Is there any way to derate the manufacturer's numbers to come up with realistic TS parameters before buying?
I have all the test equipment to measure parameters once the driver arrives but last time I just ended up paying return shipping and being back where I started. The driver was so different than the specs that it would not have worked.
I was looking at buying the Dayton Audio DCS380-15 as an example.
I have all the test equipment to measure parameters once the driver arrives but last time I just ended up paying return shipping and being back where I started. The driver was so different than the specs that it would not have worked.
I was looking at buying the Dayton Audio DCS380-15 as an example.
Not the answer you seek, but IMO one policy does not fit all products--some manufacturers consistently hold tighter tolerances than others. In addition, in most cases, their measurements are better than ours (it's their dayjob after all--they're pros). Many of these parameters are not singular values and vary with operating conditions as well. Lastly, there remain differences of definition in what a given parameter even is.
Thank you for the response grindstone. I agree with what you said, but I think I failed to fully explain my situation.
I dont mean inconsistent TS parameters between different drivers due to manufacturing tolerances.
I mean mutually inconsistent nominal TS parameters published by the manufacturer. The kind of thing that would cause REW to say "driver data fail integrity check". In other words, the values could not represent any real driver. I imagine the manufacturer must have fudged one or more of the specs that they think will make them look good.
I dont mean inconsistent TS parameters between different drivers due to manufacturing tolerances.
I mean mutually inconsistent nominal TS parameters published by the manufacturer. The kind of thing that would cause REW to say "driver data fail integrity check". In other words, the values could not represent any real driver. I imagine the manufacturer must have fudged one or more of the specs that they think will make them look good.
Edit -you beat me to it. Keeping up & tweaking slightly:
Not really, is the short answer. The longer one goes something like this:
This doesn't necessarily mean the manufacturer isn't telling porkies. In fairness, it doesn't necessarily mean the manufacturer is telling porkies either. That's another one of the grey areas. Many manufacturers these days, even some producing rather expensive drivers, simply use a production QC / QA measure of 10% to Fs. That's it. Nothing else. As a result, a driver, or batch of drivers may pass QC, but differ significantly in other spec., and they will send them out in good faith because they don't know it is. Which isn't much cop for the rest of us, obviously. You can understand it to a point -manufacturers, except the small-scale producers, simply don't have the facilities and time to do full testing on every single drive unit. Assuming they did, they'd probably have to double the retail price to cover the facilities and employee time needed.
In terms of inconsistencies in published specs., some of it is no doubt incompetence, or a simple mistake (which is more forgivable). I've seen that a few times, as have we all I suspect.
In other cases, it's often a result of taking & publishing the mean value of each individual data / parameter from a batch of test samples, which will result in some shifts in those parameters that are calculated from other data sets, relative to what is published.
Not really, is the short answer. The longer one goes something like this:
- It depends what you mean by 'derate', and 'realistic', and
- It further depends on how you measure them. T/S parameters are not fixed per se. Small stipulated small-signal, i.e. usually a low voltage drive, which gives great accuracy within his intended context. However, many manufacturers use a higher voltage drive, pointing out (quite fairly) that that is how most loudspeakers are actually used in practice. The truth, as usual, lies somewhere in between and the best policy is to measure at different voltage drives, also accounting for VC heating if relevant to a given application, and the probable mean ambient climatic conditions (which also affects things)
This doesn't necessarily mean the manufacturer isn't telling porkies. In fairness, it doesn't necessarily mean the manufacturer is telling porkies either. That's another one of the grey areas. Many manufacturers these days, even some producing rather expensive drivers, simply use a production QC / QA measure of 10% to Fs. That's it. Nothing else. As a result, a driver, or batch of drivers may pass QC, but differ significantly in other spec., and they will send them out in good faith because they don't know it is. Which isn't much cop for the rest of us, obviously. You can understand it to a point -manufacturers, except the small-scale producers, simply don't have the facilities and time to do full testing on every single drive unit. Assuming they did, they'd probably have to double the retail price to cover the facilities and employee time needed.
In terms of inconsistencies in published specs., some of it is no doubt incompetence, or a simple mistake (which is more forgivable). I've seen that a few times, as have we all I suspect.
In other cases, it's often a result of taking & publishing the mean value of each individual data / parameter from a batch of test samples, which will result in some shifts in those parameters that are calculated from other data sets, relative to what is published.
Last edited:
One idea I have for derating: Assume, the parameters most likely to be inflated are the ones people can filter by a naive "bigger number better" mindset.
Perhaps the goal then should be to avoid using the TS parameters that are most likely to be inflated. I played around with avoiding using the published specs in the following order:
SPL (nominal value not used, but rather derived from below)
Fs (nominal value not used, but rather derived from below)
Vas (nominal value not used, but rather derived from below)
EBP (nominal value not used, but rather derived from below)
Cms (nominal value not used, but rather derived from below)
BL
Qes (nominal value not used, but rather derived from below)
Qts
Qms
Mms
Rms
Re
Sd
Le
Also a separate set of equations from least to most trustworthy:
Xmax (nominal value not used, but rather derived from below)
Sd
Vd
Using the above process barely changed the result. There exists a set of consistent parameters which are within 1db of the nominal spec across the entire response.
To make a large difference I took the worst (highest) FS I could find from any combination of the published specs. I did the same for SPL (lowest) and Vas (highest). I then set all those variables and derived the others. That made the driver different enough that you would need a completely different box, and it would never be tunable down to what the nominal one could (which has been my experience buying drivers in real life)
Not sure if this is really useful as a buying tool since it would make all drivers seem bad. If people started derating drivers in this way it would at least give manufacturers an incentive to publish mutually consistent parameters.
Perhaps the goal then should be to avoid using the TS parameters that are most likely to be inflated. I played around with avoiding using the published specs in the following order:
SPL (nominal value not used, but rather derived from below)
Fs (nominal value not used, but rather derived from below)
Vas (nominal value not used, but rather derived from below)
EBP (nominal value not used, but rather derived from below)
Cms (nominal value not used, but rather derived from below)
BL
Qes (nominal value not used, but rather derived from below)
Qts
Qms
Mms
Rms
Re
Sd
Le
Also a separate set of equations from least to most trustworthy:
Xmax (nominal value not used, but rather derived from below)
Sd
Vd
Using the above process barely changed the result. There exists a set of consistent parameters which are within 1db of the nominal spec across the entire response.
To make a large difference I took the worst (highest) FS I could find from any combination of the published specs. I did the same for SPL (lowest) and Vas (highest). I then set all those variables and derived the others. That made the driver different enough that you would need a completely different box, and it would never be tunable down to what the nominal one could (which has been my experience buying drivers in real life)
Not sure if this is really useful as a buying tool since it would make all drivers seem bad. If people started derating drivers in this way it would at least give manufacturers an incentive to publish mutually consistent parameters.
Do you mean that the parameters of this woofer were off or do you ask us what the parameters of this woofer will be?
If it is about the latter, try searching on forums. Maybe someone has measured them.
Because of the difference sin the way factory software measures drivers vrs the typical kit we use, the numbers are collapsed from the T/S curves (they are not single numbers), you get 2 different sets of completely valid T/S parameters.
Having built more than a few speakers, i have found that in practise the factory numbers are a better starting point than what we measure.
Our impedance measures are useful for QC and matching.
dave
That's called "someone doesn't know what the hell they're doing" and/or "someone is rounding off each individual value, making them not consistent"
Another vaguely related issue can be that different measurement techniques drive the woofers differently, and thus yield different though VALID parameters. Yikes!
Often the issue is more that there aren't any real standards when it comes down to T/S parameters.
Keep in mind that ALL manufacturers cherry pick their data.
This often results in a sort of chicken-egg discussion with the manufacturer.
Very recently I had something similar with a very known hifi speaker manufacturer.
One I know that they clearly know what they are doing.
I always double check T/S parameters from any datasheet. More to just get a feel about what brands I can trust and what brands are making a total mess.
In this particular case the parameters were way off.
Fs and Qt were both completely out of any reasonable tolerance.
Long and difficult story/discussion short, for this particular driver not only was pink noise not usable, but even the voltage levels on the speaker itself mattered a lot to get and measure the same results.
In general results will always differ somewhat. Around 10-15%, depending on the way you measure things. In the case we are talking about 30-40%, which is kind of insane.
End of the story is that this shift in Fs and Qt was even noticeable when music was played on low levels.
In my book this is called non-linear behavior.
But did I already mention the chicken-egg discussion?
Keep in mind that ALL manufacturers cherry pick their data.
This often results in a sort of chicken-egg discussion with the manufacturer.
Very recently I had something similar with a very known hifi speaker manufacturer.
One I know that they clearly know what they are doing.
I always double check T/S parameters from any datasheet. More to just get a feel about what brands I can trust and what brands are making a total mess.
In this particular case the parameters were way off.
Fs and Qt were both completely out of any reasonable tolerance.
Long and difficult story/discussion short, for this particular driver not only was pink noise not usable, but even the voltage levels on the speaker itself mattered a lot to get and measure the same results.
In general results will always differ somewhat. Around 10-15%, depending on the way you measure things. In the case we are talking about 30-40%, which is kind of insane.
End of the story is that this shift in Fs and Qt was even noticeable when music was played on low levels.
In my book this is called non-linear behavior.
But did I already mention the chicken-egg discussion?
Depends also how the the software was designed
and what its looking for to be rejected.
And even simple programs like WinIsd will reject drivers.
But the information provided by the manufacture is accurate.
Its basic math, and to what decimal point the software wants to see.
Sometimes say Qts is listed as .64 on the datasheet.
But mathematically the software wants to see a Qts .638 based on other entered parameters.
So the software calls out a error, but in reality the difference is minimal.
and I wouldnt call the manufacture having poor data.
They say .64 but software wants to see .638
Be better if the software showed you the entered numbers and the expected numbers.
So you can make the changes yourself or decide if the data is way off.
Or just slightly different and be able to let the numbers it wants to see work.
Of course if the numbers are actually really bad from the manufacture.
then you can decide to use the driver or not.
I have seen videos of the Eminence factory where they test the drivers and if out
of tolerance by a certain % they are rejected.
Assume possible some manufactures might not have as tight tolerances.
Or possible hard hits or rough handling in shipping may cause damage
to have parameters with a higher % of error.
Or if shipped by ocean or sits in a container for extended periods.
There can be moisture or unseen salt damage.
Who knows.
and what its looking for to be rejected.
And even simple programs like WinIsd will reject drivers.
But the information provided by the manufacture is accurate.
Its basic math, and to what decimal point the software wants to see.
Sometimes say Qts is listed as .64 on the datasheet.
But mathematically the software wants to see a Qts .638 based on other entered parameters.
So the software calls out a error, but in reality the difference is minimal.
and I wouldnt call the manufacture having poor data.
They say .64 but software wants to see .638
Be better if the software showed you the entered numbers and the expected numbers.
So you can make the changes yourself or decide if the data is way off.
Or just slightly different and be able to let the numbers it wants to see work.
Of course if the numbers are actually really bad from the manufacture.
then you can decide to use the driver or not.
I have seen videos of the Eminence factory where they test the drivers and if out
of tolerance by a certain % they are rejected.
Assume possible some manufactures might not have as tight tolerances.
Or possible hard hits or rough handling in shipping may cause damage
to have parameters with a higher % of error.
Or if shipped by ocean or sits in a container for extended periods.
There can be moisture or unseen salt damage.
Who knows.
Software doesn't "reject" anything.
They work with the same formulas as how to measure and calculate those T/S parameters.
I have simulated thousands of drivers in programs like WinISD (and other ones), never came across any issues. Two digits in general is enough btw.
I also have measured and evaluated T/S of more than hundreds of speakers myself.
They all rely on just a few base parameters, (Qms, Qes, Re, Sd and VAS), the rest is all derived from that.
What I did come across on a regular basis, are mistakes in datasheets.
Either typos, or total bogus parameters (read: HEAVILY cherry picked).
Otherwise it can be a difference in measuring method.
It matters a lot when a very low level pink noise is being used, or a sine wave sweep at a very specific voltage.
Maybe I'll actually write an article about it one day, because with T/S, nothing is what it seems.
Moisture and salt damage is EXTREMELY rare.
They work with the same formulas as how to measure and calculate those T/S parameters.
I have simulated thousands of drivers in programs like WinISD (and other ones), never came across any issues. Two digits in general is enough btw.
I also have measured and evaluated T/S of more than hundreds of speakers myself.
They all rely on just a few base parameters, (Qms, Qes, Re, Sd and VAS), the rest is all derived from that.
What I did come across on a regular basis, are mistakes in datasheets.
Either typos, or total bogus parameters (read: HEAVILY cherry picked).
Otherwise it can be a difference in measuring method.
It matters a lot when a very low level pink noise is being used, or a sine wave sweep at a very specific voltage.
Maybe I'll actually write an article about it one day, because with T/S, nothing is what it seems.
Moisture and salt damage is EXTREMELY rare.
Motor strength of production units is often FAR less than initial prototypes. Make something good, then find a way to make it cheaper. I’ve often seen higher advertised Qts numbers for the same driver, a couple years after it was introduced. Sometimes they don’t even bother to update the specs. If it looks too good to be true for the money it usually is.
What happens is the magnetism inside the magnet can go down after a while.
This is even more true, when those boxes are being thrown all over the place.
In general this is easy to find, since the Fs shouldn't be affected much, mostly just the Qes.
But it can be a bit tricky, since some suspensions also harden over time.
Although in that case the Fs as well as Qms change.
So after a while it becomes a bit of a guessing game what happened.
And how manufactures are determining their T/S.
What I have seen recently, is that manufactures derive the Re from the AC impedance.
So to be very clear about the definition, the Re is the resistance at DC.
That is just how the definition was being created.
Deriving it from the impedance curve, WILL give you a different value.
What happens is that you (once again) find yourself in a chicken-egg discussion,
because the derived Re (from AC impedance) will give a more true simulation curve. There are also people on this forum who say this.
While this is totally true and I even agree that it might be a better method, in my books that is just still very misleading and misuse of the definition for Re.
(back at my study, this could actually lead to an immediate fail of your thesis in some cases, just saying).
Call it Re(AC) or something else, I don't care.
But it HAS to be clear what method was being used.
But what happens is that people now use whatever they want and the reader of the datasheet doesn't have a clue at all what method is being used.
I know this for sure about some Seas datasheets.
In the end it is frustrating as hell.
The solution?
Measure and verify the T/S yourself with a couple of different methods, it's not that difficult takes up less than half an hour.
Another solution would looking at third party websites.
Since some of them already spend some time measuring and characterizing different drivers anyway, I am hoping they could include these things in their routine as well.
Better would be to get a proper standard that companies have to comply to.
Measure and verify the T/S yourself with a couple of different methods, it's not that difficult takes up less than half an hour.
Another solution would looking at third party websites.
Since some of them already spend some time measuring and characterizing different drivers anyway, I am hoping they could include these things in their routine as well.
Better would be to get a proper standard that companies have to comply to.