|Home | Audio Magazine | Stereo Review magazine | Good Sound | Troubleshooting|
AS A READER service, AUDIO is presenting capsule summaries of the tests used in our speaker reviews. While some of the discussion may be a bit technical, we feel that many readers may want to know why these tests are performed, how they are made, and what to look for in the data that is presented. We begin these summaries with a discussion of impedance measurement; the others will follow as space and time permit.
The first measurement AUDIO makes on speaker systems is that of the magnitude of the electrical impedance as a function of frequency. The reason we provide a complete plot of this data is that there are many tell-tale performance clues contained in such a complete measurement which you can use for your own evaluation once you know what to look for.
The measurement is made by driving the speaker from a constant current source and plotting the voltage drop as a function of frequency. Enough measurement data is provided to be technically meaningful. Thus, if a tweeter level control changes the impedance, we provide measurements capable of letting you see what happens.
For the greatest majority of speakers, the phase of the impedance may be uniquely determined from the magnitude. They are, in other words, minimum phase. Because of this, the magnitude is sufficient for the needs of the review. If it were not, then we would also plot the phase.
The simplest piece of information you can gather from the impedance is the lowest value presented to the power amplifier. We identify conditions we feel may warrant your consideration in such things as paralleling speakers for patio use.
The next thing to watch for are resonance peaks. These are perfectly natural and do not indicate problems if they are present, but watch for changes in resonance frequency with level control. If changes occur, some interaction may be suspected between the crossover characteristics and level-control setting. An inductance-capacitance
crossover network is constant impedance only if properly terminated by a resistive load. Shortcuts in level-control design show up quickly with this plot.
Other more interesting observations can be made when you use the impedance plot in combination with the frequency-response measurements. That is why AUDIO uses the same graph paper for the frequency scale of these measurements. A sealed-box, direct radiator speaker, usually called an acoustic-suspension system has a second-order characteristic in low frequency response when properly designed. That is, the response should fall off at a uniform 12 dB per octave on the low end and rather quickly approach this slope on the SPL scales AUDIO provides. This type of speaker system should have only one resonance peak in the low frequency impedance.
The design of such a system involves an interplay of parameters, and there are certain alignments of parameters which yield the response characteristic the designer intended. The position, magnitude, and shape factor of the impedance resonance is a strong indicator of what the designer actually achieved.
The exact characteristics are too complicated to describe here because of the variety of alignments, but as a general rule the frequency of impedance resonance should be at or near the low frequency cutoff and have neither an excessively high nor low "Q" if acoustic-resonance effects are to be avoided.
A phase-inverter loudspeaker system, either using a vent or its acoustic equivalent of a "drone cone" passive radiator, is a fourth-order system under normal alignment. That is, the ultimate slope in response is 24 dB per octave. More degrees of freedom are available to the designer of such a system, and this is shown as a double hump in the impedance plot. Under proper alignment, the system resonance, from the acoustic point of view, is at neither of the impedance resonance peaks but is, in fact, close to the geometric mean frequency of those peaks. That is, approximately at the impedance dip between the peaks. Thus, a properly designed vented system should have its acoustic cutoff occurring near the frequency of impedance minimum between resonance peaks and should be going down at around 24 dB per octave. If, as is sometimes the case, a vented speaker dies at 12 dB per octave starting well above the proper point, it generally means that the vent is "chuffing" fruitlessly and the bass will be a bit anemic. On the other hand, impedance peaks close together and of high "Q" are warning signs of a boom box.
The lower resonance peak in a phase inverter speaker may be associated with resonance of the woofer if the system is poorly aligned and in that case may be where maximum cone excursion occurs. If this frequency is so low as to lie in the range of record warp components, then the speaker may simply drive itself into nonlinearity at higher sound levels. This effect can vary from a distinct tremolo on piano material to an unpleasant burbling and mushy bass under extreme conditions. If an impedance resonance lies below about 20 Hz, then you should consider using a rumble filter to remove subsonic signals. It makes poor advertising copy, but very good sense, when you know what to do to keep the drivers in a linear region of operation.
Occasionally you might see significant ripples in the impedance plots which do not occur at or near the stated crossover frequency. Sometimes this is an indication that two or more of the speakers in a multi-speaker system are "talking to each other." This means that a driver which should have cut off, such as a midrange, may in fact be acoustically loading a driver in another frequency range, such as a tweeter. A favorite backyard fence for such acoustic chatter is either a tweeter or midrange without a rear cover talking to the other through back pressure in the enclosure. This invariably shows up on the pressure phase spectrum of frequency response but the impedance plot also indicates its presence.
The lowly impedance plot may thus be seen to provide a bit more insight into a speaker design than you may have thought possible.
(Source: Audio magazine, Sept. 1974; Richard C. Heyser)
= = = =
Prev. | Next