swj@fluke.UUCP (Stan Jones) (08/17/83)
First, an appology for my flame about all the Horse.... about phasing etc. I had just finished clearing my news articles after 2 wks. vacation and in one pass was a bit overwhelming!!! After thinking about it overnight, I decided to add my own thoughts on the subject. An "ideal" speaker would ba a point source, or at least radiate from an area smaller than a 1/4 wavelength at the highest frequency of interest. Unfortunatly, the Physics involved prevents us from realizing this goal except when the acoustic levels are very low (as in earphones for ex.). The reason to keep the source small is so that the sound radiates evenly in all directions (or at least most of 180 degrees). The closest practical solution to this problem is to use multiple sources which are capable of the desired sound level (without significant distortion) and then combine them so they act as a point source! Therein is the challenge! In this multiple speaker system it will still be necessary to keep each speaker smaller than 1/4 wavelength diameter at the highest frequency it will radiate (to at least -20Db down the crossover curve). In addition, it is necessary to keep both active speakers in the crossover region within the 1/4 wave area. At 1khz that is 2.5 inches!!!! If it is necessary to exceed this dimension then it is preffered that it be in the vertical plane so that the horizontal spread of the sound is still as wide as possible. The critical region for phase response is in the crossover region where both speakers are active. If the acoustic output of one speaker compared to the other (as in woofer and midrange) is not in the same phase, then the pair will not behave as a non-directional source and will have significant directional effects at different frequencies in the crossover region, seriously degrading the imaging ability of the system. A 6Db/octave crossover is itself very well behaved thru the crossover region, but since the -20Db points are so far from the "crossover point" the behavior of the speakers themselves is often very poor with respect to phase. However, in the absence of instrumentation to accurately measure the phases thru this region (and a method of adjusting them) the chances of building a successful system is probably better with 6Db x-overs and off the shelf speakers. 3rd or 4th order filters dramatically reduce the phase problems contributed by the speakers since they now both only function over a small part of the spectrum, but the much more rapid phase change with frequency makes measurement and adjustment of the crossover phase response necessary for each speaker (at least by Manuf. and Model if not for each driver). It is easy to get an approximate measurement of your speakers behavoir in this respect. Get a small wideband microphone (Radio Shack has an electret mike element for approx. $1.20) and hook it to a scope. Put the mike 2 to 4 inches from the front of your speaker and be able to move it in a vertical plane from in front of the woofer to in front of the mid-range while looking at its response on the scope. Feed an audio osc. thru your amp to the speaker and compare the relative phase of the two signals picked up by the mike when in front of each speaker. The only thing we are concerned with in this simple test is how the phase compares between the two drivers. The scope must be triggered externally by the audio signal going into the system. A correctly phased system will show a change in amplitude dependant on frequency when moving the mike between drivers, but the phase (or zero crossings) will not vary. Actually, less than 20 or 30 degrees of phase shift is excellent. Run the test over the crossover region to where each driver is down to about 10% output. Be prepared for some surprises!!!!!!! Enough for now! I have touched only the subject of relative phase between drivers to maintain dispersion not absolute phase (or time). By the way, reversing the connections to a driver of an otherwise correctly phased system will cause a cancellation of output (sound) on axis and at "crossover point", and all kinds of distortion of the accoustic pattern at nearby frequencies. Stan Jones Service Products, Fluke