SB Acoustics TS parameters dreadfully inaccurate

My 2 cents, build the bloody box according to the damned manufacturer spec sheet and tweak if necessary. Theoretical discussions on TS parameters and their measurement are a dreadfull waste of time!

I hope you're right - I measured 4 JBL drivers with DATS2 and they were consistently off compared to JBLs figures. Using JBLs numbers gives a much smaller box. :D

I bought the DATs2 as I often end up with drivers out of old speakers and thought it would be useful to be able to measure them for new projects. So far none have been that close to manufacturers specs (when available) :(
 

Attachments

  • ts.jpg
    ts.jpg
    20.9 KB · Views: 341
Hi,

Here some information about how I measure TSP, maybe it can be interesting for some...

I use a Clio V7.02 system and choose the current sense mode to measure the impedance. In this way voltage is applied to the voice coil and the current is measured via an internal serial resistor of about 0.1Ohm.

I use 2.83Vrms on the voice coil. I have chosen this value because it is relevant for the mean operating point in the application. For lower and higher values the TSP become slightly different for good drivers. When the TSP are strongly dependent on the voltage level, the cause is a more non linear suspension in most cases.

To compare with the supplier published TSP, you have to know how they have measured, that is evident of course.

The sweep rate used for the impedance measurement must be very low.
In case the sweep rate is too fast the impedance peak at fs is not sampled correctly. The impedance peak becomes too low and the peak also shifts in one direction dependent of the sweep direction.

I have added an old Audax article where the loudspeaker parameter measurement is described, doing it in the old fashioned way by hand. In case you have some doubts about your equipment, it can be interesting.

Paul

View attachment Measurement of loudspeaker parameters - Audax article.pdf
 
I hope you're right - I measured 4 JBL drivers with DATS2 and they were consistently off compared to JBLs figures. Using JBLs numbers gives a much smaller box. :D

Regarding these JBL numbers, i am afraid i am not a discerning enough audiophile to apreciate important audible differences between a box designed based on manufacturer data or on home made measurements...:rolleyes:

Rooms and loudspeaker placement will certainly produce more important differences than slight misalignment of the load, a vastly overated issue, imho.
 
Last edited:
manufacturers do not change moving coils and cone materials allowing mass increases of almost 100%!
Yes, they do.
Consistent manufacturing of paper cones is tricky.

Btw, the "appropriate" signal level is an interesting concept, can you elaborate on this?
Appropriate signal = expected, working signal.
Powerful signal for sound reinforcement drivers and weak (1W or so) for hi-fi drivers.

My 2 cents, build the bloody box according to the damned manufacturer spec sheet and tweak if necessary. Theoretical discussions on TS parameters and their measurement are a dreadfull waste of time!
In stone ages, this would be OK advice.
We are living in modern times now, with science learned (or not).
 
Last edited:
Appropriate signal = expected, working signal.
Powerful signal for sound reinforcement drivers and weak (1W or so) for hi-fi drivers.

OMG! :eek: Looks like you have never heard of what small signal is all about, nor have ever read ( even superficially) those Thiele and Small papers...:)

In stone ages, this would be OK advice.
We are living in modern times now, with science learned (or not).

Big science indeed...:p
 
Regarding these JBL numbers, i am afraid i am not a discerning enough audiophile to apreciate important audible differences between a box designed based on manufacturer data or on home made measurements...:rolleyes:

Rooms and loudspeaker placement will certainly produce more important differences than slight misalignment of the load, a vastly overated issue, imho.

Fair enough - If my measurements are correct I'd either need each box to be an extra 160L per speaker, or use the smaller 'jbl' spec box and nearly twice the amplifier power with EQ. You're correct that I probably can't discern the subtle differences but I'm adding a high pass to the speakers roll off and would like it to happen at the correct frequency to work with my processors fixed 80Hz value.

As you say, it's probably better to just chuck them in a box, stick up some wall treatments and spend a few days moving them about to fix any response anomalies :p
 
As you say, it's probably better to just chuck them in a box, stick up some wall treatments and spend a few days moving them about to fix any response anomalies :p

Well i think that trusting in TS data and sims is no excuse to check and measure how the box actually behaves...

I am under the impression that most diyers are extremely suspicious with manufacturers data and picky with their sims, but seldom show any interest in double checking their own builds...:rolleyes:
 
1. Don't trust manufacturer TS data, measure it.
2. Simulate the box with the measured TS data.
3. Build the box according to the simulation.
4. Measure the box.
5. Tweak if necessary.

If you slap the drivers in to a box built according to manufacturer TS data, than you are risking an unnecessary step: destroying the wrongly built (beyond any tweaking) box and building the right one.
 
It is not easy to measure meaningful TS parameters of large drivers. Everything matters, drive levels, temperature, orientation, etc. I have used 3 different tools, AP, Klippel, and DATS, they all give different answers. Also, the alignment is going to change when the voice coil heats up in use. Not to mention power compression. Best you can do is build it, measure it, listen to it and tweak it.
 
1. Don't trust manufacturer TS data, measure it.

If your data are different, why should you trust your measurements skills better than the manufacturer's?:rolleyes:

In my case, it happens quite often and i am never quite sure what's right or wrong...

It never happened to you that different measurements made by yourself in different conditions give you different results? :rolleyes:

It also happened to me almost all the time. Trouble is I have no standard procedure to base on, and i suspect this is the reason... Serious manufacturers do have their procedures, check the SB papers, because they have no problem to be totally transparent on this point.

Did you ever have to destroy anything built based on manufacturers data? :rolleyes:

I never had to, only apply tweaks based on box stuffing, port tuning, etc... Nothing dramatic...:)
 
It is not easy to measure meaningful TS parameters of large drivers.
Yes, that is true.
Also, the alignment is going to change when the voice coil heats up in use.
Exactly! Manufacturer TS data does not include those effects. That is why preconditioning and appropriate signal level are important.
 
Last edited:
If your data are different, why should you trust your measurements skills better than the manufacturer's?:rolleyes:
Because I am that good.

It never happened to you that different measurements made by yourself in different conditions give you different results? :rolleyes:
I am setting the conditions so I know what is going on.

Did you ever have to destroy anything built based on manufacturers data? :rolleyes:
Yes.

I never had to, only apply tweaks based on box stuffing, port tuning, etc... Nothing dramatic...:)
Lucky You...
 
Drivers can exhibit quite a variation in their parameters but usually Fs/Qts and VasFs² are usually close so the final outcome is fine (ref LDC 6th Ed).

I'll put money on if a few different builders measured the same drivers, the results would be quite different.

For the last decade I've only used the manufacturer's data and all the speakers have met their targets with no surprises (except for a SEAS Nextel midwoofer). Saves a lot of time and hassle measuring with results that can be iffy.
 
It is true that Fs/Qts and VasFs² are usually close within normal production variation, but I have seen strikingly larger Qts than specified - mostly by cheap far-East drivers, but also by some Scanspeak Revelator and Classic midwoofers (Discovery series is OK). In this circumstances it leads to a wrong decision about the type of the loudspeaker box (TL vs vented vs closed vs ...).
So, measure the bloody TS parameters, it is easy and fast.
 
It is true that Fs/Qts and VasFs² are usually close within normal production variation, but I have seen strikingly larger Qts than specified - mostly by cheap far-East drivers, but also by some Scanspeak Revelator and Classic midwoofers (Discovery series is OK). In this circumstances it leads to a wrong decision about the type of the loudspeaker box (TL vs vented vs closed vs ...).
So, measure the bloody TS parameters, it is easy and fast.

Yes indeed, and these SB drivers are a good example of that. The fs is slightly high, but the Qts is much larger than it should be. As you say, using the manufacturers figures would give completely the wrong cabinet topology.

Just checked out the link from Tweet earlier in this thread, where they compare the Dayton WT3 (equiv to DATS) and the LinearX LMS system. They give extremely close agreement.

Just heard from Intertechnik, and they are going to replace the drivers. Hopefully, they will be better (and not even worse!)