Patent Application
20190394590
The present invention relates to sound reproduction and loudspeakers used in sound reproduction, and more particularly relates to the use of test signals to evaluate the performance of loudspeakers.
Test signals are widely used by audio professionals to evaluate loudspeaker performance. Heretofore, such test signals, which include white noise, pink noise, and sine sweeps, all have a relatively constant and relatively low crest-factor as a function of frequency. The difficulty with this is that the crest factors for most live signals that microphones and loudspeakers need to reproduce (speech and music) are not constant but rather increase with frequency, while having an average level that decreases with frequency. As a result, the tests performed by conventional test signals fail to produce test results that correctly reflect how the loudspeaker will perform under real life operating conditions.
The present invention overcomes the above drawbacks with conventional test signals by providing the facility to produce a test signal whose average level and crest factor more closely approximates real signals.
In one aspect of the invention, a noise generator is provided for generating a test signal for measuring the performance of a loudspeaker over an operating broad band of frequencies ranging from low to high frequencies. The noise generator is comprised of a broadband random noise source for generating broadband noise over the operating broad band of frequencies of the loudspeaker, and an impulsive noise source for generating random impulses of noise. Means are provided for equalizing the broadband noise generated by the broadband noise generating means, and for separately equalizing the random noise impulses. The equalized broadband noise and the equalized randomly generated noise impulses are combined into a composite noise signal that becomes the test signal.
In accordance with the invention, the broadband noise and the randomly generated noise impulses are equalized to produce a composite noise signal having a desired crest factor as a function of frequency. In particular, the two separate noise sources can be equalized to produce a composite noise signal having an average level and a crest factor as a function of frequency that approximates real signals. To achieve this objective, the means for equalizing the broadband noise generated by the broadband noise source is configurable to reduce the average level of broadband noise at high frequencies within the operating broad band of frequencies of the loudspeaker. Both the broadband noise and the randomly generated noise impulses can in turn be equalized to produce a composite noise signal having a crest factor that at high frequencies is larger than the crest factor of the broadband noise alone. Preferably, the broadband noise source generates pink noise and the means for equalizing the noise impulses generated by the impulsive noise source is configured to reduce the low frequency energy level of the noise impulses.
In a further aspect of the invention the reduction in the average level of the noise impulses is achieved by controlling the average rate at which random noise impulses are generated while preserving the randomness of the rate.
The invention is also directed to a method of measuring the performance of a loudspeaker over an operating broad band of frequencies ranging from low to high frequencies, comprising the steps of generating broadband noise over the operating broad band of frequencies of the loudspeaker, randomly generating noise impulses, separately equalizing the broadband noise and random noise impulses, combining the equalized broadband noise and equalized random noise impulses into a composite test signal having a crest factor, and driving the loudspeaker to be measured with the composite test signal. In accordance with this method, the broadband noise and random noise impulse are equalized to produce a crest factor for the test signal that increases with frequency.
The crest factor of a signal is defined as the ratio of peak value to the rms value of the signal's waveform. The crest factor for a sinusoidal waveform, such as that which a pure resistive load would draw, is 1.414 since the peak of a true sinusoid is 1.414 times the rms value. Crest factors in noise signals play an important role in determining whether a noise signal used to evaluate the performance of a loudspeaker accurately does so for real life operating conditions. Test signals that do not have crest factors and characteristics that cause a loudspeaker to respond as it would with real live signals, such as are present in speech and music, are not going to provide an accurate indication of the loudspeaker's true performance with real signals.
The source of impulsive noise produces impulses that occur at random intervals. Preferably, means are provided, suitably within the impulse noise source but possibly external to this noise source, for controlling the average rate at which noise impulses are produced. The average pulse (or firing) rate could, for example, be 6 pulses per second, which can be adjusted up or down. However, while the average pulse rate can be fixed, it is understood that the time interval between pulses firings remains random. It is important that this randomness be preserved. As further described below, the ability to adjust the average firing rate of the impulse noise source will provide another tool for achieving desired crest factor characteristics in the output of the noise generator.
It is seen that broadband noise source 13 and impulsive noise source 15 are situated in different noise signal paths, with the broadband noise generator being in a first noise signal path 17 and the impulsive noise generator being in a second noise signal path 19. The output of noise generator 11 is a composite signal produced by summing these two noise signals together, as denoted by the summation point 21 illustrated in
In the illustrated embodiment, processing the separate noise signals in each signal path is achieved by filters and gain controls in the signal paths, which apply separate equalizations and provide separate gain controls to the two noise signals. As shown in
As to Filter 1, its general purpose is to reduce the average level of the broadband noise at high frequencies, generally above 500-1000 Hz. The general purpose of Filter 2 is to reduce the low frequency energy in the impulse noise, generally below 500-1000 Hz. Most suitably, Filters 1 and 2 and Gains 1 and 2 are configured such that the crest factor of the composite noise signal output gradually increases with frequency to relatively high crest factors at the highest frequencies. For example, it is contemplated that the filters and gain controls in each signal path can be suitably configured and adjusted to achieve crest factors in the range of 20 dB to 30 dB at the highest frequencies, for example, above about 16 kHz. At low frequencies, crest factors can be achieved that are relatively low. For example, it may be desirable to provide for crest factors in the range of 9 dB to 13 dB. Particular low to high frequency crest factor characteristics can be established in accordance with the contemplated use of the loudspeaker under test, and, as an example, could be made to gradually increase from less than 10 dB to a crest factor ranging up to 30 dB over the frequency range of the loudspeaker.
Thus, it is seen that invention is basically a test signal which is the sum of two signals which have separate equalizations applied to them, and which preferably also have separate gain controls. The two signals are: a broadband continuous noise, and an impulsive source which fires randomly but at a prescribed average rate. The average firing rate of the impulsive noise source, the equalization of the two sources, and the relative level of the two sources are chosen to produce a composite signal which, among other things, most suitably has a crest factor that is not constant and that at high frequencies is relatively large compared to conventional test signals.
The method of configuring Filters 1 and 2 to achieve the desired composite noise signal is illustrated in
If the answer is “no” at decision point 37, that is, if the low frequency crest factor in the composite signal is not too high, it is next determined whether the high frequency crest factor is too low as indicated by decision block 39.
If at decision point 39 the answer is “yes,” that is, if the high frequency crest factor is too low, both Filter 1 and Filter 2 can be adjusted. In addition, adjustments could be made to the gain in the impulse signal path (Gain 2) and/or the average firing rate to the impulsive noise source 15. Filter 1 can be adjusted such that the high frequencies of broad band noise in the first signal path 17 are decreased; Filter 2 can be adjusted such that high frequency energy in second signal path 19 (impulse noise path) gradually increases at high frequencies (block 41). In conjunction with these filter adjustments, Gain 2 can be increased and/or the average firing rate of the impulsive noise can be decreased to achieve the contemplated very high crest factors needed at the highest frequencies (e.g. in the range of 20-30 dB).
If and once the answer is “no” at decision point 39, the configuration of the configurable and adjustable parameters of the noise generator is complete, that is, is at an end as indicated by block 43. It will be understood that the above-described configuration steps can be performed in any order.
While the present invention has been described in considerable detail in the foregoing specification and the accompanying drawings, it is understood that it is not intended that the invention be limited to such detail as necessitated by the following claims. For example, the controllable parameter for the two noise sources (broadband noise and impulsive noise) can be configured or set to achieve crest factor increases over the operating broad band of frequencies within ranges other than indicated above. The controllable parameters could be configured or set to produce lower or higher crest factors crest factors at low frequencies or higher or lower crest factors at the upper ranges of crest factors at high frequencies. Also, the increase in the crest facto with frequency could be something other than a monotonic increase, though a monotonic increase would be preferred.
This application claims the benefit of U.S. Provisional Patent Application No. 62/688,208 filed Jun. 21, 2018, which is incorporated herein by reference.
| Number | Date | Country | |
|---|---|---|---|
| 62688208 | Jun 2018 | US |
