How well certain speakers act on a certain passive filter mostly, especially within a certain format cabinet (bafflestep + diffraction).
Even more so when the goal is to keep the crossover simple and affordable.
I have come across plenty of drivers over the years that just don't play so well.
The same goes for the woofer - tweeter combination, or rather where you put the crossover point.
It's hard to explain without showing and explaining the entire story, but see it as just a unfortunate combination of a hump in freq resp, some impedance curve that wasn't ideal in combination at a cabinet size that wasn't ideal.
Important to note, is that this is basically non-existent with active filters (especially DSP)
People here always make it seem like we always ONLY go for the best thing available and that is the ONLY ever approach.
In my experience across many projects, is that it's definitely not the case for a lot of people (or companies for that matter)
They just want a certain size cabinet with certain speakers in it, often for a limited budget.
In contrary what you say, a lot of people choose speakers first, than design a cabinet accordingly to whatever looks good or fits well.
Or figure out a size they want, pick the drivers they like and make a crossover accordingly.
For example, often they just have a certain budget, need to fit it somewhere, because that's the only place they have space for a speaker and now wonder what speakers to get because they are overwhelmed by the choices that are available.
If that is good or bad isn't relevant, it's just a fact that is how the situation is.
Obviously now with experience I can predict certain things beforehand, but for people without much experience I ALWAYS advice them just to do some preliminary fiddling around to see how well things will work.
Most importantly you learn A LOT with that, without spending a penny!
But I always explain to them that it's extremely important to understand that it won't be the final crossover that will work in practice!
I think what some smart people here don't realize, is how obvious certain things seem, but for a lot of people (especially beginners) they are not.
Also, some people have different goals and approach with speakers.
For some certain looks are more important for example.
Last edited:
Sorry for the late reply, I thought I was following this thread but something got turned off...
Tracing a factory response for preliminary design simulations is a good tactic to avoid the trap. It is not a 100% fool proof method, but it gives me some confidence that the drivers I have selected will play nice with each other and with the baffle.
The situation is different with DSP active design.
I agree... When designing with passive filters, it is possible to fall into a trap where the combination of baffle gain/diffraction, driver impedance, and driver response all conspire to create a difficult problem... and it can be hard to get a satisfactory result. Whereas with a different driver (or a different baffle shape), things can fall together beautifully and we can get a great response without much crossover complexity.
Tracing a factory response for preliminary design simulations is a good tactic to avoid the trap. It is not a 100% fool proof method, but it gives me some confidence that the drivers I have selected will play nice with each other and with the baffle.
The situation is different with DSP active design.
Kimmo, I have a feature request: LEAP 5 can produce an extended set of T/S parameters based on impedances measured at several increasing voltages, then use this extended data set to accurately predict the driver response at various output levels. Unfortunately LEAP is no longer supported or sold, so this functionality is out of reach for those without a current license. It would be very useful to have this function built into VituixCAD's T/S parameter measurement routine and the driver database.
While the low voltage T/S parameters are very useful, as the driver moves through its travel, the CMS, BL, and inductance all change, and by extension all the parameters calculated from these variables. Currently I measure the T/S parameters at different voltage levels and store them as different versions of the woofer that can be selected when designing a box to show how the performance changes with level. I would love to have a more integrated way to account for the change in performance as the system plays at different levels.
While the low voltage T/S parameters are very useful, as the driver moves through its travel, the CMS, BL, and inductance all change, and by extension all the parameters calculated from these variables. Currently I measure the T/S parameters at different voltage levels and store them as different versions of the woofer that can be selected when designing a box to show how the performance changes with level. I would love to have a more integrated way to account for the change in performance as the system plays at different levels.
I don't necessarily agree that this would be very useful. Volume and tuning of the box are fixed after construction is done and product is in use. You just select which voltage produces suitable average and save that T/S data for simulation (and construction). No problem to add temporary records to database until decision is done.
One technical problem is that driver database is tab delimited multi-column text file containing one row for each T/S data set. xml is already possible (and json after some coding). Grouping field and filtering is required to keep the same file format. That is basically the same as suffix text in Model field or using Revision or Updated field for grouping.
Most of the users use and can live very well with manufacturer's data. For example I for series production which cannot rely on individual T/S data, and actual SPL and Z measurements of the box+driver system is loaded for XO design compensating possible tolerances in T/S data which was used while simulation. Most of the users who measure T/S parameters follow recommendation of measurement program, usually including some limits related to signal level and impedance range.
One technical problem is that driver database is tab delimited multi-column text file containing one row for each T/S data set. xml is already possible (and json after some coding). Grouping field and filtering is required to keep the same file format. That is basically the same as suffix text in Model field or using Revision or Updated field for grouping.
Most of the users use and can live very well with manufacturer's data. For example I for series production which cannot rely on individual T/S data, and actual SPL and Z measurements of the box+driver system is loaded for XO design compensating possible tolerances in T/S data which was used while simulation. Most of the users who measure T/S parameters follow recommendation of measurement program, usually including some limits related to signal level and impedance range.
An absolute +1 for this.
Having some idea and view how a system is gonna perform with large signals, is absolutely mandatory for any professional or serious system.
It's actually one of the few reasons why some people will get disappointed after spending all there time and hard earned money on a design.
For professionals it's a step that's often forgotten, even more so for sound reinforcement.
It's actually something you have a look at BEFORE you choice your volume and tuning. Especially because compensation is very hard to do afterwards.
It's one of the things I miss the most and it's very unfortunate that besides LEAP everyone still seems to be still stuck at small signal parameters.
Which are everything except representable for how a system actually gonna perform.
In some professional cases even totally useless.
The workaround I use atm, is just make a couple of "other" or "new" drivers with the changed parameters.
It kinda works, but is very fiddly to setup and is not as good as what LEAP does.
What is the problem? Why not measure impedance etc at large signal? What is a large signal?
Personally i find the 100mV rms norm to be ridiculous for bass drivers , also manufacturers are not consistent following this norm, nor is it even close to the use case the speaker is mostly intended for( home , pa, etc) . Also a large resistor value (22 - 49 Ohm etc) for Rsense does not make sense in my view.
I tested the purifi drives for my project at 100mV, 1V, 2,8V and 4V rms. And yes the results vary, be it from 100mv to 1v is the biggest variation, from 2,8 to 4 it is pretty minimal.
So i use the 2.8Vrms imedance/ts parameters. And do not care anymore about the 100mv icm large Rsense measurements.
Personally i find the 100mV rms norm to be ridiculous for bass drivers , also manufacturers are not consistent following this norm, nor is it even close to the use case the speaker is mostly intended for( home , pa, etc) . Also a large resistor value (22 - 49 Ohm etc) for Rsense does not make sense in my view.
I tested the purifi drives for my project at 100mV, 1V, 2,8V and 4V rms. And yes the results vary, be it from 100mv to 1v is the biggest variation, from 2,8 to 4 it is pretty minimal.
So i use the 2.8Vrms imedance/ts parameters. And do not care anymore about the 100mv icm large Rsense measurements.
It's called large signal parameters, not "a large signal".
You can't put just a number to that, because it totally depends on the type of speaker.
Or better, how the BL(x), Kms(x), Le(x) and Re act.
Especially the non-linearity of the BL can cause quite some trouble here.
Which you ideally want to be aware of BEFORE settling in on a certain design, tuning, volume and cabinet.
Last edited:
If these parameters values vary with strength of applied voltage, how would you tune, choose volume and cabinet? Those aspects are fixed once chosen, the driver behaviour is not.
You can magically predict how a driver is gonna perform in a certain size cabinet?
That's a very unique talent............
Or, in practice, we have 5 different drivers with all their pros and cons.
I very very very rarely just select just one driver and settle the entire design around it.
That's just often not the reality.
The real answer is that there are multiple ways to design something.
Which is fully dependent on context and practicality how to approach.
So therefore there is no right or wrong way of order.
Last edited:
That's a great question people should ask! 👍
There are multiple ways to approach this.
Like always, there is not one single clear answer.
For real PA stuff, you could focus way more on large signal performance, because that's often how speakers operate.
Otherwise another example is to tune the speaker in such a way that it fits nicely in between.
Depending how and where the parameters shift, hence simulating this is such a great thing to have!
That way we at least have a bit of sense and visual, instead of relying on magical talents to be perfectly predict this beforehand.
So, if i understand this correctly, the problem is more how many times one has to do measurement the driver parameters and take it from there.
This may require to select another driver for your project to make a choice.
It therefore depends on how strong the parameters change with signal strength for given driver as a basis for selecting the driver you want to proceed with and then "make it fit in between" is just one choice.
This may require to select another driver for your project to make a choice.
It therefore depends on how strong the parameters change with signal strength for given driver as a basis for selecting the driver you want to proceed with and then "make it fit in between" is just one choice.
Of course not. Who said that we don't need to investigate available data of potential drivers before purchasing, or measure and simulate further after purchasing if data is not available? No one so don't speak sht.
Swarm's original goal was to simulate driver+cabinet combination with T/S parameters measured with different signal levels. That requires measuring of actual drivers if/because data is not available. So there can be few voluntary steps before selecting driver to the project i.e. building actual cabinet. Point is that we already have actual driver if we want to measure T/S with different signal levels or jump directly to building prototype where we can measure the rest. Simulation with small signal T/S is (hopefully) done before all that in case enclosure is conventional which can be simulated with simple box simulators. If not, measuring with different levels does not produce useful data for box simulation.
Putting all the trouble and effort in to make a prototype, instead of being able to simulate and investigate all kinds of scenarios, makes absolutely very little sense.
Currently you make it sound like a certain design and driver is made and designed in a sorta vacuum.
Never to be used and seen ever again.
Which is obviously not the case.
The same drivers are often being used later, sometimes in an entire new design again.
Are you saying to get through all the trouble again making, building and buying an entire new prototype again?
That's quite the waste of time, when you had the opportunity to measure all relevant data earlier.
In the beginning it could create a tiny bit more work, but eventually people can built up a database.
On top of that, we are getting more and more information from drivers made with Klippel systems from known sources.
Behavior can be distilled from there as well.
This can also be done with poorman's Klippel method, since it's the relative changes here that are important.
So the argument that you have the driver anyway, becomes completely irrelevant.
Unless you enjoy doing extra work and spending money again next time.
^How many poor man's Klippel parameters you have shared for VituixCAD community? How much authorized VituixCAD users have shared that kind of data? Assuming that answer stays zero or very close, do you think that majority would do something else than investigate datasheet, read Voice coil magazine, discuss in forums, simulate with standard T/S, purchase driver, build a speaker and tune it? It's normal practice - not a vacuum. In addition, poor man's Klipper data will not answer all the questions no matter who pays or measures. I'm not against wider simulations, and program allows saving multiple T/S for the same driver. How difficult this is to understand? Nothing more to say in this forum and especially for you.
I don't understand what you're trying to say? There are plenty of options in VituixCAD that the majority doesn't use?
How is it for any importance how much I share?
It's about providing and creating a platform and workaround for the entire community.
Or to say it on an easy way, trying to get rid of the ever guessing game.
Besides, you make it sound that improving such things is a bad thing.
I am pretty sure you know it's the right way of doing preliminary research and investigation.
Guessing and tuning is normal and frustrating practice.
The point you're missing here is that the guessing game can be an awful lot less.
Resulting in a lot less wasting time, effort and money (building prototypes isn't free either).
Fact is it's unnecessary and only being done because there is currently not such a program to simulate these things like LEAP is doing.
Btw, I am totally confused for the need to get all personal here, I would like to keep the conversation on the subject of loudspeaker simulation.
I think I am giving very reasonable arguments on the subject here.
Thank you.
How is it for any importance how much I share?
It's about providing and creating a platform and workaround for the entire community.
Or to say it on an easy way, trying to get rid of the ever guessing game.
Besides, you make it sound that improving such things is a bad thing.
I am pretty sure you know it's the right way of doing preliminary research and investigation.
Guessing and tuning is normal and frustrating practice.
The point you're missing here is that the guessing game can be an awful lot less.
Resulting in a lot less wasting time, effort and money (building prototypes isn't free either).
Fact is it's unnecessary and only being done because there is currently not such a program to simulate these things like LEAP is doing.
Btw, I am totally confused for the need to get all personal here, I would like to keep the conversation on the subject of loudspeaker simulation.
I think I am giving very reasonable arguments on the subject here.
Thank you.
Kimmo,
I think I've found a small bug in "Tools/Enclosure/Filter":
If you enter a Rs value different from "0" (let's say 1 Ohm for example), even if "La box" is not selected, this 1 Ohm value is reflected in SPL graph and results (like in info tab).
Hope I've been clear
Anyway thanks again for this amazing software
I think I've found a small bug in "Tools/Enclosure/Filter":
If you enter a Rs value different from "0" (let's say 1 Ohm for example), even if "La box" is not selected, this 1 Ohm value is reflected in SPL graph and results (like in info tab).
Hope I've been clear
Anyway thanks again for this amazing software