Subwoofer and linearity

[JustAFrog]

Hello people,


I try to measure a subwoofer of mine in order to compute a linkwitz transform, and I am suprised to see that it is highly non-linear.


at 1Vrms, fs = 51Hz
at 5Vrms, fs = 41Hz
at 7Vrms, fs = 38Hz (!)
I didn't go any further.



39Hz is the value given by the manufacturer.
I measured that by holding the thing in my hand (but if it rests on the floor it only change fs by one or two hertz)



Is it normal such un-linearities at a rather small voltage?

Could it behave better properly enclosed ?It's a rather cheap car subwoofer, should I try to get a better one?



It is that one :

https://www.focal.com/sites/www.focal.fr/files/shared/catalog/document/fiche-technique-27v1.pdf


Thank you very much for your time, I hope you can enlighten me

 

[Michael Bean]

If you were testing the raw driver alone, not installed in a box, you were probably driving it past it's linear operating range. Use small signals for the driver alone and large signals for in-box testing.

Mike

 

[JustAFrog]

If you were testing the raw driver alone, not installed in a box, you were probably driving it past it's linear operating range. Use small signals for the driver alone and large signals for in-box testing.
Mike

Thank you for your answer
Do you think it's ok to have fs=51Hz when 39 is expected ?

 

[ICG]

Thank you for your answer
Do you think it's ok to have fs=51Hz when 39 is expected ?

It was a new driver, wasn't it? It's normal for drivers to have a stiffer suspension comming fresh out of production. If you wobble them in for a while with a big excursion (or play music for a week), the fs goes down. So far nothing unusual. Measure the parameters before and after and you'll find they are consistently and do not lead to a different enclosure.

 

[bentoronto]

Hope you will post your findings as you produce them.

I don't know how the novel function you are reporting (resonance in relation to drive) relates to the familiar parameters (such as harmonic distortion). Or only secondarily.

Mountains of conflicting opinions about break-in for drivers.

B.

 

[academia50]

Mountains of conflicting opinions about break-in for drivers.



Yes

Matrix-Hifi

The corner/speaker burn

 

[AdamThorne]

"I measured that by holding the thing in my hand (but if it rests on the floor it only change fs by one or two hertz)"

You hand is much too mushy. On the floor is the way to go. But if there is a rear vent on the driver you have to leave that unobstructed. Some folks will build a jig to hold a big driver if it has a rear pole vent.

 

[JustAFrog]

It was a new driver, wasn't it? It's normal for drivers to have a stiffer suspension comming fresh out of production.

I got it second hand, I have no idea of it's background. One thing sure is that it didn't got used for a long time. I made a quick test yesterday, stressing it a bit and measure it again, it didn't changed the result.

Hope you will post your findings as you produce them.

I don't know how the novel function you are reporting (resonance in relation to drive) relates to the familiar parameters (such as harmonic distortion). Or only secondarily.

I'll do what's to be done to hook up the speaker to the computer to have some nice data, and mount the speaker in a cabinet so a can push the volume more. I'll try to get a nice 3D curve "impedance/frequency/driving tension"


Shall I post the results here or start a new thread ?

 

[TBTL]

I would be interested in that plot. You can post it here.

 

[ICG]

I got it second hand, I have no idea of it's background. One thing sure is that it didn't got used for a long time. I made a quick test yesterday, stressing it a bit and measure it again, it didn't changed the result.

A lot of drivers have to be broken in again after longer periods of time not used. It's mostly (but not exclusively) the surround like the tough-elastic treatment of a lot of PA drivers but can also apply to the spider or parts of the cone (fullrange drivers).

Just to verify what you wrote: The fs changes with the power/excursion even after playing it reasonably demanding for a while?

The Thiele-Small-Parameters are mostly small signal parameters. They can change a lot with high power, excursion and mounting.

On high power the heat of the VC increases the Qts because of the higher resistance (power compression), the excursion can change the inductance when the coil leaves the homogenous magnet field, a high excursion can get the suspension into progression or DC offset (the latter often in conjunction with power compression) or simply a 'default' offset off the geometrical resting position. If the driver isn't vented behind the spider and/or dustcap, the air compression also changes the suspension, which, in turn of course change the parameters increasingly.

The mounting on a baffle, in a horn, sealed, vented etc changes a lot too. Free air measurements of the driver can falsify the measurements because of the cancellation back/front can lead to an excursion which is non-linear anymore because of the progressive suspension. Also, the pole core bore and other ventilation measurements may not be blocked because of the trapped, compressed air adding up the suspension and the driver has to stay upright because of the possible offset because of the mass of the VC/cone and gravity.

 

[bentoronto]

+1

Yes, for sure, all kinds of parameters in flux or, at least, not properly addressed by static T/S (electric analogy) models. For example, the recently explored variable inductance of voice coils.

Beats me why a 100 year old driver technology remains the standard today. Or why elaborate sims are calculated based on parameters that flux - and screw up passive crossovers.

Many of these issues are overcome by multi-amping and active XO (esp DSP with 48dB/8ave slopes).

Ben

 

[stoneeh]

Any supposed impact of break-in can be easily put into numbers - just measure TSP of the driver fresh out of the box, measure them again after break-in (including a proper rest period), and compare.

Here for example is a measurement series (German; use translate) that can only find temporary effects of break-in over multiple speakers; after a rest period, parameters return to standard: Chassis einbrennen - Was bringt es wirklich? - Der-Akustische-Untergrund
Note: their measurements show yet a difference between 1 and 10 days of rest.

Other measurement series either show no difference, or show a difference up until maybe a couple of hours of rest period. None that I've seen have given the DUT multiple days of rest, after which it is likely all changes have reverted to their out-of-the-box value.

 

[bentoronto]

... other words, the "TSP" will be changing every day you haven't energized your drivers, or energized them enough, or the weather changes, or.... and that holds only for some drivers but not for others.

And if you own the wrong kind of drivers, it seems inescapable that designs need to consider that flux and be insensitive to it.

Is it fair to say flux doesn't apply directly to drivers other than Rice-Kellogg cones? Or just nobody has checked?

B
Footnote: electrostatic bias voltage can change with the weather. But the impact is small and not freq dependant, other than the whole band the ESL plays.

B.

 

[academia50]

A lot of drivers have to be broken in again after longer periods of time not used. It's mostly (but not exclusively) the surround like the tough-elastic treatment of a lot of PA drivers but can also apply to the spider or parts of the cone (fullrange drivers).

I see it as a..... myth ?

An argument without much technical support. Time can work in favor of finding the T / S

Also, the pole core bore and other ventilation measurements may not be blocked because of the trapped, compressed air adding up the suspension and the driver has to stay upright because of the possible offset because of the mass of the VC/cone and gravity.

This is decidedly a myth, when it comes to speakers of good build quality.
My JBL coaxials have not been changed from their original position in 40 years and the VC works perfectly centered. The spider takes care of that, and I've already changed the foam environments 2 times in all this time ...

It seems to me that if some people read the link that I provided above, we could discuss some more interesting aspects, we were talking about the "softening" of speakers, I think ......

 

[head_unit]

Shall I post the results here or start a new thread ?

I'd be curious to see that here. Like the full impedance curves at each level.

I am also curious: what are the details of your entire measurement setup? And yes as noted, drivers should be clamped (NOT hung!) and if near a floor or wall there is a bit of effect.

Thiele-Small parameters are SMALL SIGNAL. If the woofer is moving much, then it's a different thing.

51 to 39 seems a lot. But Focal is a huge manufacturer, they are not fools, so something is odd.

 

[JustAFrog]

Hello guys, I didn't forgot you
I was rather busy doing some woodworking, fixing my amplifier channel, soldering some few thing to be able to use my sound card to measure automatically, now I have to do a bit of coding to generate and analyse the data

I did the measurement of Fs with the xy mode of my oscilloscope, finding the frequency where the load is resistive, I measure current with a 10Ω in series with the speaker. I plan to use the two channel of my sound card to measure the tension at the resistor and the tension of the whole (speaker + resistor).

I mounted the speaker in cabinet, and now the frequency seems way more stable, accrediting the thesis than I fell off the linear zone, the speaker being in free air rather than in a small closed cabinet (27liters).

 

[GM]

Matrix-Hifi

Hmm, dead link for the pdf, though thankfully archived: Wayback Machine

 

[head_unit]

I measure current with a 10Ω in series with the speaker

There are two classical approaches: big resistor, or really small like 0.1 ohm, which give somewhat different results. You're in between-nothing wrong with that, but you may get somewhat different numbers. Now as far as Fs, the oscilloscope should be fine. What voltages are you seeing across the WOOFER? Fs will depend some on drive voltage though I would not expect as drastic as 51 to 39 or whatever. I wonder if some nonlinearity is tripping out the oscilloscope method. Try repeating with like 100 or 1000 ohms maybe?

 

[JustAFrog]

Hello folks. I have been torturing a beautiful and old Toshiba SB-230, and frying some few low value resistor. I managed to get some few usable data
Here the result for the speaker: The thing is non linear at all.

The speaker is mounted in a closed 27L cabinet. Nothing to fear about weird extreme extrusion and all.

As you can see, with 1000mV, fs is about 60Hz.
With 8000mV, fs is 55Hz (theoretical value is 51.6Hz), I am pretty sure if I increase the voltage it'll lower again, maybe to tend to 51.6Hz (?).

So… in the graphic you can see the behaviour of the thing. vertical axis : impedance
horizontal from left to right : frequency
front to back : voltage (in mV)


Maybe for this particular speaker 0.1Watt isn't "small signal", I guess 16 watt isn't either. (it needs 30W ? 100W ? or it is just non linear no matter what ?) I wont torture this amp anymore to find out

I am curious if you can replicate a similar experiment.
I am not equipped to analyse that in the acoustic domain. I might do it soon.

(on the data you can see how my poor amplifier was struggling driving a 4ohm load)

 

[JustAFrog]

What voltages are you seeing across the WOOFER? Fs will depend some on drive voltage though I would not expect as drastic as 51 to 39 or whatever. I wonder if some nonlinearity is tripping out the oscilloscope method. Try repeating with like 100 or 1000 ohms maybe?

It really depends, I looked from a range to few mV to some few volts. With bigger value resistor I wont be able to have some few volts on the speaker (I dont have an amplifier powerful enough).