BURST OF CREATIVITY
At the Summer Consumer Electronics Show this past June, NAD appealed to the audio press to support a movement for the reconvening of the Electronics Industries Association (EIA) Amplifier Standards Committee. The company's aim was the adoption of a new power rating which NAD tentatively called Musically Effective Burst Power (MEBP). The company further suggested that "a new EIA amplifier standard should also mandate that all power ratings be expressed in decibels, corresponding to the way the ear hears, rather than in linear 'watts' " and went on to say that a "dBW-based rating system would go a long way toward reducing consumer confusion and rationalizing amplifier design." Since both concepts were originated by me when I was chairman of the Institute of High Fidelity (IHF, now the EIA) Amplifier Standards Committee, I heartily approve of the concept. It might be a good idea, however, to review how the present standard came to be and what NAD's objection to it is.
During the time when the IHF was developing its Standard Methods of Measurement for Audio Amplifiers (1975 to 1978), the FTC issued its now (in)famous rules for permitted disclosures of home amplifier power in ads and spec sheets. The FTC rules stated that if a power specification were to be claimed for a home amplifier, the amplifier must be preconditioned for one hour at 1/3 rated power and the primary power rating must be based upon the amplifier's ability to deliver continuous power, measured in watts, with all channels driven, into a specified impedance, over a specified bandwidth, at a specified maximum level of total harmonic distortion. It was left up to the manufacturer to choose the impedance, bandwidth, and THD level, but all three were required to be stated as part of the power disclosure. Eight ohm loads and a bandwidth of 20 Hz to 20 kHz were so commonly cited that many thought those numbers had been imposed by the FTC. The FTC ruling was issued because some manufacturers on the fringes of high fidelity had been taking liberties with the existing IHF music-power rating and were making grossly exaggerated claims for their products. The committee I chaired was well aware that music is not composed of constant-level sinusoids and that the FTC, in an attempt to protect the music listening public, had neither understood nor served its needs very well. Nonetheless, the law was the law, and to write a standard that purposely flouted it did not seem appropriate. Thus, the IHF/EIA primary power rating was designed to be compatible with the FTC regulation, and a dynamic headroom specification was created that, according to the foreword of the standard, "addresses itself to the power output capability of an amplifier when that amplifier is called upon to handle music waveform signals, rather than continuous sine-wave signals." Furthermore, the dBW (power expressed as dB, re: 1 watt) was introduced as "an alternate recommended logarithmic power-rating scale" (italics mine). What we were attempting to do--or at least what 1 was attempting to do, since I can't now speak for the other committee members in this--was to avoid going to the mat with the FTC (which would have made us appear as an industry group trying to cover up deceptive advertising by our members), while leaving the door open for perceptive manufacturers to create products that had the headroom needed to reproduce real music at high listening levels without the gargantuan and costly power supplies and heat sinks that would be needed for a high FTC power rating. The committee chose not to call the new specification "music power" because that was what caused the FTC to intervene in the first place. Nor did we wish to use the term "dynamic power" because we feared that two "power" ratings potentially would be confusing.
Speaking for myself, I had hoped that specifying dynamic headroom in dB might convince manufacturers to start rating power (at least as a supplement) in dBW. Since human perception of the relative loudness of two sounds is approximately proportional to the logarithm of their acoustic power ratio, a dB-based power rating system seemed to me to make good sense.
Certainly, if dynamic headroom was meant to indicate how much louder an amplifier would play music than the FTC power rating might suggest, a dB based specification would be most appropriate. Hence, dynamic headroom was born.
The committee now needed to come up with a suitable test signal and methodology. Of all the members, only Hitachi produced experimental data that related to the duration of peak levels in music. Their studies suggested that music peaks seldom exceed 20 mS in duration and occur no more frequently than once per second. On the basis of this, the IHF dynamic headroom test signal was established as a 20-mS burst of a 1-kHz sine wave, repeated at 0.5-S intervals, superimposed on and phase-coherent with a continuous 1 kHz sine wave at a level 20 dB below the burst. The continuous sine wave was meant to represent the average music power; the more stringent 0.5-S interval (rather than 1S) was adopted for ease of measurement.
The dynamic headroom rating was to be established by observing the output of the amplifier on an oscilloscope and determining the level at which the output stage went into clipping on the high-level portion of the burst. The dynamic headroom rating was then defined as "the ratio of the average power of a sine wave, having the same peak-to-peak voltage [as the signal at clipping], to the continuous average power output rating ... of the amplifier, expressed in decibels." The committee realized that the accuracy of this measurement would depend on how clearly one could establish the clipping point by visual observation. Since most solid-state amplifiers clip sharply, visual observation could be quite precise.
For those that do not, the measurement would be less precise--but it was, after all, an auxiliary measurement, not a primary one, so some imprecision in some instances could be tolerated.
As is often the case-especially when a standard is developed in the United States and affects products that are, for the most part, designed and manufactured in the Far East-publication of IHF-A-202 was greeted with sublime indifference. Even now, 11 years after publication, certain significant innovations of the standard (notably the input terminations and control settings to be used when measuring sensitivity and signal-to-noise ratio) are often ignored. Dynamic headroom did seem to stick, however, and most high-fidelity amplifiers, whatever their origin, now often carry a specification for it.
This is usually in the range from 1.5 to 2 dB, which is about what one might expect from a rather loosely regulated power supply that provides a higher output voltage when its filter capacitors are fully charged.
Unless the power supply is very well regulated, a dynamic headroom of 1 or 2 dB comes almost automatically. It is possible, however, to achieve much greater dynamic headroom by using a "smart" power supply-that is, one with two output levels. Normally the amplifier operates from its higher voltage supp y and can deliver substantial short-term output power; as the amplifier heats up, it switches to a lower voltage supply to limit the power dissipation and thus the need to provide extra heat-sinking and power-transformer capacity. The low-voltage supply determines the continuous (FTC) power rating; the high-voltage supply establishes the short-term power output or dynamic headroom. With this type of supply, one can design for whatever dynamic headroom is desired, and one can provide that headroom for much longer periods than the 20 mS called for in the IHF/EIA standard. In fact, the designer decides how long to provide high power; a little extra heat-sinking and transformer capacity can double, triple, quadruple, or further increase the short-term power duration arbitrarily. NAD was one of the first companies to capitalize on this concept, which was, after all, the purpose of having a dynamic headroom specification in the standard. For years, NAD amplifiers and receivers have carried rather modest continuous power ratings but have had dynamic headroom specifications higher than most of their competitors. As such, they represented good value in the minds of many music lovers.
Back to the present: At the June CES, NAD made a plea to abandon FTC continuous power as the "primary" power rating and replace it with a dynamic power rating based on response to a test signal similar to the IHF dynamic headroom signal but of substantially different duration and duty cycle. For the time being, NAD suggested a 200-mS tone burst repeated at 2-S intervals and superimposed on a continuous tone 12 dB lower than the burst. (For the future, NAD suggested they would develop a more complex test signal.) The 200-mS burst is said to correspond to the peak durations that NAD found to be typical of music, and the test signal's 15% average-to-peak ratio is said to impose the same thermal stress on the amplifier as compressed rock music.
To demonstrate the validity of its proposal, NAD played various selections through its own 50-watt (FTC) Model 7100 "Power Envelope" receiver and through five competitive receivers that were FTC-rated from 60 to 90 watts per channel. The outputs from all receivers were matched and padded down before driving the speakers so that, even at reasonable listening levels, all devices would be driven into clipping during peaks. The square of the output voltage (which, into a resistive load, is proportional to output power) was displayed on an oscilloscope so that the attendees could visually observe clipping as it occurred. Indeed, the NAD 7100 did not clip until much higher output levels than the competitive products, and the peak duration of the musical sforzandi often exceeded 20 mS in duration. The demonstration was so dramatic that one waggish technical editor called for the Standards Committee to be reconvened there and then.
As committee chairman (I'm told one never really gets rid of the job unless one resigns from it), I am in favor of reconvening the committee to examine the NAD proposal and the data generated in support of it. The difference between a 20-mS and a 200-mS peak duration is clearly substantial and should be investigated by one or more independent parties. In observing NAD's CES demonstration, t noted that most often (but not always), long-duration clipping occurred on very low-frequency signals whose period (the duration of a single cycle) exceeded 20 mS. If those signals are present in music at sufficient level to cause clipping, the peak duration, ipso facto, will exceed 20 mS. The time at which the Hitachi study was performed implies that it was based on analog tapes or discs, so it is quite likely that very low frequencies were not present in the recordings at full level. NAD (properly) used CD recordings, which can carry arbitrarily high levels of low-frequency tones. If the above hypothesis is true, NAD is likely to be on the right track.
Nonetheless, further investigation is needed. As stated above, in NAD's demonstration all amplifiers were forced into clipping on a fairly regular basis. This is fine for demonstration, but it is not the ideal way to conduct an experiment to establish the duration and recurrence of peak levels in music. (Indeed, upon questioning, an NAD spokesperson stated that the original experiments were not performed with the demonstration setup.) In my opinion, one must perform the experiment below the clipping point, and one must categorize the peaks according to their duration and level--not a simple task by any means. In the NAD demonstration, it is conceivable that short-duration peaks, substantially higher in level than the long-duration ones, were occurring and being clipped even more frequently. NAD's position, if I understand it correctly, is that short-duration clipping cannot be heard and therefore can be ignored.
Even granting that the NAD tone burst is more representative of music than the present IHF burst, I foresee problems using it as the basis of the primary power rating. It will be difficult to quantify distortion or bandwidth when using a noncontinuous test signal (I can conceive of ways to do it, but they are likely to be of questionable accuracy), so the NAD proposal may bring us back to the days of rating power independently of bandwidth or distortion. Many would consider this a step in the wrong direction.
Furthermore, NAD did not make clear how its proposed test signal would be used to rate power. The IHF/ EIA dynamic headroom methodology, i.e., driving the amplifier to the clipping point on the tone burst, can be justified to establish an auxiliary rating such as headroom, but it is not appropriate (in my opinion) for a primary rating. With some amplifiers-notably those with NAD's "soft-clipping" circuit-it is very difficult to get an accurate reading using a "clipping-point" criterion.
I commend NAD's work and believe the Amplifier Standards Committee should review it to determine whether the present dynamic headroom test should be modified. At the same time, the committee can decide whether the results of such a test are or are not appropriate for use as an amplifier's primary power rating. Change, in this regard, would imply a change in FTC regulations, which presumably would require political action.
(adapted from Audio magazine, Nov. 1989)
= = = =