Which Thiel-Small equivalent circuit to use?

[andy2]

According to the paper, there are two types of equivalent circuits to analyze a speaker system: one is called acoustical analogous circuit and the other is the electrical equivalent circuit. One can be converted to the other and vice versa.

If you need to see how the speaker system responds, which one would you use?

For example, if you need to understand how your filter affects the frequency response, which circuit to use?

[soongsc]

Originally posted by andy2
According to the paper, there are two types of equivalent circuits to analyze a speaker system: one is called acoustical analogous circuit and the other is the electrical equivalent circuit. One can be converted to the other and vice versa.

If you need to see how the speaker system responds, which one would you use?

For example, if you need to understand how your filter affects the frequency response, which circuit to use?

I've used the electrical equivalent before, you can use it to determine whether the right amount of power is going though the driver over a frequency range.

What does the "acoustical analogous circuit" look like is there a web site that expains it?

[andy2]

Originally posted by soongsc


I've used the electrical equivalent before, you can use it to determine whether the right amount of power is going though the driver over a frequency range.

What does the "acoustical analogous circuit" look like is there a web site that expains it?

Go to this website:
http://www.readresearch.co.uk/HTML/papers.htm

You can download the original paper and you should see it talks about these equivalent circuits.

[f4ier]

The all-electrical model should be the one to use to see the impedance curve. It's also the one to use to load your filter and extract the electrical filter transfer function. You then convolve this filter transfer function with the raw acouctical transfer function to get the actual filtered response.

If you are using a circuit simulator, it's easiest to use the full electro-mechano-acoustical model.

Remember that these models are only valid within your speaker's linear range. These models are no where near complex enough to characterize cone breakup for one thing. They're really only useful for relatively low frequency simulation.

cheers

[soongsc]

Originally posted by f4ier
The all-electrical model should be the one to use to see the impedance curve. It's also the one to use to load your filter and extract the electrical filter transfer function. You then convolve this filter transfer function with the raw acouctical transfer function to get the actual filtered response.

If you are using a circuit simulator, it's easiest to use the full electro-mechano-acoustical model.

Remember that these models are only valid within your speaker's linear range. These models are no where near complex enough to characterize cone breakup for one thing. They're really only useful for relatively low frequency simulation.

cheers

I have found the the parallel RCL model in series with the RC sometimes does not fully agree with the impedance curve tested.

[f4ier]

parallel RCL model in series with the RC

Did you mean in series with a series RL circuit?


They don't agree 'cause real speakers have lossy voice-coils. You'd need a bit more components to model lossy coils. A parallel RL in series with the above circuit slightly improves the model. However, it'd still be an academic exercise as the improved electrical model is useless without the actual raw acoustical transfer function of the speaker.

[soongsc]

Originally posted by f4ier


Did you mean in series with a series RL circuit?


They don't agree 'cause real speakers have lossy voice-coils. You'd need a bit more components to model lossy coils. A parallel RL in series with the above circuit slightly improves the model. However, it'd still be an academic exercise as the improved electrical model is useless without the actual raw acoustical transfer function of the speaker.

Yes, it should be RL. With the additional parallel RL, would it interact with the filter in a similar manner as the actual driver?

Since now the electical model is different from the TS model, it seems questionable how well the acousical TS model is.

[f4ier]

The model interacts with the filter as per circuit theory. But yes -- there is more similarity with the real thing; the model is an improved version after all.

The simulated, raw, acoustical transfer function is not completely accurate anyway (see my first reply).

[andy2]

Originally posted by f4ier
The model interacts with the filter as per circuit theory. But yes -- there is more similarity with the real thing; the model is an improved version after all.

The simulated, raw, acoustical transfer function is not completely accurate anyway (see my first reply).

What models then are being used by the commercially available software such as lpscad ...?

[f4ier]

LEAP, SoundEasy, LspCAD, Speaker Workshop, Crossover Simulator, Jeff's Excel spreadsheets (at the FRD Group) all use models of filters with actual impedance and acoustical measurements of speakers. The pair of measurement sets are stored in separate text files with frequency - magnitude - phase format per line as one data point. The files could contain any number of lines; the more data points the higher the resolution of course.

It's not as simple as sticking a microphone in front of a raw speaker while stepping through some test tones and then recording the SPL. You've got to eliminate your room's acoustics from your measurements. There are several measurement methods each with their +/- points. LMS, for example, uses gated sine sweeps to measure SPL. SoundEasy, LspCAD, Speaker Workshop use the Maximum Length Sequence method to measure SPL.

Have a look at the manuals of the above-mentioned software for more information