Thiele-Small Parameters measurement using dual added mass method.

Bernard

Member
Thread Starter
Joined
May 22, 2017
Messages
59
Hello,
icbcodc https://www.avnirvana.com/threads/speaker-impedance-measurement-circuit-and-sound-card-output.6860/post-50908 gave me the idea to measure one of my loudspeakers with this method and to try the Speakerbench control tool : https://speakerbench.com/apps/app1 .
Here is the result: " Fit quality: Sorry :-( "
I attach the zma files and the corresponding JSON file. The test can last one to two minutes.
A 40 g mass was used because it roughly corresponds to the Mms I always got with my old measurements.
It seems that I have to use a lower mass.
I put a voltage of 0.316 V corresponding to -10 dB FS ? Is it right ?
With this method, the T° has no influence on the calculation of Vas but the atmospheric pressure (mB) does.
Next measurements next week!
 

Attachments

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
3,180
40g does seem high, perhaps try 15g and 30g. Best to check the voltage at the driver terminals with a sine wave (100 Hz, say) and make sure it isn't too high, but I would expect it is OK if 316 mV is the output voltage.

What resolution did you use to export the.zma files? It appears a little coarse.
 

Bernard

Member
Thread Starter
Joined
May 22, 2017
Messages
59
40g does seem high, perhaps try 15g and 30g. Best to check the voltage at the driver terminals with a sine wave (100 Hz, say) and make sure it isn't too high, but I would expect it is OK if 316 mV is the output voltage.

What resolution did you use to export the.zma files? It appears a little coarse.
I haven't changed any of the export settings.
Would 48 PPO be enough?
 

icbcodc

Member
Joined
Apr 29, 2020
Messages
78
Hello,
icbcodc https://www.avnirvana.com/threads/speaker-impedance-measurement-circuit-and-sound-card-output.6860/post-50908 gave me the idea to measure one of my loudspeakers with this method and to try the Speakerbench control tool : https://speakerbench.com/apps/app1 .
Here is the result: " Fit quality: Sorry :-( "
I attach the zma files and the corresponding JSON file. The test can last one to two minutes.
A 40 g mass was used because it roughly corresponds to the Mms I always got with my old measurements.
It seems that I have to use a lower mass.
I put a voltage of 0.316 V corresponding to -10 dB FS ? Is it right ?
With this method, the T° has no influence on the calculation of Vas but the atmospheric pressure (mB) does.
Next measurements next week!
Sorry my English is poor, I don't understand how to use Speakerbench control tool and what it is for.

About Dual added mass method, I just followed John's advices, and I remember his basic principle is one mass > 25% MMS AND the other mass > 50% MMS.

T/S parameters are low signal parameters , so you don't have to use high V, but I'm not sure the accurate threshhold of low signal.

Sorry about that ,I'm a new guy.

You'd better ask John for it.
 

Bernard

Member
Thread Starter
Joined
May 22, 2017
Messages
59
I don't understand how to use Speakerbench control tool and what it is for.
How to use Speakerbench test ?
https://speakerbench.com/app
Click on MANUAL
Click on 6. Sample data files
On this line, click on JSON file to download the file named « Z_L16RNX.json » which contains the 3 ZMA files on your pc
Then return to https://speakerbench.com/app
Click on « 2. CALCULATE (FIT) DATA » or https://speakerbench.com/apps/app1
Then « Drag & Drop or Select JSON object » , « Z_L16RNX.json » in our example
The "Fit quality" tab returns a diagnosis of the measurements. « Model test » tab indicates the deviations between the data and the model.
If the impedance curves do not overlap, we have to measure again until the correct weights are found.

To create your own data file, you have to export the data of your 3 impedance measurements with ZMA extent.
In REW, to export a file in text mode, it's simple.
step 1: click on File
step 2: click on Export
step 3: click on Export measurement as text
Use custom range: 10 Hz to 10000 Hz
Use custom resolution: 96 PPO

When the 3 text files are created, they are renamed to zma by changing the txt extension to zma.
Then we go straight to the site https://speakerbench.com/apps/app0 to collect our data.
Speakerbench will create your Json container.

Let's go!
 

Attachments

icbcodc

Member
Joined
Apr 29, 2020
Messages
78
How to use Speakerbench test ?
https://speakerbench.com/app
Click on MANUAL
Click on 6. Sample data files
On this line, click on JSON file to download the file named « Z_L16RNX.json » which contains the 3 ZMA files on your pc
Then return to https://speakerbench.com/app
Click on « 2. CALCULATE (FIT) DATA » or https://speakerbench.com/apps/app1
Then « Drag & Drop or Select JSON object » , « Z_L16RNX.json » in our example
The "Fit quality" tab returns a diagnosis of the measurements. « Model test » tab indicates the deviations between the data and the model.
If the impedance curves do not overlap, we have to measure again until the correct weights are found.

To create your own data file, you have to export the data of your 3 impedance measurements with ZMA extent.
In REW, to export a file in text mode, it's simple.
step 1: click on File
step 2: click on Export
step 3: click on Export measurement as text
Use custom range: 10 Hz to 10000 Hz
Use custom resolution: 96 PPO

When the 3 text files are created, they are renamed to zma by changing the txt extension to zma.
Then we go straight to the site https://speakerbench.com/apps/app0 to collect our data.
Speakerbench will create your Json container.

Let's go!
So my understanding is this is a online verification tool which can verify the impedance measurement data generated by other software such as REW?

I think this online app can simulate the impedance plot according to the measurement data just like REW.

But REW has a simulation to verify the impedance measurement already, why should I try another tool?

Cause my English is poor and I'm not familiar with it and don't know whether it's working properly like REW either.
 

Bernard

Member
Thread Starter
Joined
May 22, 2017
Messages
59
don't know whether it's working properly like REW either.
Hello icbcodc,
The tool created by Mr. Futtrup and Mr. Candy makes it possible to check whether the impedance curves we created with REW correspond to their model developed to obtain the most accurate TS parameters with that new method.
The curves sent for verification show that the masses I used are too high.
Let me think that the two people quoted are as qualified as John.
Speakerbench displays a graphical representation of the data created by REW sent to the site.
Send your 3 zma files and send us the analysis result.
I have a full-mass error of 1.37. It's up to me to redo measurements with different masses until I find the right ones. That's all.
 

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
3,180
Claus and Jeff are the originators of the dual added mass method. The model used in SpeakerBench is the same as the model used in REW (Ritter 3PC), but they have some proprietary fit techniques which might give a different fit to REW for a given set of files,. For the test data I used the results were the same.
 

icbcodc

Member
Joined
Apr 29, 2020
Messages
78
Hello icbcodc,
The tool created by Mr. Futtrup and Mr. Candy makes it possible to check whether the impedance curves we created with REW correspond to their model developed to obtain the most accurate TS parameters with that new method.
The curves sent for verification show that the masses I used are too high.
Let me think that the two people quoted are as qualified as John.
Speakerbench displays a graphical representation of the data created by REW sent to the site.
Send your 3 zma files and send us the analysis result.
I have a full-mass error of 1.37. It's up to me to redo measurements with different masses until I find the right ones. That's all.
Hi Bernard

Thank you for your information.

I found the basic information of the methodology of DUAL MASS METHOD , 《An Added-Mass Measurement Technique for Transducer Parameter Estimation

There are two authors of this paper: Jeff Candy and Claus Futtrup , and the online tool "https://speakerbench.com/app " you recommended is also created by Jeff Candy and Claus Futtrup , so I think this tool is reliable enough,but just a little complex for me.

I will try it tomorrow.
 

icbcodc

Member
Joined
Apr 29, 2020
Messages
78
Hello icbcodc,
The tool created by Mr. Futtrup and Mr. Candy makes it possible to check whether the impedance curves we created with REW correspond to their model developed to obtain the most accurate TS parameters with that new method.
The curves sent for verification show that the masses I used are too high.
Let me think that the two people quoted are as qualified as John.
Speakerbench displays a graphical representation of the data created by REW sent to the site.
Send your 3 zma files and send us the analysis result.
I have a full-mass error of 1.37. It's up to me to redo measurements with different masses until I find the right ones. That's all.
Hi Bernard

Thank you for your information.

I found the basic information of the methodology of DUAL MASS METHOD , 《An Added-Mass Measurement Technique for Transducer Parameter Estimation

There are two authors of this paper: Jeff Candy and Claus Futtrup , and the online tool "https://speakerbench.com/app " you recommended is also created by Jeff Candy and Claus Futtrup , so I think this tool is reliable enough,but just a little complex for me.

I will try it tomorrow.
 

Bernard

Member
Thread Starter
Joined
May 22, 2017
Messages
59
Hi Bernard

Thank you for your information.

I found the basic information of the methodology of DUAL MASS METHOD , 《An Added-Mass Measurement Technique for Transducer Parameter Estimation

There are two authors of this paper: Jeff Candy and Claus Futtrup , and the online tool "https://speakerbench.com/app " you recommended is also created by Jeff Candy and Claus Futtrup , so I think this tool is reliable enough,but just a little complex for me.

I will try it tomorrow.
These informations can also be found in the REW help file at Thiele-Small Parameters. :)
John goes to the trouble of writing a top-notch help manual and most people don't read it. :(
 
Last edited:

icbcodc

Member
Joined
Apr 29, 2020
Messages
78
These informations can also be found in the REW help file at Thiele-Small Parameters. :)
John goes to the trouble of writing a top-notch help manual and most people don't read it. :(
Yes, John've done a lot hard work to make REW and help file more useful.

But sometimes I cann't understand the help file clearly cause my English is not good.

There are too many details for a new guy to know to confirm so that he can finish the T/S parameters from zero.

This forum gives a good chance to communicate online with John and you people, it's a wonderful experience.
 

Bernard

Member
Thread Starter
Joined
May 22, 2017
Messages
59
An on line translator such as DeepL may help you : https://www.deepl.com/translator

I put this sentence on left : Les traductions sont correctes dans les langues que je connais, le français, l'anglais, le polonais et l'allemand. Est-ce compréhensible dans votre langue, que je suppose être du chinois ?
DeepL has translated : 我所知道的语言,法语、英语、波兰语和德语的翻译都是正确的。这句话用你的语言可以理解吗,我猜是中文?
Then, I put that translated sentence in the left windows.
DeepL has translated in French: La traduction est correcte dans les langues que je connais, le français, l'anglais, le polonais et l'allemand. Est-ce compréhensible dans votre langue, je suppose que c'est du chinois ?

NB: I would like to point out that I don't own any shares DeepL
 

icbcodc

Member
Joined
Apr 29, 2020
Messages
78
An on line translator such as DeepL may help you : https://www.deepl.com/translator

I put this sentence on left : Les traductions sont correctes dans les langues que je connais, le français, l'anglais, le polonais et l'allemand. Est-ce compréhensible dans votre langue, que je suppose être du chinois ?
DeepL has translated : 我所知道的语言,法语、英语、波兰语和德语的翻译都是正确的。这句话用你的语言可以理解吗,我猜是中文?
Then, I put that translated sentence in the left windows.
DeepL has translated in French: La traduction est correcte dans les langues que je connais, le français, l'anglais, le polonais et l'allemand. Est-ce compréhensible dans votre langue, je suppose que c'est du chinois ?

NB: I would like to point out that I don't own any shares DeepL
Thanks Bernard.

It looks like a good translater, the Chinese version is correct. :-)
 

andyc56

Member
Joined
Jun 2, 2017
Messages
33
I found the basic information of the methodology of DUAL MASS METHOD , 《An Added-Mass Measurement Technique for Transducer Parameter Estimation

There are two authors of this paper: Jeff Candy and Claus Futtrup , and the online tool "https://speakerbench.com/app " you recommended is also created by Jeff Candy and Claus Futtrup , so I think this tool is reliable enough,but just a little complex for me.
I found a PDF of what looks like a preprint of this paper in case anybody is interested.
 

jcandy

New Member
Joined
Dec 3, 2017
Messages
3
In diagnosing what went wrong, I find the "Model Test" tab to be the most revealing (see attached image).

32755


Here you can see that the model describes the unweighted driver impedance well, but does poorly on the 20g and 40g measurements. My suggestions would be to reduce the masses (as John has suggested) to 30g and 15g, and make sure to measure the added masses accurately. Beyond that, I would clarify that it is important to:

- do the heaviest (double-mass) measurement first, then the single-mass, then unweighted
- remove masses very, very carefully between steps
- make sure applied voltage is constant between measurements

If you want to give it another try I'll be happy to look at the new data.
 

Bernard

Member
Thread Starter
Joined
May 22, 2017
Messages
59
If you want to give it another try I'll be happy to look at the new data.
I'll be happy to send you the results as soon as I receive a more accurate scale.
In the meantime, since I still have my 4 weights of 10 g, tomorrow I will measure both speakers with the little brother of the WT Pro, the Woofertester 2 which still works with USB audio codecs.
 

jcandy

New Member
Joined
Dec 3, 2017
Messages
3
Yes, this is much better. Should be fine for modeling purposes. What did you do differently?
 

Bernard

Member
Thread Starter
Joined
May 22, 2017
Messages
59
Yes, this is much better. Should be fine for modeling purposes. What did you do differently?
Three changes:
- reduction of the 4 Polos weight from 10 to 7.5 g each, following John's advice [ https://en.wikipedia.org/wiki/Polo_(confectionery) ];)
- sweep level at -8.00 dBFS by adjusting the gain knobs so that the peak indicators flashed briefly at the loudest input volume;
- measurements in the right order: M2, M1, M0
 

jcandy

New Member
Joined
Dec 3, 2017
Messages
3
Very good. Thanks for using Speakerbench, and thank alot to John for facilitating it.
 
Top Bottom