Speaker Impedance measurement ,circuit and sound card output

[icbcodc]

Free air means the driver should not be in a box or baffle or very close to a surface. It is best if it is supported vertically, as it would be when in a speaker.

O.K. I will try to make it vertical when I do a measurement.

 

[icbcodc]

Free air means the driver should not be in a box or baffle or very close to a surface. It is best if it is supported vertically, as it would be when in a speaker.

Hello John.

I've made a shelf to support the speaker vertially, and made a impedance measurement.

But when I'm calculating the

Free air means the driver should not be in a box or baffle or very close to a surface. It is best if it is supported vertically, as it would be when in a speaker.

Hello John.

I'm back for some questions.

I made five soundcard calibrations in preference menu,the five results are with minor difference.

Five curves

Do these minor differences matter?

Do I need to "make cal file" in Preferences menu for further impedance measurement calibaration?

 

[John Mulcahy]

Do these minor differences matter?

No.

Do I need to "make cal file" in Preferences menu for further impedance measurement calibaration?

No, impedance calibration uses different calibration files that are generated by REW when you go through the open circuit/short circuit/reference resistor impedance calibration steps

 

[icbcodc]

No.

No, impedance calibration uses different calibration files that are generated by REW when you go through the open circuit/short circuit/reference resistor impedance calibration steps

Clear.

Thank you john.

I made the speaker vertically when measuring.

I use this kind of mass from the car damping material.

I added mass like this.

Impedance curve

T-S Parameters without break in:

T-S paremeters from vendor without break in:

They are quite different.

Should I make the supporter firmer,it can swing a little bit cause I cut too much of the board.

 

[John Mulcahy]

The Mms and Bl figures are similar, even though that looks like a lot of mass to add. Probably better with closer to 14g and, if you make a second added mass measurement for a dual added mass result, about 7g for that. Easiest to split the mass into 4 parts so two can be removed for the second measurement. Attach them closer to the centre than the edges.

 

[icbcodc]

The Mms and Bl figures are similar, even though that looks like a lot of mass to add. Probably better with closer to 14g and, if you make a second added mass measurement for a dual added mass result, about 7g for that. Easiest to split the mass into 4 parts so two can be removed for the second measurement. Attach them closer to the centre than the edges.

Thank you for you explaination.

So you mean I should try twice for different added mass value: 7g and 14g, and split 14g to 4 parts, 2 parts for 7g, 4 parts for 14g?

 

[John Mulcahy]

Yes. Measure first with all 4 pieces, then with two (on opposite sides of the cone), then none.

 

[icbcodc]

Yes. Measure first with all 4 pieces, then with two (on opposite sides of the cone), then none.

Good idea,I'll try it tomorrow.

 

[icbcodc]

Yes. Measure first with all 4 pieces, then with two (on opposite sides of the cone), then none.

Hi John

I made 4 pieces

2piecees

another 2 pieces

7g mass

14g mass

Impedance measurement, this time I didn't do calibration again,just use the data saved in REW system.

T/S parameters with 14g Mass T/S parameters with 7g Mass

Sd is 123.3 cm2, cause the center of the cone is not movable,so its area is not contained in Sd.

T/S from vendor, they are still different, is there something wrong in my measurement?

 

[Bernard]

Hi John

I made 4 pieces
View attachment 32271

2piecees

View attachment 32272

another 2 pieces

View attachment 32273

7g mass

View attachment 32274

14g mass

View attachment 32275

Impedance measurement, this time I didn't do calibration again,just use the data saved in REW system.

View attachment 32276

T/S parameters with 14g Mass T/S parameters with 7g Mass
View attachment 32277

Sd is 123.3 cm2, cause the center of the cone is not movable,so its area is not contained in Sd.

T/S from vendor, they are still different, is there something wrong in my measurement?
View attachment 32279

Hi icbcodc,

What an interesting thread!
You wonder if your results are right. They are because you have to know that you can only compare what is comparable. We never get the same results as the manufacturer because the conditions are different.
You measure a loudspeaker (A) in a room (B) and the manufacturer measures a loudspeaker (a') in a room (b'). You use the REW software (D) with a sound card (E) under a temperature (F) and the manufacturer uses the LMS software (d') with a sound card (e') under a T° (f').
The ambient temperature T° changes the speed of sound C and it also has its influence.
C is calculated with the formula C = 331.3 × square root (1 + (T° / 273.15)) .

Your measurements are accurate and your perseverance is remarkable. By following your thread, I can see my own mistakes and questions. It is an instructive thread.

With your two curves, all that remains is to calculate the TS parameters of your loudspeaker with the dual-mass method.
You will inevitably obtain TS data different from those of the manufacturer and it will be normal.

When you measure your second loudspeaker, the data will be different from the first one and this will be normal too. This is due to manufacturing differences.

Well done and keep up the good work.

 

[icbcodc]

Hi icbcodc,

What an interesting thread!
You wonder if your results are right. They are because you have to know that you can only compare what is comparable. We never get the same results as the manufacturer because the conditions are different.
You measure a loudspeaker (A) in a room (B) and the manufacturer measures a loudspeaker (a') in a room (b'). You use the REW software (D) with a sound card (E) under a temperature (F) and the manufacturer uses the LMS software (d') with a sound card (e') under a T° (f').
The ambient temperature T° changes the speed of sound C and it also has its influence.
C is calculated with the formula C = 331.3 × square root (1 + (T° / 273.15)) .

Your measurements are accurate and your perseverance is remarkable. By following your thread, I can see my own mistakes and questions. It is an instructive thread.

With your two curves, all that remains is to calculate the TS parameters of your loudspeaker with the dual-mass method.
You will inevitably obtain TS data different from those of the manufacturer and it will be normal.

When you measure your second loudspeaker, the data will be different from the first one and this will be normal too. This is due to manufacturing differences.

Well done and keep up the good work.

Thank you Bernard for your encouragement.

Yes, you're right, the value should not be the same,but I think the values of QTS and F0 are too different.

AND I know VAS is hard to be measured accurately, so 20% difference is acceptable.

And I found I made a misunderstanding, there is function called "dual-mass method" in REW, I did not know it.

I will try it tomorrow again.

Do you think the supporter board of the speaker should be firmer? cause it is swing a tiny when I push the top up side of it.

 

[John Mulcahy]

You can use the measurements you have to run the results for dual added mass. That will also produce a plot of Bl on the Impedance & Phase graph, it should be flat in the region around Fs if the measurements are good. You could add some diagonal struts from the back and front of the base to the region near the driver mount to stiffen the frame.

 

[icbcodc]

You can use the measurements you have to run the results for dual added mass. That will also produce a plot of Bl on the Impedance & Phase graph, it should be flat in the region around Fs if the measurements are good. You could add some diagonal struts from the back and front of the base to the region near the driver mount to stiffen the frame.

Hi John
I rerun the Thiele-Small Parameters with "Dual added mass" method using the impedance information I measured this morning.

Fs is not changed, Vas (10.90 litres) looks better，Qts (0.708) looks

and the BL curve is not available

I will stiffen the frame tomorrow.

 

[John Mulcahy]

Can you attach the mdat file with the 3 measurements? Not sure why Bl should be N/A.

 

[icbcodc]

Can you attach the mdat file with the 3 measurements? Not sure why Bl should be N/A.

Of course.

FYI.

 

[icbcodc]

Can you attach the mdat file with the 3 measurements? Not sure why Bl should be N/A.

I'm reading the following famous book by Joseph D'Appolito again and try to get some guidance.

So the "Added Mass method" is for Vas calculating, three basic equations list below:

So in T/S parameter, the quality of cone diagram should be Mmd or Mms?

In his book:

1. the quality of added mass should not be larger than that of cone diagram,and should be larger than 60% of that of cone diagram.

2. The quality of dded mass should keep accuracy like 0.1g at least.

3. When using added mass,the new Fsam value should be less than 75% of free air Fs, if not, increase the quality of added mass till Fsam down to 75% of Fs。

4. The quality of added mass should not be larger than that of cone diagram。

So, in my case, the Mms value of QS-216 is around 13g, so 60% percent of 12g should be 7.2g, so the quality of added mass should between 7.2g to 13g。

Maybe 12g or 12.5g added mass is better than 7g or 14g?

 

[John Mulcahy]

Thanks. I have fixed the Bl display for the next beta release, however there is a problem with the measurements. The motional model fit is very different between them, I think the most likely reason is that the measurements of the added masses are not accurate. Even small errors in the mass figures make a large difference to the results, a precision scale is needed and when removing the masses from the cone it should be moved as little as possible.

 

[John Mulcahy]

On your question about masses, using smaller masses is OK, but not too small, for dual added mass I would advise not less than 25% of Mms for mass 1 and not less than 50% for mass 2. There has been a lot of research on more accurate motional and impedance models for drivers since that book was published, and more accurate methods for fitting those models to the driver impedance measurements. Some relevant papers are listed in the REW help. The dual added mass method is one of the results of that research. The model fitting methods used in REW and other modern software do not require such large changes in the resonant frequency.

 

[icbcodc]

On your question about masses, using smaller masses is OK, but not too small, for dual added mass I would advise not less than 25% of Mms for mass 1 and not less than 50% for mass 2. There has been a lot of research on more accurate motional and impedance models for drivers since that book was published, and more accurate methods for fitting those models to the driver impedance measurements. Some relevant papers are listed in the REW help. The dual added mass method is one of the results of that research. The model fitting methods used in REW and other modern software do not require such large changes in the resonant frequency.

Yes, thank you for your advises.

I think Dual added mass is a new way, the latest chinese version of 《Testing Loud Speakers》 is based on the English version in 1998,two decades ago,that's awful for people like me. Even about REW, we don't have much latest information in Chinese.

If you didn't help me,I couldn't have gone so far.

I've found the 1000ohm Rsense in his circuit is bad for impedance measurement in REW working with a USB soundcard. So I need to refer to his book carefully.

Thank you for making REW a super cool software, I used REW to test my room acoustic and optimized it last year.

BTW, I draw a circuit and a formula, is this the basic way you calculate the Impedance in REW?

 

[John Mulcahy]

BTW, I draw a circuit and a formula, is this the basic way you calculate the Impedance in REW?

That is the principle, but the input impedance of the soundcard, the series and shunt impedance of the test leads and the frequency-dependent gain difference between the channels must also be included so the actual calculation is more complicated.

 

[icbcodc]

That is the principle, but the input impedance of the soundcard, the series and shunt impedance of the test leads and the frequency-dependent gain difference between the channels must also be included so the actual calculation is more complicated.

Yes, I understand,the real world is always complicated.

This thread will be a good tutorial or FAQ for me.

 

[Bernard]

Hello John,

About the dual added mass method, on this page of their site: https://speakerbench.com/manual/#measure , Mr. Futtrup and Mr. Candy say :
We recommend a stepped-sine signal measurement such that all frequency points are measured under steady-state conditions. It is our experience that a fast Chirp or Farina sweep will not result in acceptable precision. If your equipment only offers sweep options, we recommend a slow sweep to achieve reasonably high accuracy. Note: REW and DATS V3 both use a sweep.

How slow do you think the sweep should be ?

Thank you.

 

[John Mulcahy]

I discussed that comment with Claus and did testing with all REW's sweep lengths and found no difference for sweeps above 5 seconds (256k at 48 kHz sample rate, for example). The noise filter was more important for obtaining reliable results, since sweeps are more susceptible to the effects of external noise. REW already shows a warning if the selected sweep is too short.

 

[Bernard]

I discussed that comment with Claus and did testing with all REW's sweep lengths and found no difference for sweeps above 5 seconds (256k at 48 kHz sample rate, for example). The noise filter was more important for obtaining reliable results, since sweeps are more susceptible to the effects of external noise. REW already shows a warning if the selected sweep is too short.

Thank you, John. It's good to know that.

 

[icbcodc]

Thanks. I have fixed the Bl display for the next beta release, however there is a problem with the measurements. The motional model fit is very different between them, I think the most likely reason is that the measurements of the added masses are not accurate. Even small errors in the mass figures make a large difference to the results, a precision scale is needed and when removing the masses from the cone it should be moved as little as possible.

Hello John
I've stiffened the frame,it's firm now.

the mass quality is not very accurate, it's accuracy is around 0.1g-0.2g.

The accuracy of DC resistance of speaker is around 0.1-0.2ohm。

Lastest T/S Parameters after stiffening the frame.

T/S parameters from vendor.

Is the frame for speaker good enough now?

Is it necessary to use a more acurate electronic scale？ such as 0.01g precision?