Inductance modelling

[3ll3d00d]

Interesting thread developing on diya - http://www.diyaudio.com/forums/subwoofers/320189-inductance-cancellation-techniques.html - regarding more accurately modelling the effect of inductance (and building that into hornresp making use of data that is captured by rew or arta).

[Ricci]

I've been following it. Not much time to post but I lurk over there a lot. 

[lilmike]

Following along with that too. I'm very interested in modeling inductance effects properly. I have a hard time changing things we can measure accurately to account for the stuff we're not looking at correctly, so I look forward to what comes of this. I'll hopefully be building and measuring a few things soon.

[3ll3d00d]

I see he's released a build with this feature in -> http://www.diyaudio.com/forums/subwoofers/119854-hornresp-post5404994.html

[lilmike]

Thanks for the heads-up.

Sadly, it looks like the WT2 data does not have the complex inductance parameters in a format Hornresp can use. 

I've sent them an email to see what  (if anything) can be done.

[3ll3d00d]

7 hours ago, lilmike said:

Thanks for the heads-up.

Sadly, it looks like the WT2 data does not have the complex inductance parameters in a format Hornresp can use. 

I've sent them an email to see what  (if anything) can be done.

you can import the measurement into REW and let it calculate them for you.

This is probably a nice thing to add to all the driver measurements from data-bass.

[lilmike]

Yeah, I suppose I could do it that way, if I had saved the sweeps. The WT2 doesn't save the individual sweeps like that. 

In my case, I could also just use ARTA or REW and measure things directly. 

[Ricci]

Will WT2 export text? That may be enough to import into REW and let it calculate.

[Ricci]

On ‎4‎/‎15‎/‎2018 at 4:41 AM, 3ll3d00d said:

you can import the measurement into REW and let it calculate them for you.

This is probably a nice thing to add to all the driver measurements from data-bass.

I have all of that it's just picking a format to report it in. LIMP offers several. Also the website isn't setup to accept it. I'd have to put it in the notes or on the forums.

[SME]

IIRC, WT2 exports to ZMA, which REW can import.  I just tried it on one of my saved measurements, but I can't make REW calculate the parameters because all I have is an impedance measurement of the driver in the box.  To calculate the inductance parameters, it must also calculate the rest of the T/S parameters which requires a free-air baseline and another measurement with added mass or in a test box of known volume.

[mwmkravchenko]

The complex inductance capability is finally incorporated in to Hornresp. Nice to see. Much of the effort belongs to bolserst and David McBean.

It's a situation of providing the correct frame work and the proper papers behind that framework. Hornresp is getting pretty interesting these past few months. Moving closer to LEAP kind of computational simulation capabilities.

[3ll3d00d]

I gave it a go with some DATS measurements for a FaitalPro 12RS1066 but get a giant value for Rss (see https://www.avnirvana.com/threads/imported-impedance-measurements-produces-giant-value-for-rss.2687/ for details). I can't say I was super careful with the z measurements but the calculated data seems reasonable with respect to the published params so not sure what is going on.

[SME]

Per the response there, it looks like your RSS is effectively infinite.  That may be perfectly fine for the model, but I haven't looked at the details.

I agree with the comment about the added mass being too high.  IIRC, you want something like 10-20% of Mms or else the error are potentially high.  Also, one issue with impedance measurements that may or may not apply here is that they aren't accurate unless the driver is mounted to a very rigid baffle and in a forward firing orientation.  In an up-firing orientation as when set on a bench top, the parameters can come out very wrong.  That's why I don't have any free-air impedance sweeps of any of the drivers I bought.

[3ll3d00d]

That was the weight it took to reduce Fs by about 25% which is (as far as I recall) approximately what you are aiming for. I don't believe 10-20÷ of mms is a target I have ever seen mentioned either, that seems much lower than I recall being used. (Edit: the "testing loudspeakers" book advises to start with 60% of cone mass which is about 100g, not much less than the actual weight used)

I am aware of the assorted issues around applying the delta mass method though, as commented in that thread, the resulting basic params are not wildly out of spec.

[SME]

I recalled wrong.  I was probably thinking about how much to reduce Fs by and not how much delta Mms to have.  Thanks for the correction.

Adding an Mms of 60% of cone mass should reduce Fs by 20%, so that seems to be about the right ballpark.

[lilmike]

I did hear back from the WT2 devs. They're researching Thorborg's approach and evaluating adding that capability to the software.

If I do a custom impedance sweep with the WT2, I can export the data as text. That's pretty much the same thing as I'd do with ARTA or REW though, so why not just use those tools?

[Ricci]

I've been talking with Bolserst through email. He's the guy doing a lot of the leg work on this with David McBean (HornResponse creator) over at DIYAudio forum. I've sent him some measurement files that he is going to look at. This is another one of those community projects that might turn out to benefit a whole lot of people. 

[deepthoughts]

On 4/24/2018 at 1:27 AM, lilmike said:

I did hear back from the WT2 devs. They're researching Thorborg's approach and evaluating adding that capability to the software.

If I do a custom impedance sweep with the WT2, I can export the data as text. That's pretty much the same thing as I'd do with ARTA or REW though, so why not just use those tools?

Taking any impedance measurement with WT2 or DATS gives simple text files.  They might classify it with their own suffix, but you can rename any of them as text or other types of files after opening in a generic text editor.  Most often the hiccups come from errant header info that might miss a " or * or simply isn't formatted as the program is looking for.  That's really all the different suffixes are telling us... what the header info might contain.

I'll have to do some hunting through the thread unless anyone can point to a summary of how they are calculating and how many elements they are using for the complex Le model.

I know in my own experience, until you have a very complex model, the most useful results come from the ability to make adjustments to best correlate with a simple, near field measurement of the driver in a known sealed box or similar.  Particularly where shorting rings and sleeves are used, you have to pick what parts of the curve you want to match related to which parts of the frequency response interest you most.  I look forward to seeing what they have put together for the latest update.

[3ll3d00d]

@deepthoughts I think http://www.diyaudio.com/forums/subwoofers/320189-inductance-cancellation-techniques.html#post5390118 summarises it and https://www.roomeqwizard.com/help/help_en-GB/html/thielesmall.html#top gives the full name of the Thorborg paper

[mwmkravchenko]

There are a few more papers other than the Thorborg paper that Bolserst used. If anyone wants the papers let me know and I'll be happy to forward them to you.

[deepthoughts]

2 hours ago, 3ll3d00d said:

@deepthoughts I think http://www.diyaudio.com/forums/subwoofers/320189-inductance-cancellation-techniques.html#post5390118 summarises it and https://www.roomeqwizard.com/help/help_en-GB/html/thielesmall.html#top gives the full name of the Thorborg paper

Thanks for the searching.  A quick scan looks like the models add/account for the missing aspects not represented in a simple 3 element, shorted inductance model which LspCAD has had since version 5.25.  One bit not immediately clear is the units of the semi-inductance, Ke.  Does this just come through as a parallel inductance to Le which reduces to an inductance and resistance in parallel or is Ke somehow variable with frequency?  I thought I caught mention of Ke varying with the square root of frequency.  Is this figured/modeled as a frequency dependent inductance element?

 

-Mark

[lilmike]

12 hours ago, deepthoughts said:

Taking any impedance measurement with WT2 or DATS gives simple text files.  They might classify it with their own suffix, but you can rename any of them as text or other types of files after opening in a generic text editor.  Most often the hiccups come from errant header info that might miss a " or * or simply isn't formatted as the program is looking for.  That's really all the different suffixes are telling us... what the header info might contain.

...

I was hoping that the WT2 devs would be able to incorporate Thorborg's approach into their calculated T/S parameters so I could get semi-inductance parameters for HornResp, so I asked them to consider it. Keith got back to me and said that the calculation of the parameters is trivial, the complicated part is incorporating the changed approach into the rest of their simulator. 

If they can, great. If not, I am not too fussed, as I also have ARTA and could always use REW.

Now, I just need some time to actually build and measure speakers...

[mwmkravchenko]

I completely forgot about this spreadsheet.

Thorborg advanced inductance spreadsheet

If I remember half of what I have forgotten, I'd be more dangerous