SB Acoustics TS parameters dreadfully inaccurate

Jump to Latest
C

cs

Member
Joined 2005
Tweet said:
That would give an internal series resistance of 101.5 ohms and an output current of only 12.554 milliamps. Now would that be sufficient to drive a large woofer and get accurate T/S parameter measurements ?

C.M
Click to expand...

I guess the problem is how do you define 'accurate', and what test level do you design for ? For the SB drivers considered here, the Qts varies from around 0.4 at 1-2V test level, but nearly 0.6 at 0.1V input. They require completely different port tunings (or no port at all).
 
H

hollowboy

Member
Joined 2007
Tweet said:
{...] an output current of only 12.554 milliamps. Now would that be sufficient to drive a large woofer and get accurate T/S parameter measurements ?
Click to expand...

By definition, yes:

"These are the physical parameters of a loudspeaker driver, as measured at small signal levels"

cs said:
I guess the problem is how do you define 'accurate', and what test level do you design for ? For the SB drivers considered here, the Qts varies from around 0.4 at 1-2V test level, but nearly 0.6 at 0.1V input. They require completely different port tunings (or no port at all).
Click to expand...

That could potentially be a good thing: if the system is ~flat at high power, and high Q at low power, that's a built-in loudness curve.

speculating on this topic
 
GDO

GDO

Member
Joined 2009
hollowboy said:
That could potentially be a good thing: if the system is ~flat at high power, and high Q at low power, that's a built-in loudness curve.
Click to expand...

Mmmm... If the loudspeaker is driven somewhat hard, the voice coil will heat up, making some parameters drift, as explained in this paper from PHL AUDIO ... in french.

http://cyrille.pinton.free.fr/electroac/lectures_utiles/lesage.pdf

It will be noted that the electrical coefficient QES varies in proportion to Δθ. QTS varies in a lesser proportion because QMS remains constant. It is as if the strength of the magnetic circuit decreased with increasing temperature of the coil.

Thus, if a 66-liter Bass-Reflex load given at 58 Hz is the best choice for
operation at room temperature, the enclosure should be enlarged and the vent should require a lower frequency tuning when the operating temperature of the coil rises ...

This is obviously not possible, but if we choose a volume Vb of 85 liters given to Fb =42 Hz, we see that in the cold one has something intermediate between a B4 and C4 to become
A QB3 more rounded at the cut between 100 and 150 ° C.
Click to expand...

In a word, BR tuning is more about tweaking than rocket science!:yikes:
 
Last edited:
GDO

GDO

Member
Joined 2009
Sorry but i am no idealist and prefer to rephrase:

Naive CAD cannot be accurate, and real world has not to be blamed for its flaws...:D

And TS culture is both naive or misunderstood. Like Dats and stuff like that: just too good to be true...
 
Last edited:
F

Fernando R

Member
Joined 2017
Speaker box design is rather tolerant to compliance variations. As long as the resulting Fs variations maintain a proportional variation with Q values things should be OK. Things even out and the impact in the final box response is minimal. What screws up a design is when Mms variations or BL variations change the Fs and Qt relationship. At that point you have a different driver.


Compliance variations can be the result of break in or test levels as well as production. Large Mms, Re or BL variations are production issues and no amount of break-in or change in test voltages/currents will resolve that.
 
D

DaveFred

Member
Joined 2018
Paid Member
cs said:
I have now returned the drivers to SB, and they have tested them. What they found is that the measured TS parameters are different to those on the datasheet, but quite close to those on the datasheet for the SB17NAC (natural aluminium, not black) version. It seems that the NBAC datasheet is preliminary, and not correct.

My measurements were still quite different (higher fs and Qts) even from the NAC data, so there is still a discrepancy. They also did measurements at a much lower level (0.1V), and that gave values much closer to my results. This pointed the finger at the DATS using too low a signal level, but it is set to the default of +4dBu (1.23Vrms) so it should be OK.

As a sanity check, I also tested some Scanspeak Discovery woofers I had purchased previously (18W-8434G00) with the DATS, and they measured very close to the SS datasheet. Interestingly, Ulrik at SB said that whereas SB always use a constant voltage to measure their drivers, Scanspeak always use constant current.
Click to expand...

And they are still using that data sheet.

I found this thread after measuring with DATS V3 and also finding the TS parameters way off on the SB17NABC.

Did you ever use them in a build?

What alignment did you go with?
(sealed/ported/liters/vent size/etc.)

Thank you,

David.
 
L

leon2216

Member
Joined 2020
I also found this problem. I think it is because DATS cannot measure DCR correctly, which is obviously too large compared with the DCR measured by the bridge, resulting in the larger QES during calculation, thus improving THE QTS. I used NE315-8 to test, and the DCR measured by the bridge was 6.3 ohms, SPEC was 6.4, and DATS measured 7.1. The final QTS 0.4 is far from SPEC 0.35
 
celef

celef

Member
Joined 2012
T

TMM

Member
Joined 2007
leon2216 said:
I also found this problem. I think it is because DATS cannot measure DCR correctly, which is obviously too large compared with the DCR measured by the bridge, resulting in the larger QES during calculation, thus improving THE QTS. I used NE315-8 to test, and the DCR measured by the bridge was 6.3 ohms, SPEC was 6.4, and DATS measured 7.1. The final QTS 0.4 is far from SPEC 0.35
Click to expand...
From memory (and I could be wrong on this), DATS does not directly measure DCR but instead just does a frequency sweep and fits a T/S model to the resulting impedance curve (20-20kHz). The resulting Re (and all other parameters) is the value which makes the T/S model fit the curve best. 7.1 vs 6.4 is not a huge difference in Re to be honest. It's not going to change the measured Q by a great deal.

If your driver has a very low Fs and high Le then Re is going to deviate from DCR the most since Re has little effect on the resulting impedance over the frequencies that the DATS sweeps.

The issue is not how DATS measures but the limitation of the T/S model itself. The T/S model is both highly simplified and is only valid for one drive level. If you are expecting DATS to spit out the same T/S parameters as the manufacturer when you measure at a different level, then you're misunderstanding the limitations of the T/S model.

Technically, if you listen at low levels then designing around the T/S model as measured by DATS will give you a better result than designing around the T/S model measured at screamingly loud SPL.
 
Last edited:
L

leon2216

Member
Joined 2020
hank you for your reply. I just used DATS to export THE ZMA file and simulated it on Speakerbench, and the results were the same with dats, proving that the DCR you mentioned was simulated and not involved in the calculation of Qes. I also have a pair of Wavercor speakers. The LMS measurement results were 0.47 for QTS and 0.56 for Dats, which was unbelievable, although Dr. Joseph D"Appolito in Testing Loudsperkers compared the difference between MLSSA and CLIO. However, the DCR of THE CLIO at that time was measured by the bridge. In addition, the VOLTAGE CLIO measured by MISSA was low, and the difference was not as large as it is now. The QTS of Dats was significantly larger (at least 6.5 "units or above) than that of LMS or SPEC, and the VAS did not decrease. This leads to large errors in design values
 
You must log in or register to reply here.
Top Bottom