T/S parameters valid for normal listening levels.

[vinylnvalves]

Saw this thread over on DIYaudio https://www.diyaudio.com/community/thre ... bo.400699/
Suggests that speakers may not the as the simulations predict. It’s probably not a surprise, as all empirical methods “laws” are just an engineering best guess at a point in time.

[IslandPink]

Sounds intriguing - will try and dig into it when I get chance, this will be intermittent at present.

[vinylnvalves]

I always measure and voice the speakers in room, at normal listening levels, as I found measuring at low levels gives a slightly different FR sweep. I have always assumed it was the contribution of the room, that was the nonlinear variable.

[Daniel Quinn]

My suggestion would be : infinite baffle

[pre65]

My suggestion would be : infinite baffle

What led you to that conclusion ?

[Scottmoose]

Of course. Thiele / Small parameters (which are just a way of expressing the EM behaviour of a drive unit) are not strictly speaking fixed, but can vary with voltage drive.

In the original intent, small signal values were measured at relatively low levels [natch], and a lot of home & pro. measurement methods still do use that approach -not least because it's quick & cheap as you technically only need a couple of close-tolerance resistors & suitable leads. But moving coil drive units aren't always consistent, either mechanically or electrically (the latter depending on the overall design of the motor). Better types in theory should remain well-behaved until you either drop into the noise floor, or get to drive levels where the coil is moving out of the linear operating region, and some do, but that's not always a given.

Most published values are derived from one selected point on the drive curve. Manufacturers like Seas, Scan, SB, MA etc. publish data at specified, usually quite high drive levels since that ensures they are representative of typical usage conditions rather than leaving any room for doubt. So even though in theory it shouldn't make too much of a difference -for safety, you really want to be using representative drive levels to design an alignment, just to be sure, and make further modifications as necessary to account for the presence of any series R in the circuit, and / or the output impedance of the amplifier, since this also has the effect of altering the EM damping (i.e. it changes the effective Qes and thereby Qts). If you don't, the alignment can / will change, as shown in the link -extent depending on circumstance.

[rowuk]

When we are talking about Thiele/Small, we are basically ONLY talking about the lowest frequencies that a driver can produce. The "audibility" of a moving target at that lowest octave should not be a huge problem unless we are overdriving the voice coil at a frequency where the box is limiting motion of the cone. We must also remember that the room at those tuned frequencies also has huge effects on output reaching the ear.
For any reasonable playback levels indoors, I suspect that we have some wiggle room.

[Scottmoose]

Right, Thiele / Small parameters only define the behaviour of the driver in the acceleration (rising response) band up to the mass corner frequency in Hz ( [2*Fs]/Qts) while factors like Rms start to fall away even sooner, at roughly Fs/Qes.

Wiggle room there definitely is; I'd always recommend using whatever data yields the largest box size for a desired set of results so there's room for adjustment as desired. Unfortunately, the recent fashion for low-efficiency, high-excursion drivers with limited mechanical damping has helped leve the door open for some fairly significant variations though -I've measured a number of drivers from 'a certain company' for instance which were showing that at only fairly modest variations in drive level. That company produces many equivalent drivers in both 4ohm & 8ohm variations -interestingly, it was always te 8ohm types that showed the largest deviations, sometimes approaching 45%, which makes me a mite wary about drivers of this type in general -notwithstanding those from Scan, which are about as consistent as it gets.

[chris661]

I always measure and voice the speakers in room, at normal listening levels, as I found measuring at low levels gives a slightly different FR sweep. I have always assumed it was the contribution of the room, that was the nonlinear variable.

I found different: I used a directional and sensitive microphone close-up to a test speaker (Kef egg), and found the frequency response stayed consistent at levels from middling to inaudible.

The penultimate sweep, I could only hear blips as it went through the kHz range where the ear is most sensitive. 6dB below that, nothing.


Some noise did get in, which showed up in the THD curves.

I'll dig around and see if I can find the REW file.


Chris

[Scottmoose]

I've never run across any FR variations -not other than caused by some form of mechanical defect (rubbing coil, tinsel leads etc.), or in a handful of cases excess excursion physically distorting the cone. Those aside that at least should remain consistent; T/S values can be a different matter depending on driver; even then I suspect at least a proportion of the differences sometimes found are down to when the drivers are measured -which I'd lay odds is in 99% of cases is when they are new (or newly acquired) and partly become conflated with the suspension breaking in.

[Nick]

I would have thought that given the driver introduces distortion, that is a result of it being somewhere non linear. Being non linear would make me think there was some level dependent value somewhere in the system. So given that I would be surprised that there would not be some level dependent behavior. As to if those changes are reflected in T/S parameters I don't know.

[brig001]

Looking at Klippel tests, excursion seems to be the main thing that makes parameters vary - if excursion is low, the driver is more linear. I dare say some effects cancel and others sum though

These are for my drivers:
https://www.daytonaudio.com/images/reso ... esults.pdf

Look at the variation in FS for example

Loads of info at Klippel, but It's all beyond me to be honest...
Brian

Diet: Can't spell

[Scottmoose]

I would have thought that given the driver introduces distortion, that is a result of it being somewhere non linear. Being non linear would make me think there was some level dependent value somewhere in the system. So given that I would be surprised that there would not be some level dependent behavior. As to if those changes are reflected in T/S parameters I don't know.

It depends how you're defining distortion. Frequency response, i.e. any deviation from a ruler-flat line from 0Hz to infinitely high is defined as linear distortion. Assuming you're working within the mechanical limits of the components, it's basically fixed and doesn't in itself show deviations.* Non-linear distortions (harmonic, IM etc.) are largely born in the motor & related components, and do show variation with drive level etc.

T/S parameters (generic handle) are slightly different, although in practice they're more closely related in behaviour to the latter. What we call T/S parameters are just the electromechanical properties of the drive unit at & around resonance; fundamental, small & large signal -real or mathematical analogies from electrical filter theory. Because the motor is ultimately non-linear at higher excursions you get deviations in those just as you do HD etc.

*External factors may come into play e.g. FMD but they're not innate to the electromechanical behaviour of the driver itself.

[Nick]

distortion, that is a result of it being somewhere non linear.

I would have thought that would be a clear indication how I an defining distortion

[brig001]

Frequency response, i.e. any deviation from a ruler-flat line from 0Hz to infinitely high is defined as linear distortion. Assuming you're working within the mechanical limits of the components, it's basically fixed and doesn't in itself show deviations.

I don't think that this is true
When you look at the Klippel report I posted, FS changes significantly with excursion from 63.4Hz to 43.0Hz
Unless its effect is cancelled out by similar changes in Qts, Rms, Cms, Vas etc.
It won't affect my speaker too much since I have a small sealed box, but it could show up in others

Brian