AVS Forum banner
1 - 10 of 10 Posts

·
Registered
Joined
·
2,260 Posts
Discussion Starter · #1 ·
Two questions about driver measurement that got no response in the DIY section.

First, T/S parameters are "small". I have seen no definition of small. Small for a 4 inch mid could be very different that for a 18 inch sub. I cannot find anything in any of my documentation, AES, D'Apolitto, Bullock etc.


Second, Measuring VAS using delta compliance. Is there a rule of thumb on the chamber size? Possibly close to the expected VAS? A ratio of fs? Kind of like trying to get over 25% shift for delta mass. I have noticed that the results tend to vary more when Vb is more than twice VAS.
 

·
Registered
Joined
·
4,649 Posts
The "Small" in Thiele-Small is the name of one of the originators...


Rule of thumb I have seen is to use a cube the size of the speaker being measured, e.g. a 1 foot cube for a 12" speaker. It just has to be a known volume.


Some of the parameters are small signal parameters, some are large signal parameters, and some are neither.


The small signals used to measure these parameters are small enough (low enough in amplitude) that the mechanical behavior of the speaker being measured is essentially linear. You have a lot of latitude in what to use. And yes, what will work for one speaker may not work for another speaker.
 

·
Registered
Joined
·
882 Posts
Yeah, in mathematics (and engineering) the word "small" is generally used semi-formally to mean something along the lines of "sufficiently small that the particular object being studied looks linear", or put another way, "small enough that the first order approximation is true up to negligible error". Here it's being used to mean that the driver's motion depends linearly on the input signal. Double the signal => double the output; add two signals => driver's motion is a sum of the two separate motions. FWIW I've also seen people use "small" in the context of higher-order nonlinear models, to mean "small enough that my model holds up to negligible error".


In this case you might also think of it like "small enough that none of the large signal parameters will make a difference". For example, the driver becomes hugely nonlinear as you give it a signal that makes excursion approach xmax. So any signal causing that much excursion is not small, by definition of small.
 

·
Registered
Joined
·
2,260 Posts
Discussion Starter · #4 ·

Quote:
Originally Posted by SpectralD /forum/post/17018243


Yeah, in mathematics (and engineering) the word "small" is generally used semi-formally to mean something along the lines of "sufficiently small that the particular object being studied looks linear", or put another way, "small enough that the first order approximation is true up to negligible error". Here it's being used to mean that the driver's motion depends linearly on the input signal. Double the signal => double the output; add two signals => driver's motion is a sum of the two separate motions. FWIW I've also seen people use "small" in the context of higher-order nonlinear models, to mean "small enough that my model holds up to negligible error".


In this case you might also think of it like "small enough that none of the large signal parameters will make a difference". For example, the driver becomes hugely nonlinear as you give it a signal that makes excursion approach xmax. So any signal causing that much excursion is not small, by definition of small.

You got the crux of the question exactly. How small is too small (insuficent suspension action) and how large is no longer small for a given driver. I was wondering if anyone has done any testing to come up with a generic close rule, like one tenth of Pe.

If I change from 100% to 50% (not measured the actual values) of my WT2, I do get different results. How do I pick a level that best matches the models? Should we be advocating for a standard value so that published and user measured values are more in line? ( like making note of the method delta mass or delta compliance along with the added mass and chamber volume in the spec.

Example: I first used delta mass to measure the VAS of a small driver. It came out as 6l where the spec was 5. A tad larger difference than "optimistic". Measured delta compliance in 8l. Came out as 5.5. Not too bad. Found a reference the OEM measured at 4.5, so I reduced my chamber to 5. It then measured 5.3. I reduced the drive to 25% and repeated. 5.2 All were done with identical preconditioning.


We need someone to do it with a wide range of drivers, from little 4 full ranges to big hoonkin 18 pro's. Plot it all out and give us a guideline!
 

·
Registered
Joined
·
2,260 Posts
Discussion Starter · #5 ·
And yes, I have searched through Dick Small's AES papers for a guide. I do not have a copy of his full big blue volume. The only one I have seen is with a friend in Colorado.
 

·
Registered
Joined
·
882 Posts
Seems to me like you want to start a thread in the DIY section about how to determine the linear operating range of a driver, and what sort of nonlinearities occur at either end. That might get the responses you're looking for.
 

·
Registered
Joined
·
2,260 Posts
Discussion Starter · #7 ·
I did. Slow getting some action, so I tried "theory" thread. I know most of the theory here is practical application stuff, but why not give it a shot? Besides, maybe it will drage a few over to DIY and we can get them interested in building better sound.
 

·
Registered
Joined
·
2,260 Posts
Discussion Starter · #9 ·
Yea, titles matter. I asked a question about VAS and I think everyone assumed I was talking about our most entertaining friend, not measuring equivalent compliance!
 

·
Registered
Joined
·
428 Posts
I looked in Dr. Leach's electroacoustics book, and he recommends keeping the AC current to the speaker constant at 10 mA RMS, though he gives no theoretical justification for it. This is messy though, as in general it requires changing the amplitude of the source each time its frequency changes.


I looked there for some recommendations on box volume for the VAS measurement, but there were none. However, he does say, "The internal dimensions of the test box should not exceed lambda/16 to lambda/8 at the test frequencies". Presumably this is because larger boxes must be modeled as a series compliance and mass element, rather than just a compliance element as he describes in section 3.3 on the plane wave tube.
 
1 - 10 of 10 Posts
Top