WHY LARGE SIGNAL PARAMETERS, NOT SMALL

[Quentin Jarrell 671]

I have modeled several enclosures below the show the difference between large signal and small signal T/S parameters.  To gather large signal parameters we use what klippel calls LSI (Large Signal Identification).  The following is Klippel's definition of LSI:

 

LSI modules measures the linear, nonlinear and thermal parameters of transducers operated in free air or in a sealed or vented enclosure. The transducer is operated  under normal working conditions (vertical position and fixed clamped in a laser stand or enclosure). Starting in the small-signal domain the amplitude is gradually increased up to limits admissible for the particular transducer. The maximal amplitude is determined automatically using the identified transducer parameters and general protection parameters describing the thermal and mechanical load.  The identification of the model parameters is performed in real time with an adaptive system. It is based on the estimation of the back EMF from the voltage U(t) and current signal I(t)measured at the electrical terminals. The identified model allows locating the sources of the nonlinear distortion and their contribution to the radiated sound. The dynamic generation of a DC-part in the displacement, amplitude compression and other nonlinear effects can be investigated in detail. After the initial identification a temporal parameter variation and long-term thermal effects can be investigated. The data is stored in the stand-alone processor unit and can be transformed via the USB interface to a connected computer for visualization and interpretation.

 

 

Basically this means the analyzer collects T/S parameters with the transducer under normal operating power in either free air or in an enclosure.  We model in the typical free air method at one meter.  The transducer below is the DSS Ethos dual two ohm 15.

 

SMALL SIGNAL PARAMETERS (cone moved 1.7mm) FOLLOWED BY LSI

 

Re          3.99          3.99

Qms       7.479        6.69

Qes        .712          .57

Qts        .65             .53

Fs          41.9          28.6

Mms       362.9g      362.9g

Cms      .04mm/N    .09mm/N

BL         23.149       23.149

Vas       36.379L     78.0361L

SD         804.25       804.25 cm^2

SPL       87.7 1w/1m        87.7 1w/1m    (NOTE: spl efficiency increases as power is applied however we still rate it at 1 watt @ 1meter as most modeling programs use this factor, it will not affect                                                                          response)

 

GREEN is LSI 

RED is small signal

 

 

First up a recommended ported enclosure with small enough volume to fit in a vehicle with tuning to give a little bump in response.  3 cft tuned to 32 hertz

 

 

 

 

Next is a bit larger enclosure to give bump in response in the low 30's with a mid 20's F3.  5 cft tuned to 28 hertz

 

 

 

 

Next up is an enclosure with a maximumly flat response with lowest possible F3  6.7 cft tuned to 22 hertz

 

 

 

 

And finally a sealed enclosure with a qtc of .7  You might say they look too terribly different but look at the difference in enclosure size.

 

 

 

 

 

 

 

 

[ramteid 17]

So what about compared to the real life counterpart? Do the subs react similar as the large signal parameters show in winisd?

[shizzzon 461]

So what about compared to the real life counterpart? Do the subs react similar as the large signal parameters show in winisd?

 

Minus environmental acoustical gain, yes

[ramteid 17]

So what about compared to the real life counterpart? Do the subs react similar as the large signal parameters show in winisd?

Minus environmental acoustical gain, yes

I thought i had read that in the loud speaker cookbook but was not for sure. So why do subwofer companies not do large signal parameters instead?

[Impious 1,476]

So what about compared to the real life counterpart? Do the subs react similar as the large signal parameters show in winisd?

Keep in mind it's all realistically an approximation of real world performance. It is very difficult to accurately model real world large signal performance due to the number of variables. The above quote doesn't specify exactly how the large signal parameters are defined (I.e. what power, heat, or linear limits are at the point of parameter definition). The large signal analysis above could be a "snap shot" of the driver's parameters at a particular point along it's performance or power curve. If more or less power is applied then the parameter shift from small signal will be different. Based on the same Re on both measurements it doesn't appear significant coil heating is taken into account.....as the coils heats up due to the power being applied Re will increase which will increase Qes....so there's another source of potential difference from the large signal parameters above.

This is why small signal parameters are used.....they are a linear measurement, and modeling works best when things are linear. When things become non-linear it becomes very hairy very quickly. That said Q and Klippel are correct, small signal isn't 100% "real world" and large signal analysis is a closer approximation, albeit with it's own set of problems.

[Quentin Jarrell 671]

So what about compared to the real life counterpart? Do the subs react similar as the large signal parameters show in winisd?

Keep in mind it's all realistically an approximation of real world performance. It is very difficult to accurately model real world large signal performance due to the number of variables. The above quote doesn't specify exactly how the large signal parameters are defined (I.e. what power, heat, or linear limits are at the point of parameter definition). The large signal analysis above could be a "snap shot" of the driver's parameters at a particular point along it's performance or power curve. If more or less power is applied then the parameter shift from small signal will be different. Based on the same Re on both measurements it doesn't appear significant coil heating is taken into account.....as the coils heats up due to the power being applied Re will increase which will increase Qes....so there's another source of potential difference from the large signal parameters above.

This is why small signal parameters are used.....they are a linear measurement, and modeling works best when things are linear. When things become non-linear it becomes very hairy very quickly. That said Q and Klippel are correct, small signal isn't 100% "real world" and large signal analysis is a closer approximation, albeit with it's own set of problems.

 

 

As he said this is all an approximation.  Parameters change greatly depending on a lot of factors.  As the cone moves most parameters change constantly.  The LSI parameters are a snap shot of the driver at xmax.  Since we most expect to play to xmax this is what we use.  As pointed out Qes will increase with heat.  I wasn't going to go into it but since it was brought up I'll explain why Re was left the same.  The point at which the large signals were generated the coil had increased by 46 kelvin (~46 celcius increase from room temperature at start of testing).  Coil Re raised to 4.69 ohms.  However it took ~350 seconds for the coil to reach this temp.  If we were to list this Re customers would first assume their subwoofer is bad if measuring Re while it is cold with a DMM.  Also it could cause the customer to load the amp with an impedance too low thinking the sub's impedance was higher than actually is when beginning to play.

 

It's all about getting a close round about approximation.  Even the box the driver is in will change the driver's parameters.  Large is just a better approximation over small as we don't play our subwoofers to only 2mm.  What is important is both tests are done in the same exact manner.  The only difference is with large signal the driver is moving as it would be powered up.  The reason why Thiele and Small did not do large signal testing back when was they were dealing with much smaller drivers.  A driver that is only rated for 5 or ten watts will not see as dramastic parameter shifts at 1 watt versus 5.  Not enough to make a difference when it comes to the enclosure.

[ramteid 17]

Alright, thanks for the info.

[///M5 2,833]

 

So what about compared to the real life counterpart? Do the subs react similar as the large signal parameters show in winisd?

Minus environmental acoustical gain, yes

I thought i had read that in the loud speaker cookbook but was not for sure. So why do subwofer companies not do large signal parameters instead?

The good ones do small, large and A LOT more.  Then others for instance don't even know what t or s are.  Kevin for instance.

 

Just refreshing IMO to see someone who is offering a driver bring up discussion points that are not only relevant, but show that hiding information is not the protocol of the business.  Definitely gives me a lot more confidence in doing business with Quentin.

[ramteid 17]

So what about compared to the real life counterpart? Do the subs react similar as the large signal parameters show in winisd?

Minus environmental acoustical gain, yes

I thought i had read that in the loud speaker cookbook but was not for sure. So why do subwofer companies not do large signal parameters instead?

The good ones do small, large and A LOT more. Then others for instance don't even know what t or s are. Kevin for instance.

Just refreshing IMO to see someone who is offering a driver bring up discussion points that are not only relevant, but show that hiding information is not the protocol of the business. Definitely gives me a lot more confidence in doing business with Quentin.

Yes, it is nice to see and makes it more likely that ill buy some of his products in the future. Edited September 22, 2013 by ramteid