X-over design questions

[Milzie]

I'm trying to learn a little about how to design/build a crossover (newbie) and have a few questions that I cant seem to find an answer to.

1) Why are different gauge inductors available and what effect will be seen by changing the gauge? In other words if a specific gauge inductor is specified on a crossover schematic, what will change if a larger or smaller gauge inductor is used of the same inductance value? Or is it just typically use a larger on the woofer and smaller on the tweeter?

2) If an impedance compensation circuit is used on a driver, does this effect the values of the other crossover components. Example, If a driver (fs =50Hz) had a 12dB LW crossover at 2700Hz already built, then an impedance compensation was added after the original design, would this effect the values for the cap and coil for the original 12dB LW x-over (need to be changed?) If they did need to be changed, would a new impedance graph need to be generated in order to calculate the new values (see question 4)?

3) Same as question 2 but for a tweeter attenuation circuit?

4) If one were to design a crossover simply based on the driver manufacturers impedance and sensitivity graphs using the little calculators that can be found all over the net, do you use the impedance at the crossover frequency taken from the graph, or do you simply use 8ohms (for an 8ohm driver)

I guess what I am trying to figure out, is if I can build a crossover based on the little calculators you can find all over the net, and get somewhat satisfactory results without too much tweaking and measuring.

Also a question on sensitivity graphs.
Some manufacturers graphs appear very smooth, while others show some roughness to the curves (compare a Morel MW-164 to a Vifa P17WJ-00-08 for example). Is this due to accuracy of the measurement equipment (1/3 octave vs. 1/6 octave for example), or is it because some manufacturers smooth the curve by putting in a line of best fit, or is it simply because one driver is significantly better than the other? How audible would this rough section be if for example comparing the 2 above drivers?

Thanks,
Jeff

[SY]

Your bottom line question: it's highly doubtful that you'll get a satisfactory crossover using a simple calculator. If you have any hope of getting close, it's by the use of measuring equipment and speaker CAD programs. If you're not ready to do that, use the resources of someone like Madisound, who will (for a fee) design you a crossover for the drivers you buy from them.

Calculators do work well for calculating bass response in cabinets- assuming, of course, that the drivers you buy actually have the same Thiel-Small parameters as the spec sheet where you get the info to input into the calculator.

As for the more detailed questions, ALL crossover components and the driver's actual impedance interact. There is NO one answer to "what gauge coil should I use?" or "what happens if I put a Zobel network across a woofer?"

[fdegrove]

Hello Milzie,

2) If an impedance compensation circuit is used on a driver, does this effect the values of the other crossover components. Example, If a driver (fs =50Hz) had a 12dB LW crossover at 2700Hz already built, then an impedance compensation was added after the original design, would this effect the values for the cap and coil for the original 12dB LW x-over (need to be changed?) If they did need to be changed, would a new impedance graph need to be generated in order to calculate the new values (see question 4)?

To give a short answer to this isn't simple but let me say just this:

You could put an impedance correction network at the input of the x-over (even at the output of an amp taken the cables with it) and that would then linearize (or approach that ideal) the impedance of all LS combined.
A more common way to do this is to correct each individual speaker (mostly woofers and midranges) with an optimimized network.

It may influence the other components down the line but that will very much depend of the type of filter in use.

Cheers,

[Milzie]

Thanks for the replies

How about the last little bit regarding the smmoth vs. rough sensitivity graphs?

Jeff

[fdegrove]

Hi Milzie,

Also a question on sensitivity graphs.

Graphs like that aren't all that meaningful.
As long as you don't see gross abberayions you should be fine.

Cheers,

[SY]

Milzie: Any resemblance between a spec sheet curve and what your driver will do in your box with your baffle is sheer coincidence. The manufacturers' specs will also give you no idea of what the acoustic center offset will be for your drivers and your baffle geometry.

A few hours spent studying Dickason's "Loudspeaker Design Cookbook" will do you a lot of good. It's a pretty straightforward treatment.

[planet10]

The following are generalizations:

1) The guage of the wire usually effects the resistance of the coil (every coil is L+R, not textbook L)

2,3) An impedance compensation circuit changes the impedance the XO sees so yes the filter needs to change.

4) not likely

every manufacturer will measure things a bit differently, some will smoth the curves, some will average a number of curves, some will give you raw curves. These should only ever be considered a very ruff idea of what they actually will do.

dave

[FrankDIY]

Following is my own opinion based on my experience on loudspeaker design:

1) Same answer as Dave

2, 3 and 4) Impedance compensation network is absolutely neccessary in a design to linearize the impedance of the driver. Not using trick will greatly affect the crossover network. The difference between the expected curve and the measured curve will be very different. The impedance compensation network has to be before the crossover network (between the driver and the crossover network).

Building loudspeaker is not only buying high-tech drivers, putting them in a nice and exotic box. This can impress friends but not audiophile. The most important thing in loudspeaker design is measurement, measurement and measurement When you get the driver, measure all the real parameters by hand or with a software. Compare them with the one of the manufacturer. You can have a difference up to 30%. Redo the same thing after break-in (100h). After break-in, using the real parameters, you can build you impedance compensation network and re-measure driver impedance curve. From that curve, you can determine the impedance at your crossover point and than build your crossover from that impedance. And then measurement again...

All the graphs/parameters of the manufacturer are usually used only to get a rough idea of the driver.

I will greatly appreciate different opinion than mine in order to learn new things.

Good luck!