to the simulations here: Simulating high inductance more accurately
Second driver.
WinISD Pro gives a curve that's similar to the uncorrected Hornresp curve*, and therefore not particularly close to what's measured. The corrected (Lossy Le enabled) Hornresp curve is bang on.
I urge you to try this for yourself.
* The difference is that Hornresp shows the response will drop over a couple of hundred Hz, while WinISD thinks it'll be flat to 50kHz.
Chris
Compared to the simulations here: Simulating high inductance more accurately
Second driver.
WinISD Pro gives a curve that's similar to the uncorrected Hornresp curve*, and therefore not particularly close to what's measured. The corrected (Lossy Le enabled) Hornresp curve is bang on.
I urge you to try this for yourself.
* The difference is that Hornresp shows the response will drop over a couple of hundred Hz, while WinISD thinks it'll be flat to 50kHz.
Chris
I did.
I already said that, are you kind of slow?! I've already said that numerous times! The low end is correct at BassCADe, while at hornresp IT IS BY NO MEANS! What is it I want from such a tool? A correct, reliable response in the bass to be sure the enclosure works. And, unlike hornresp, it does exactly that. You want to simulate above? Use BoxSim or similar tools which use actual measurements.
I have to say, I didn't expect anyone ever would be quite that learning resistant.
Attachments
Thanks Chris. Just highlighting the important parts that seem to be lost on our Le-adverse forum member
I did.
I already said that, are you kind of slow?!
Great, now go look at the MEASURED response in my second link.
On the Hornresp screenshot, the grey curve is the uncorrected reponse (same as WinISD, and your simulation), and the red curve is the Lossy Le corrected response.
Note the following:
- The uncorrected Hornresp simulation and the WinISD simulation are the same below 100Hz
- NEITHER OF THEM ARE EVEN A TINY BIT CLOSE TO WHAT'S ACTUALLY MEASURED.
- The measured response is actually very close to the Hornresp sim with Lossy Le enabled.
To suggest I'm slow would be rather close to being a personal attack. Careful. We don't want you getting kicked off the forum when there's clearly so much learning to be done!
Chris
Edit, so that we are one million percent clear, I'm talking about this:
Wut? After you enter the t/s into Hornresp you can verify the parameters as Brian mentioned by double clicking Sd. It should be the same parameters you are entering into WinISD so t/s differences are not an issue, they are not different.
Ok, take a look at that sim you just posted and look at the one Chris linked to which includes the measurement. The measured and Hornresp simulated graphs have a 7 db hump at 35 hz (compared to 100 hz level). The sim you did shows a 2 db hump at 40 hz. How is that even remotely similar?
Let's understand that all these different programs are doing different calculations. The basic theory is the same but some include things like port end corrections while some don't.
Hornresp is capable of simulating lossy Le more accurately that WinISD and whatever program you used for your sim.
WinISD is capable of simulating losses, it can do at least 2 of these 3 - Ql, Qa and Qp. IIRC the box losses are set particularly high by default so if that isn't changed it can have an effect vaguely similar to lossy le. WinISD can also simulate le in the later versions if you check the checkbox but probably not particularly well. Why would Le even be an option that needs a checkbox anyway? It's a vital part of the sim.
Who knows what BassCAD is doing? If it doesn't even accept an Le parameter it can't be doing much right. Maybe it's simulating Ql, Qa and Qp, maybe not. Maybe it's doing other things. But clearly it is not producing an accurate sim.
With that out of the way, I'd like to move back to using a set of six of these drivers for low-frequency reinforcement in PA settings.
In light of the lower-than-expected linear cone movement available, I've moved the port tuning frequency from 45Hz up to 60Hz, which has gained some efficiency and mechanical power handling.
Since this cabinet won't provide much low-frequency extension, I'm going to call it a bass module (or something like that), rather than an actual subwoofer.
This was a quick simulation of a down-firing ported box. The rear chamber and port is as standard for a ported box, but to simulate a down-firing subwoofer, you need to use an offset-driver setup and a couple of sections that relate to the height of the cabinet off the ground, and its circumference.
To come up with the areas all I did was say it'll be 40mm off the ground, and with three drivers arranged in a circle (ie, there's rotational symmetry), you get a small area at the terminus, (pi*driver radius*height) at the driver's tap point (S2), and (cabinet circumference*height) at S3. Simple maths, but if anyone wants to know how to simulate a down-firing subwoofer properly, that's how to do it.
Shown is maximum SPL with 4mm of excursion and a 50v limit on the amplifier, and the 24v input required to reach around 4mm of excursion.
Also shown is the maximum SPL from the main speakers (grey trace in the top image) with 50v and 5mm Xmax limits. This bass module should have a useful amount of headroom, and for acoustic artists with the occasional bass guitar or stompbox, enough low-frequency extension, too.
The main speakers will operate with a highpass filter, and they're simulated flown. I've tested them,
I'm going to work a little on optimising a 60Hz design - the downfiring one was just playing around to see what could be had. It might turn out a bandpass design has a place here.
Chris
Edit - cross-posted with JAG
In light of the lower-than-expected linear cone movement available, I've moved the port tuning frequency from 45Hz up to 60Hz, which has gained some efficiency and mechanical power handling.
Since this cabinet won't provide much low-frequency extension, I'm going to call it a bass module (or something like that), rather than an actual subwoofer.
This was a quick simulation of a down-firing ported box. The rear chamber and port is as standard for a ported box, but to simulate a down-firing subwoofer, you need to use an offset-driver setup and a couple of sections that relate to the height of the cabinet off the ground, and its circumference.
To come up with the areas all I did was say it'll be 40mm off the ground, and with three drivers arranged in a circle (ie, there's rotational symmetry), you get a small area at the terminus, (pi*driver radius*height) at the driver's tap point (S2), and (cabinet circumference*height) at S3. Simple maths, but if anyone wants to know how to simulate a down-firing subwoofer properly, that's how to do it.
Shown is maximum SPL with 4mm of excursion and a 50v limit on the amplifier, and the 24v input required to reach around 4mm of excursion.
Also shown is the maximum SPL from the main speakers (grey trace in the top image) with 50v and 5mm Xmax limits. This bass module should have a useful amount of headroom, and for acoustic artists with the occasional bass guitar or stompbox, enough low-frequency extension, too.
The main speakers will operate with a highpass filter, and they're simulated flown. I've tested them,
I'm going to work a little on optimising a 60Hz design - the downfiring one was just playing around to see what could be had. It might turn out a bandpass design has a place here.
Chris
Edit - cross-posted with JAG
Last edited:
118dB will be fine - the midbass driver is capable of around 110dB, so there's headroom in the bass. Let's remember that the 15"s I use regularly are capable of around 125dB - I'm not expecting to replace them.
What I'm after here is a system that'll sound full-range, and provide a enough clean output for a church or small pub gig. Looks to me like a 60Hz ported box of some flavour will probably do it, but I'm going to keep trying different things until I find something that maximises the output from these drivers.
Chris
What I'm after here is a system that'll sound full-range, and provide a enough clean output for a church or small pub gig. Looks to me like a 60Hz ported box of some flavour will probably do it, but I'm going to keep trying different things until I find something that maximises the output from these drivers.
Chris
This is a far better idea for general PA.
For reference, the vast majority of 15" on a stick type PA bins are already rolling off from 80hz and sound fairly 'full' for general PA use especially with a 63hz bump on your average 15 band EQ or powered mixer. Usually more than adequate for background music, DJ's etc.
Dont be fooled by quoted bass extension on commercial cabinets. Most of the big name boxes are actually quoting -10db limits for extension and the powered cabinets often use dynamic equalisation(JBL eon is a great example) that makes them sound deep at low SPL but near limiter is barely able to hit 60hz with any authority.
Using your SIM and commercial 'inflated ratings' it would be sold as a bass module with 48-200hz frequency range and 124db peak. They often quote 6db or more as a 'peak' rating and often that's centred on any peaks in the operating bandwidth before eq. And yep, they kinda all do it...
I would go and make a cheap prototype cabinet from MDF(don't spend any serious time on finish) and give it a test with suitable filtering and power and see how it goes
For reference, the vast majority of 15" on a stick type PA bins are already rolling off from 80hz and sound fairly 'full' for general PA use especially with a 63hz bump on your average 15 band EQ or powered mixer. Usually more than adequate for background music, DJ's etc.
Dont be fooled by quoted bass extension on commercial cabinets. Most of the big name boxes are actually quoting -10db limits for extension and the powered cabinets often use dynamic equalisation(JBL eon is a great example) that makes them sound deep at low SPL but near limiter is barely able to hit 60hz with any authority.
Using your SIM and commercial 'inflated ratings' it would be sold as a bass module with 48-200hz frequency range and 124db peak. They often quote 6db or more as a 'peak' rating and often that's centred on any peaks in the operating bandwidth before eq. And yep, they kinda all do it...
I would go and make a cheap prototype cabinet from MDF(don't spend any serious time on finish) and give it a test with suitable filtering and power and see how it goes
Last edited:
To further my point about inflated ratings(to give you a bit more confidence on what ACTUAL PA subs are like in regards to response) have a look at the Mackie SWA-1501. A 500wrms powered bandpass sub. Bit of an industry standard for DJs etc for quite a while.
It has a quoted response down to 36hz, a peak SPL of 128db (125 cont)but when you look at the specs, you can see The 36hz is actually -10db referenced from 120hz but from 60hz is more like 15+db down. The response curve is more or less one big jump centred at 60hz and dropping fairly fast either side.
The 128db peak will also be referenced to that 60hz hump(not at all averaged from 36-120hz) and if you work backwards, (125 db continuous) would need a 1w sensitivity of 97 db or so(perfectly reasonable) yet their response chart isn't shown as 1w/1m, it is simply referenced to 100hz at 100db. Lots of number fudging!
It is most likely to be 97db sensitive at 60hz, meaning at 36hz it's more like 82db 1w!
The most important part of my comments is the absolute importance of 60-80hz output capacity as that is where you really need it most in PA work!
It has a quoted response down to 36hz, a peak SPL of 128db (125 cont)but when you look at the specs, you can see The 36hz is actually -10db referenced from 120hz but from 60hz is more like 15+db down. The response curve is more or less one big jump centred at 60hz and dropping fairly fast either side.
The 128db peak will also be referenced to that 60hz hump(not at all averaged from 36-120hz) and if you work backwards, (125 db continuous) would need a 1w sensitivity of 97 db or so(perfectly reasonable) yet their response chart isn't shown as 1w/1m, it is simply referenced to 100hz at 100db. Lots of number fudging!
It is most likely to be 97db sensitive at 60hz, meaning at 36hz it's more like 82db 1w!
The most important part of my comments is the absolute importance of 60-80hz output capacity as that is where you really need it most in PA work!
Attachments
Last edited:
Rest assured, I did.
The comment on adjusting the port length and the reason for doing so was hard to miss, particularly as it was reiterated by ICG immediately after you originally made it. I wondered at the time why he felt the need to repeat the information.
I tried to download a copy of BassCADe to see if it can model in 1 Pi and 0.5 Pi space, and take path length differences into account when combining outputs, but the program was deemed to be suspect by my Norton Antivirus software and the downloaded file was unfortunately automatically deleted.
I have now managed to download a copy of BassCADe 3.4.4.0 by circumventing my Antivirus software, and taking the chance that the program is indeed safe.
It is not clear to me what assumptions are made as far as solid radiation angle is concerned, and acoustic path length differences from the observation point to the direct radiator and port outlet do not appear to be taken into account when simulating a "reflex vented box". The software seems to use a simple conventional Helmholtz resonator model, where the enclosure is specified by volume only, and is treated as a pure acoustic compliance.
Last edited:
In effect what you are saying is that BassCADe automatically knows under what conditions, and by how much, to reduce the specified Bl motor strength based on the given driver Thiele-Small parameter values, and a fixed value for Le. It is simply not possible for any simulation program to do that.
It is important to understand that there is nothing magic about the BassCADe bass-reflex model. It is a essentially just a copy of the original classic model developed by Neville Thiele and Richard Small. WinISD also uses the same T-S model, which is why the results for the two programs are so similar.
Hi just a guy,
Nice to see you posting again from time to time. Hope things are continuing to go well.
Kind regards,
David
Preaching to the choir there - here are a couple of articles I wrote on the subject that you might find interesting.
Spec Wars: Looking Inside Loudspeaker SPL Specifications - ProSoundWeb
and the follow-up: Finding The Limits: Pursuing Useful Loudspeaker Specs, Including SPL - ProSoundWeb
Chris
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.