Measurements

Measuring the frequency response of a horn in an-echoic environments is meaningless, since the horn operation by nature depends on corners/walls. These response curves should therefore merely be considered a rough indication of the actual performance. Some peaks and dips are inevitable products of reflections from walls/furniture, etc. The measured cabinet was placed smack dab in the corner for these measurements.

Graph 1 Near-Field response from 30Hz to 20 000Hz, measured close to the driver.

Graph 2 Near-Field response from 30Hz to 1 KHz, measured close to the mouth

Comments

The measurements reveal some interesting facts:

1. There is a rise in the response curve at around 30-35Hz. This is caused by "eigentones" in my living room, and corresponds well to calculated resonant frequencies of the room.
2. Measured close to the driver, there is notable roll-off from ca. 300-400Hz. Compare this to the corresponding rise in response measured close to the mouth from ca. 300-400Hz downwards. They seem to sum up pretty well, and how did I plan this, you might ask. Well, I didn’t. In addition to the normal roll-off of the driver, what we see is a "horn design freebie", A bonus, as it were. As the horn starts to load the driver effectively, the heavy load will constrict the cone’s movements. As a result, potentially annoying mismatches between horn and driver gets smoothed out.
3. The "wavy" response curve between 35 and 50Hz is a normal ill effect of a finite, foreshortened horn. These peaks and dips get more severe the more the horn is foreshortened. The exact resonant peak frequencies can be calculated using this formula:
   f(n) = sqrt(n²(c/4l)²+fc²), where:
   l = effective horn length
   n = 1,2,3…
   Between these peaks, corresponding dips will occur.
   The fact that the ripple occurs at frequencies somewhat lower than the formula yields is an indication that the corner placement does what it’s intended to, effectively lengthen the horn. Not a lot can be done to remedy this response irregularity, short of building a horn that expands to the full mouth area. Bear in mind, though, that this effect will be most prominent very close to the mouth. Also, the previously mentioned "self correcting" effect will smooth this ripple out a bit, since the load on the driver will decrease in the dips, and increase in the peaks.
4. The rise in frequency response in the 5k-10k area is to be expected from an 8’’ full-range driver, and in this design there’s not a lot we can do about it, except try to tweak the driver. In other design situations, one could employ horn loading at the front of the driver in order to heighten the sensitivity from, say 500Hz to 4KHz.

Other measurement issues:

• It should be pointed out that there is a tolerance of about +/-2db in the equipment used.
• Also, I suspect that the roll-off from 15k to 20k is a bit more severe than is evident from Chart 1. This is probably due to reflections and/or erratic measuring microphone HF response correction. Other measurements I have done suggest a roll-off of about 15-20db at 20KHz. Again, this roll-off must be expected, but there are remedies to this. More efficient high frequency horn loading can be obtained by fitting a larger phase-plug on to the pole piece of the driver.