Zaph|Audio - SB Acoustics 2-Way : Final Crossover Measurements

Final Crossovers - Selected Measurements

Measuring the results of any design is always recommended. These designs were all auditioned using the Digital Filter function in SoundEasy where mistakes of assembly simply cannot occur. When the final crossovers are constructed mistakes can easily occur, so it's always a good idea to measure the assemble crossovers in the system. This also acts as a final sanity check that the design target was achieved.

The measurements here are for those crossovers actually constructed. Some additional measurements were taken of some impromptu testing/tweaking of one crossover setup when additional felt was tested. The final crossover/scheme that I will keep for personal use includes some additional felt applied around the tweeter with a small tweak in the crossover to adjust for it. That was not incorporated into the previous design pages to keep the design as plain and straight-forward as possible. It's an easy change to make if you have any interest in it. It's certainly not required, just the way that I prefer to do it. It does also help to explain why trying to get a flat response without diffraction control may be counter-productive.

This set of measurements arose because something sounded just a little off when listening. I had forgotten to include the trap for the tweeter Fs in my haste. Seeing the actual crossover prompted me to try one without the trap. A check in software showed that dropping the 20mfd cap to 10mfd in the tweeter crossover was all that was necessary. Measurements confirmed that it could be left out if an "ideal" LR4 highpass isn't an issue.

You can see the change made with the trap left out vs. without it and lowering the cap highpass around the crossover area. Red is the response with the trap accidentally left out. Yellow has the correction with the trap eliminated and blue is with additional diffraction control (a small felt triangle).

The additional felt for diffraction control and with the changes made to eliminate the trap are reflected below.

To better demonstrate the change that can occur with a simple application of felt for diffraction control, let's look at the measurements of the last crossover with and without the felt. The difference is surprising, even to me, although I've got a lot of experience with felt. The change actually exceeds 2db at 6k.

The actual perceived response change is smaller than these measurements would indicate. The reason is that we perceived more than just the direct response. The off-axis has an influence. A single measurement when there is no diffraction control is not representative of the power response, that is, the integrated response that is the entirety of all the off-axis area. When diffraction control is used, a single on-axis measurement is closer to representing the power response for a single driver above the baffle step, roughly 3kHz up for a tweeter on a typical baffle. If you look back at the quasi-anechoic large baffle measurement for this tweeter, you'll see that it does not have the broad peak at 6kHz.

Peaks such as that shown here at 6kHz may make one try to equalize it out when designing the crossover, but doing so may not be appropriate.

Change due to felt also is apparent in off-axis measurements. We can see just how much influence it can have in the measurements below. These were taken at 90 degrees (mic parallel to the baffle) on the inside (short) edge of the baffle, 1m from the tweeter axis. This is a system response with the crossover in place. It's easy to see that the area affected is very much in line with that of the on-axis responses above.

The remaining measurements were made before the trap error was found, but had been left in due as a demonstration. These are the system Response, 5 Degrees Off-Axis, With (Yellow) and Without (Blue) Felt Triangle Around Tweeter

One benefit of good diffraction control is the increased uniformity in the off-axis. Typical listener placement is on-axis to 5 degrees or a bit more off-axis. We can see this demonstrated in the measurements below that show the on-axis with the additional felt and at 5 degrees off-axis, to the inside (closer) edge.

Finally, let's look again at several responses overlayed. These are more of a re-cap. The system responses of the original crossover are shown without felt (yellow), then with the felt added (blue), then after padding changes to compensate slightly for the felt added (red).

David L. Ralph © 2009