Yes Understood.
But its been mentioned by the OP that some parameters entered
failed a Integrity Check.
Hence the software rejects it.
The entered values fail.
Winisd also has a Integrity Check
If it fails, The software shows you which parameters dont match
the proper formulas.
It also does not allow you to save or run the model.
Hence rejects it.
So yes as noted it can be user error entering the parameters.
Or many times the manufacture has rounded up the parameter values.
I can see now, the OP is also looking for what parameters can be easily fudged in a datasheet. And how we can possibly derate the values.
Its been mentioned that FS is a good place to start since most manufactures
use 5 to 10% average for rejection ratio.
And when most but not all drivers are new Fs tends to be slightly higher
I would also agree I have seen manufactures fudge Le Values and Xmax values and Vas values. looking at older data sheets.
Far as Xmax , im sure they love laser measured methods
since it tends to make Xmax higher.
Is there some resource which tells you which parameters are most likely to be affected by manufacturing variations? Physical dimensions should be fairly consistent, but the other stuff.... how much of a spread should one expect?
In purely electronic simulations we have something called Monte-Carlo simulation where the response is calculated based on the tolerances of each component in the circuit to get an idea of the worst case scenario, i.e. what would happen if the components were at the end of their tolerance limit - upper or lower, depending on which would hurt more.
In purely electronic simulations we have something called Monte-Carlo simulation where the response is calculated based on the tolerances of each component in the circuit to get an idea of the worst case scenario, i.e. what would happen if the components were at the end of their tolerance limit - upper or lower, depending on which would hurt more.
Yes, I am familiar with monte-carlo calculations.
Besides electronics, I have a physics background as well, so I am actually used to error or inaccuracy analysis.
Even when doing simulations.
These calculations can sometimes get a little long and tedious, although these days one can just use a math solver for these things
The problem is that the T/S parameters themselves are already derived when being measured. So to get those tolerances or errors, the manufactures has to provide which variables they use (are controlled) and what other parameters are calculated from that.
In general Sd, Re, Mms are very easy to control.
And although BL is a relatively easy parameter to keep steady, it is a hard parameter to measure directly.
Next would be the Rms, which is also pretty constant over the frequency range.
The hardest is the Cms.
Which is just a bit of a weird parameter by itself, since it heavily depends on material properties. It is also the least linear of the bunch.
The static compliance is totally different than the dynamic compliance for example.
I have to admit, I have never created a speaker/driver from scratch myself.
Although I did a lot of work and calculations on a device to measure those parameters for passive radiators.
But I have experience ordering speakers to a certain spec.
I get a sense that a lot is done by trial and error (which is pretty standard in acoustics in general).
After a while a manufacturer just has a database of material properties, which were measured by just trying and verifying.
In general the whole specific field feels almost a little voodoo.
I have never seen any people in the community sharing their experiences and knowledge (which is extremely unfortunate), and when I ask on a pro B2B level about what is going on, I always get very cryptic answers back.
The amount of spread seems to be even a deeper secret.
Which from an electronics point of view is totally crazy, since often it is clearly stated in the datasheet.
And even an hard selling point for some IC's
Besides electronics, I have a physics background as well, so I am actually used to error or inaccuracy analysis.
Even when doing simulations.
These calculations can sometimes get a little long and tedious, although these days one can just use a math solver for these things
The problem is that the T/S parameters themselves are already derived when being measured. So to get those tolerances or errors, the manufactures has to provide which variables they use (are controlled) and what other parameters are calculated from that.
In general Sd, Re, Mms are very easy to control.
And although BL is a relatively easy parameter to keep steady, it is a hard parameter to measure directly.
Next would be the Rms, which is also pretty constant over the frequency range.
The hardest is the Cms.
Which is just a bit of a weird parameter by itself, since it heavily depends on material properties. It is also the least linear of the bunch.
The static compliance is totally different than the dynamic compliance for example.
I have to admit, I have never created a speaker/driver from scratch myself.
Although I did a lot of work and calculations on a device to measure those parameters for passive radiators.
But I have experience ordering speakers to a certain spec.
I get a sense that a lot is done by trial and error (which is pretty standard in acoustics in general).
After a while a manufacturer just has a database of material properties, which were measured by just trying and verifying.
In general the whole specific field feels almost a little voodoo.
I have never seen any people in the community sharing their experiences and knowledge (which is extremely unfortunate), and when I ask on a pro B2B level about what is going on, I always get very cryptic answers back.
The amount of spread seems to be even a deeper secret.
Which from an electronics point of view is totally crazy, since often it is clearly stated in the datasheet.
And even an hard selling point for some IC's
Excellent reply, b_force, but rather depressing for us.
The standard design procedure would be to select a driver based on published specs, simulate and design the enclosure/ horn, whatever, then buy the driver and construction materials.
Instead, we have to either:
1. Buy the drivers, test each driver separately for TS parameters, design the enclosure for each based on those, and then build and assemble it.
or
2. Buy the drivers, design the enclosure based on published data, then adjust the enclosure dimensions - stuffing/variable volume insert, etc. - while using a SPL measuring setup.
Perhaps during simulation we should be testing for various values of (for instance) Cms - a little above and below the published value - and see how badly these changes affect the response.
The standard design procedure would be to select a driver based on published specs, simulate and design the enclosure/ horn, whatever, then buy the driver and construction materials.
Instead, we have to either:
1. Buy the drivers, test each driver separately for TS parameters, design the enclosure for each based on those, and then build and assemble it.
or
2. Buy the drivers, design the enclosure based on published data, then adjust the enclosure dimensions - stuffing/variable volume insert, etc. - while using a SPL measuring setup.
Perhaps during simulation we should be testing for various values of (for instance) Cms - a little above and below the published value - and see how badly these changes affect the response.
Well, luckily we also have a community , and people like Hificompass, as well as VoiceCoil Magazine
Measurements and compare | HiFiCompass
Items about: Voice Coil Test Bench | audioXpress
Especially the Klippel measurements are extremely useful!
And a couple of others like Zaph, but he hasn't updated in years, and some magazines Hobby Hifi and Klang+Ton here in EU if you can read a little bit of German, although basically you just have to look at the graphs.
And a whole bunch of other ones I am forgetting at the moment.
Zaph|Audio
Oh, and diBirama as well
(although distortion plots are extremely limited unfortunately )
diBirama italian diy loudspeakers, autocostruzione, Diego Sartori
What I do often, is cross-correlate results between all those 3rd party measurements, since most of them are only relative numbers (only valid under these specific conditions)
You won't find exact numbers, but you certainly see certain trends.
After a while you also get a sense what brands know what they are doing and what brands are kind of all over the place.
Or what brands are just basically plain lying in their datasheets.
(read: cherry picking to the max).
Even the frequency response is sometimes totally bogus.
You will also see that certain drivers that are more expensive will perform only marginally better or not better at all.
Next is to try to think in ballpark figures.
You can kind of design in a way to allow some spread or tolerances.
One of the reasons why for example closed box systems are so much easier.
Since a multi-sub solution is the only good way to attack those room modes, the mid-woofer only needs to go from 80-100Hz (or can be even higher in some cases).
This is of course a little trickier if you want to design a full-range system.
But yeah, sometimes it is even trial and error for us unfortunately.
Again, from an electronics points of view that feels totally insane.
Even those load lines from old tubes were a better start in the whole design.
, and people like Hificompass, as well as VoiceCoil Magazine Measurements and compare | HiFiCompass
Items about: Voice Coil Test Bench | audioXpress
Especially the Klippel measurements are extremely useful!
And a couple of others like Zaph, but he hasn't updated in years, and some magazines Hobby Hifi and Klang+Ton here in EU if you can read a little bit of German, although basically you just have to look at the graphs.
And a whole bunch of other ones I am forgetting at the moment.
Zaph|Audio
Oh, and diBirama as well
(although distortion plots are extremely limited unfortunately
)diBirama italian diy loudspeakers, autocostruzione, Diego Sartori
What I do often, is cross-correlate results between all those 3rd party measurements, since most of them are only relative numbers (only valid under these specific conditions)
You won't find exact numbers, but you certainly see certain trends.
After a while you also get a sense what brands know what they are doing and what brands are kind of all over the place.
Or what brands are just basically plain lying in their datasheets.
(read: cherry picking to the max).
Even the frequency response is sometimes totally bogus.
You will also see that certain drivers that are more expensive will perform only marginally better or not better at all.
Next is to try to think in ballpark figures.
You can kind of design in a way to allow some spread or tolerances.
One of the reasons why for example closed box systems are so much easier.
Since a multi-sub solution is the only good way to attack those room modes, the mid-woofer only needs to go from 80-100Hz (or can be even higher in some cases).
This is of course a little trickier if you want to design a full-range system.
But yeah, sometimes it is even trial and error for us unfortunately.
Again, from an electronics points of view that feels totally insane.
Even those load lines from old tubes were a better start in the whole design.
I am going on a (big) little side note here.
Really curious about any thoughts on this!
The whole thing around T/S parameters always baffled me.
Or more the lack thereof in some datasheet and even history.
Just a extremely brief history, feel free to correct or nuance certain things.
The concept of mechanical–electrical analogies was already known for ages.
We are talking basically way before 1900.
But even from 1900, it still took around 50-60 years to really get a bunch of parameters that are usable to work from (and still only to a little extend).
btw, obviously, all the highest respect to those early engineers and pioneers!!
But from an historical and physics point of view, it is kind of confusing.
In other fields these techniques on similar difficult problems were used for many decades?
It just makes you wonder? Why took it so long to get something going?
And still until this day, there are still no standards compared to datasheets for electronics?
But the second thing that really makes me wonder.
For example KEF was already publishing those parameters in their datasheets.
In the 80s for certain, but (maybe the real KEF expert can tell), as far as I know also already around 1970-1975 and such.
At the same time, Philips for example, only gave very limited information.
Even far in the 80s; Philips Loudspeaker Databooks
Were those T/S not known by Philips, didn't they agree with the idea of T/S parameters, or did they just simply never published them?
(and there is still a big secret safe with all the original parameters )
I have some old books from 86/87.
Seas, Focal, Scan Speak, Vifa, Isophon, Audax, Celestion, Kef etc can be found in those databooks. But never Philips.
As far as I know, they did sell speakers for the DIY market or other OEM/B2B ?
The reason is not because I am particularly fascinated by Philips (although a little).
But more to try to understand why and how certain manufactures (like Philips, but there are many others) publish certain data, or rather withhold the data.
Really curious about any thoughts on this!
The whole thing around T/S parameters always baffled me.
Or more the lack thereof in some datasheet and even history.
Just a extremely brief history, feel free to correct or nuance certain things.
The concept of mechanical–electrical analogies was already known for ages.
We are talking basically way before 1900.
But even from 1900, it still took around 50-60 years to really get a bunch of parameters that are usable to work from (and still only to a little extend).
btw, obviously, all the highest respect to those early engineers and pioneers!!
But from an historical and physics point of view, it is kind of confusing.
In other fields these techniques on similar difficult problems were used for many decades?
It just makes you wonder? Why took it so long to get something going?
And still until this day, there are still no standards compared to datasheets for electronics?
But the second thing that really makes me wonder.
For example KEF was already publishing those parameters in their datasheets.
In the 80s for certain, but (maybe the real KEF expert can tell), as far as I know also already around 1970-1975 and such.
At the same time, Philips for example, only gave very limited information.
Even far in the 80s; Philips Loudspeaker Databooks
Were those T/S not known by Philips, didn't they agree with the idea of T/S parameters, or did they just simply never published them?
(and there is still a big secret safe with all the original parameters
)I have some old books from 86/87.
Seas, Focal, Scan Speak, Vifa, Isophon, Audax, Celestion, Kef etc can be found in those databooks. But never Philips.
As far as I know, they did sell speakers for the DIY market or other OEM/B2B ?
The reason is not because I am particularly fascinated by Philips (although a little).
But more to try to understand why and how certain manufactures (like Philips, but there are many others) publish certain data, or rather withhold the data.
Last edited: