Modeling a driver in an electronics CAD application

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hello

I am trying to, as easily as possible, to design electronic crossovers for speakers that already have passive ones.

My plan this far has been to take a driver's impedance graph, then model a circuit, no matter how complicated (I usually end up with total 8 to 10 resistors, inductors and capacitors) that has the same impedance related to frequency.

Then I can draw freq. response curves of the crossover networks with a) dummy resistance loads and b) the drivers as loads. Then it goes as easy as designing an electronic crossover with the same frequency and phase response as the original passive one had.

My first question is... is there any easier (a standard) way to model the drivers?

My second question is... do I get correct results this way? If not, where and why do I go wrong?
 
Thinkcat

Dedicated applications already exist to do that. Download SpeakerWorkshop from SpeakerWorkshop.com. Ideally, you should do acoustic and electrical measurements of the drivers you intend using to get the best results (more accurate and quicker to do). Also check here: The FRD Consortium.

The biggest hurdle I see is getting your drivers measured if you do not have a measurement microphone. Perhaps someone in your area can help with this.
 
I do not have the speakers made that I am planning this for. Then also I'm lacking in measuring equipment. I just know that I can build satisfactory speakers by doing everything as instructed.

Therefore my point is not exactly designing new speakers, but instead, modifying pre-existing instructions of a pre-existing kit. Is this the best way? I'm not sure, probably not. But is this good enough? My hope is to make this as much "good enough" as would be following the original instructions with a passive crossover.
 
The resonance is fairly easy to model, but the inductive rise of the speaker is much more difficult to model. It is best to use measurements. If you have impedance measurements we could talk further about this.

As Shaun O. Said, Speaker Workshop is a very inexpensive measurement system. An impedance jig might cost you very little or even be "free" depending on how well stocked your parts bin is. IF you already have a passive circuit design, the impedance will get you to a transfer function, which will get you to a start at modeling the passive filter with an active circuit.
 
Let's have an example.

Seas had a house-designed kit called Embla, that used their drivers, and all the drivers have measurements available in the datasheets. I intend to trust that the designer of the passive crossover has done his job well enough, so that I do not need to modify the intended frequency response anymore. And thus I will not need to do any measurements.

Jorma Salmi, the speaker design mastermind at Gradient, made a slightly improved version of the original crossover. Main differences being the need for one less capacitor and a slightly smaller series inductor for the bass-midrange (here smaller being a little less expensive). Surprise being the practically identical frequency response of these two crossovers.

http://koti.mbnet.fi/thinkcat/Embla/Pass_suotimet_ja_elem.png has, on the upper left part, the original crossover. Lower, currently connected, still on the left, is the Gradient's crossover. Going right from there the drivers should be self-explanatory. The text there is Finnish, so don't even try to get that.

The responses are at

http://koti.mbnet.fi/thinkcat/Embla/Gradient_vs_Seas.png

Pink and green being the Gradient's crossover, red and blue being the original Seas crossover.

The simulated impendances of the drivers (and actual impedances of the circuits in the previous diagram) are at

http://koti.mbnet.fi/thinkcat/Embla/Elem_imped.png

These follow as closely as possible the ones in the manufacturer's datasheets.

The active crossover is at

http://koti.mbnet.fi/thinkcat/Embla/Akt_suodin.png

And the frequency and phase responses are compared at

http://koti.mbnet.fi/thinkcat/Embla/Akt_pass_vertailu.png

I went silly to ensure that these follow each other as closely as possible. This might not be necessary. See

http://koti.mbnet.fi/thinkcat/Embla/Seas_vs_res.png

The green lines are the bass-midrange. The line with only a single bend is the original Seas crossover driving an 8 ohm resistor, and the line with more waves is the simulated driver instead of the resistor. Same holds about the tweeter for the red lines.

At -18dB level, the blue is the sum of the crossover responses with the resistors, and gray is the same with the simulated drivers. And black (at -30dB level) is the difference between driving the resistors and the simulated drivers.

The point in this last graph being, what would happen if I simply forgot about simulating the drivers and went straight ahead with a little simplification.

Assuming I did simulate and calculate the thing right. Did I?
 
I'd say, still get SpeakerWorkshop. All you'll need is to build a simple jig using resistors and perhaps some switches. You could then measure the transfer function of the crossover sections in-circuit (i.e., measured at the driver terminals) and save those as your target response curves for the electronic crossover.

Good luck.
 
Thinkcat said:
These follow as closely as possible the ones in the manufacturer's datasheets.

The point in this last graph being, what would happen if I simply forgot about simulating the drivers and went straight ahead with a little simplification.

Assuming I did simulate and calculate the thing right. Did I?

I think using traced impedances is a guess at best and wrong at worst. The woofer impedance looks a bit strange with the drop at the top and regardless you can see the +/- 2dBish variations between your resistive and real impedance simulations - assuming I am following your pictures. How would you think you can neglect that?

Obviously speaker impedance doesn't matter in the active circuit, but it definitely matters in the passive one.

I can't comment on the correctness of your sims, it would be asking way too much time to verify.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.