Speaker Project

Prototyping

Now that we have the nearfield response of the woofer. This verifies the box size and port size and we can go on to starting to worry about the crossover and overall frequency response of the speaker system.

We begin by testing for the On-Axis frequency response of the woofer. This can be done using a gate ( a start time and stop time for the measurement) which is necessary to remove reflections off walls and floors. I put the microphone about 1 foot from the driver and the direct sound so much overwhelms the indirect sound that I don't have to worry much about gating.

The low end response is inaccurate for a measurement like this because of room effects (which is why we did the nearfield test) but the high end response is very accurate. When doing this test make sure that the incoming volume levels are pretty high (with a max above 15K - I used 25K) for best accuracy. You can use the Vu meter (View / Vu Meter) for good feedback here. Note the Vu Meter can be dragged anywhere on the screen by picking it up on the border.

Anyway, I used a sample size of 32K and a single sample here and got the following on-axis results.

There's not a lot of detail in this particular chart, but it shows the measured response of the driver pretty well. Note how little detail there is below about 200Hz due to the small sample size. There is plenty of detail above, however.

The next step is to splice the measured on-axis response with the merged nearfield response. Here's a chart showing both responses pre-splice.

We splice the two datasets by opening the On Axis resource and then selecting Calculate / Splice. Select the merged response for the B dataset (to splice to) and when finished we get a new dataset. Here I've spliced the two datasets at 350Hz because it looked about right. Sometimes you will want to move the nearfield response up or down to match where it splices - you can do that by opening it and selecting Transform / Scale and adding or subtracting some number of dB. 

Here's a chart showing the spliced response

This chart shows the original on-axis in heavy black, the nearfield in red, and the spliced response in blue. I used the on axis frequency resolution so the low end resolution is minimal here. I think for nicer low-end results I would splice using the nearfield frequency resolution (the Use the B frequency resolution option in the dialog). Well, the blue curve is fine for crossover arithmetic and we know the low end response.

By the way, notice the 3dB peak at around 600Hz. If you recall we had an impedance irregularity at that frequency. Here's the matching frequency response irregularity - caused by the enclosure back wall reflection. It might be worth using a non-rectangular enclosure or trying more damping material or a waffle back to reduce this reflection. I'll leave that for later.

Also note the dip at 100Hz. We don't have much low frequency information but I also did a farfield test (non-gated) which showed the same irregularity when the microphone was farther away. That's the floor reflection (the woofer is about 4 feet off the ground) and an unpleasant but necessary part of putting a woofer above the floor.

Next we start working with the tweeter.

On to the next page.

last updated March 12, 2001