Dear Editor (May 1972)

Phase Linear Amp

Dear Sir:

I am most happy to see the increase in quality of material evident in the Feb. 1972 issue of AUDIO. It is far too seldom that one sees material as good as that done by H.F. Olson, or Stanley and McLaughlin, in a popular magazine. The authors and AUDIO are to be congratulated.

In the same issue, however, I find six pages devoted to the observations and opinions of Mr. Robert Carver. Now I have no objection to the presentation of unorthodox views, but I think they should be clearly stated and the author's evidence supporting his viewpoint should show why present opinion is in error. As far as I am concerned, Mr. Carver has failed completely in both respects. My reading of the material seems to indicate that he has discovered that given equal rms ratings on three amplifiers, the one having the highest IHF m rating will sound the best. He then goes on to tell us that this proves the superiority of amplifiers having poor supply regulation.

Now really, gentlemen, this is ridiculous. Anyone familiar with amplifiers and rating methods could tell you that the "great discovery" that Mr. Carver has presented to us is an obvious fact well known for many years. The conclusion he reaches is however unusual, to say the least, for someone who is supposed to be in the quality amplifier business. He does not tell us what is wrong with the usual approach in the design of quality equipment of putting a reasonably well regulated supply in the amplifier so it can deliver something near its IHF m rating when tested for rms output. Does he object to the cost of the larger transformer and additional capacity needed, or what?

Having led us astray, Mr. Carver then leads the unwary into complete error. He proceeds to inform us that voltage losses in the average transistor amplifier power supply are approximately 30%. Now this may be true of his power supplies, but not of any of the other amplifiers sold as high fidelity products. In general, the full load voltage will be no more than 5% lower than the no load voltage in a typical amplifier. Thus in his example the supply that has 63 volts output at full mild have no more than 67 to 68 it, with no load, not 95 as suggests.

We are next presented with two pages telling us of the shortcomings of an unidentified protection circuit. Now once again the fact that some types of protection circuits cause very undesirable side effects is also rather well known. Had Mr. Carver shown in schematic form the type circuit or circuits that are prone to the problems he described he would have done us some service. This information was not included, however, and we are left to wonder just what kind of protection circuit he is talking about.

Mr. Carver next attempts to make a case for high power amplifiers, but does not indicate what type speakers would require this much power to reproduce the sound level of an actual piano. There is really no argument on this point though. If speaker efficiency is known, the amount of amplifier power required to generate a desired sound pressure level can be quickly calculated. With the normal type of electrostatic or acoustic suspension speakers used in sound systems, high amplifier power will be required to match the sound levels of an actual piano playing the type material suggested.

Having established that we should have high power amplifiers, Mr. Carver again reverts to complete error. He tells us that supply voltages of over 200 volts are required to produce an output of 350 watts with an 8.0 ohm load. This may be true if we are using Mr. Carver's power supply which has a 30% loss, but a more normal power supply with 5% or so regulation would only require a d.c. supply voltage of 155 to 160 volts.

We are finally told that of the methods to get the desired high power output Mr. Carver's choice of high voltage transistors is the best. If this is his opinion of the possible choices, fine, but why doesn't he tell everyone that this approach also involves some undesirable aspects. High voltage transistors of the type he describes have much poorer frequency response than epitaxial base types having lower voltage ratings. Many times by as much as an order of magnitude. They also can only be obtained in NPN polarity making necessary a quasi-complimentary output circuit. The large junction capacities make it necessary to add something to the circuit to control secondary crossover notches. The quasi complimentary circuit is inferior as far as basic linearity is concerned and will cause the amplifier to have higher distortion figures than a similar full complementary system. With all the facts known, some people just might prefer some other method of obtaining the desired high power.

In short, it rather appears that you have published some very bad material along with the good. I will hope that this was just a slip and that AUDIO will not contain any trash like this in the future.

-Daniel Meyer Southwest Technical Products San Antonio, Texas

We reviewed the Phase Linear 700 last June and I subsequently asked Robert Carver to tell us something about his design philosophy--which may be unorthodox but certainly produces the results!

-ED.

Dear Sir:

Thank you for giving me the opportunity to reply to Mr. Meyer's letter in the same issue in which it will appear.

Mr. Meyer seems to object primarily to my claim that voltage losses in the typical transistor power amp are approximately 30%. He makes the flat statement that, "In general the full load voltage will be no more than 5% lower than the no load voltage in a typical amplifier." We need not play any "whom do you trust" games to resolve this discordant point: Since we know that power at clipping is proportional to the square of the d.c. supply voltage, we need only compare IHF m and rms ratings advertised for several popular receivers and power amps. Such a survey reveals that the ratio between IHF m and rms power under ideal conditions (1,000 Hz and rated distortion) never is greater than 0.8, implying power supply losses never less than 10%. Since these ratings are taken at rated distortion, the IHF m figure does not perfectly reflect the no-load d.c. power supply voltage of the amplifier, thus making even 10% a low number for a best-case situation.

Without naming brands and citing specific voltage drops, I can go no further along this line of argument except to say that, given Mr. Meyer's admission that high power is not a bad idea, and that a 155 volt d.c. supply is necessary to produce 350 watts rms of audio across 8 ohms, I must point out that a power supply which doesn't "droop" at all would require either an infinitely large power transformer, or an electronic regulator equal in complexity and cost to an additional channel of amplification. Even settling for Mr. Meyer's favorite figure of 5% regulation would require our Phase Linear 700 amplifier's power transformer to contain five times as much copper and iron as it does now (the 700's d.c. supply drops 25% under full load), and we at Phase Linear don't wish to increase the incidence of hernia among audiophiles with state of the art, but power hungry, speaker systems.

In any case, power supply regulation is Mr. Meyer's game, not mine. The output operating area charts included in my article show why the change in power supply voltage from no load to full load as evident with an unregulated should be viewed an advantage not a disadvantage. The extra operating area allows this type of amplifier to supply higher power for short durations or into higher than 8 ohm speaker loads than an amplifier with the same rms rating and a regulated power supply could ever provide. I am not suggesting that conventional techniques are in any way wrong, I am simply saying that there is a better way.

The only design problem posed by the use of a power supply whose unloaded voltage rises to 200 volts is that of selecting the proper output transistor, as was discussed on page 34 of my article. I don't know how old Mr. Meyer's transistor reference books are, but a fast look at any printed in the last two years will show that the epitaxial base power transistors preferred by Mr. Meyer has an upper frequency limit only one third that of the triple diffused power transistors used in the Phase Linear amplifier. As a matter of fact, the new Phase Linear 1000-watt, two-channel instrumentation amplifier with its 0.5 mHz power. bandwidth will make its debut later this year as, among other things, further verification of the basic Phase Linear power amplifier design philosophy. Only through the use of triple diffused transistors can a 0.5 mHz power bandwidth be attained.

Certainly some clever circuitry is required to realize low distortion in a pseudo-complimentary amplifier, but we think the performance of the Phase Linear 700 as evidenced by its review in AUDIO magazine attests to the fact that these problems have been more than adequately solved by our engineering, development staff.

As a final point, I must comment on latter of protection circuits. It I possible to draw pieces of same names in the example of misbehaving protection circuitry without stepping on a lot of toes, but the circumstantial information given about the test parameters, coupled with some knowledge of currently available brands of high power amplifiers and their ratings, should enable Mr. Meyer (as it did many other audiophiles who have written and called me in the last month) to identify the amplifier in question (and for that matter the speakers used in the piano allegro test). As for the type of protection circuitry guilty of the misbehavior, it is sufficient to point out that it was a very common VI-sensing system in a popular amplifier, and further, it is not possible to safely restrict an amplifier to exactly its safe operating area by sensing output transistor emitter current alone. Either a "loophole" in the protection will result, wherein a particular type of overload will go undetected and initiate destruction of the amplifier, or a misbehavior will result, whereupon distortion will be generated during operation within the safe area limit. The comprehensive energy limiting circuit in the Phase Linear amplifier does not suffer from either of the aforementioned ills.

In summary, we see that the Phase Linear approach to amplifier design has produced an amplifier which can supply more power at lower distortion into a wider range of loads and for a more reasonable price than any other amplifier currently available; it can even be carried around by a single human being of average physical development. This achievement represents a system approach to power amplifier design that we are rather proud of and we consider it rather unlike a gentleman or scientist to label as trash an effort to communicate this philosophy to the readership of a technical high fidelity magazine such as Aunto.

-Robert W. Carver Phase Linear

Why not Watts (rms)?

Dear Sir:

Your editorial comment (AUDIO, p. 16, Feb., 1972) regarding the specification of amplifier power in rms watts accurately defines what is meant by this terminology. Essentially, we would agree that this is a "truthful" power rating. However, an academic nit-picker can compute a quite legitimate quantity for rms power and therein lies another booby trap that the audio industry should avoid.

After the publication of my recent letter [1] in the Journal of the Audio Engineering Society, McKnight [2], Eargle and Locanthi [3] have pointed out a very real pitfall that could come about if some unscrupulous salesman ever discovers the true theoretical meaning of rms power. Essentially, the booby trap is simply that this is another method of inflating a power rating ("bigger numbers sell") outside of the laboratory; that is, the theory show that rms power = 1.225 x (average power). Eargle and Locanthi clearly point out the danger of this theory and I think it well to quote their words of wisdom.

This means that a 100-watt average-power amplifier can quite accurately be called a "122.5-watt rms" power amplifier! Likewise, a manufacturer of an 81.3-watt average-power amplifier could accurately and honestly refer to his device as a "100-watt rms" power amplifier-while his equally honest but less informed competitor would refer to the same device as an "81-watt rms" device! Rms power can be computed, but it is irrelevant; there is no wattmeter which reads in terms of it. As long as nobody bothered to compute it, rms power could easily be reconciled and identified with average power in the minds of most engineers. But now that it has been shown, quite accurately, to be just one more inflated power rating, it should no longer be ignored. We all want to measure power accurately, and there are a number of statements which could be used in amplifier specifications to accomplish this. For example, "This amplifier has a power rating of 100 watts, average sinewave power (28.3 volts rms at 0.1% third-harmonic distortion measured across an 8-ohm resistive load)." A statement such as this accurately defines the terms it uses and states to a certain degree the conditions of the test. An approach of this sort is sorely needed in the high-fidelity and commercial sound fields.

To this I add my hearty amen! Prof. J. Robert Ashley Univ. of Colorado

[1] J. Audio Eng. Soc., Vol. 19, No. 9, p. 793 October 1971.

[2] J. Audio Eng. Soc., Vol. 20, No. 1, p. 46, Jan./Feb. 1972

[3] J. Audio Eng. Soc., Vol. 20, No. 1, p. 45, Jan./Feb. 1972

Dolbyized FM

Dear Sir:

In the letter which you have entitled, "Is Dolby Necessary?" (AUDIO, January, 1972), Mr. Clyde Wade, in describing the broadcast practices by which fidelity is suppressed, seems really to be asking whether or not FM itself serves any purpose.

If, as he suggests, FM is preponderantly a medium for the transmission of highly compressed, limited, and distorted program material, and the dynamic range of such broadcasts is rapidly decreasing, what, indeed, is the purpose of any attempt to improve reception quality and audience coverage?

On the other hand, there are FM stations which are committed to quality FM service, and which maintain high standards in programming as well as signal characteristics. Many of these stations have expressed interest in the Dolby system because it serves these ends, others have already started Dolby broadcasting for limited periods and several others are considering full-time Dolby broadcasting.

Like other broadcasting developments, such as the availability of color TV transmissions, the Dolby system requires listener equipment changes only if the listener wishes to obtain the noise reduction benefit the system provides. Since Dolby and non-Dolby broadcasts are compatible, as in the color example, there need be no forced obsolescence. In fact, the Dolby encoded FM signal is clearly and entirely within the FCC's fidelity standards.

Mr. Wade's final point is that the Dolby system's use "bypasses the reasons" for FCC transmitter power limits. This implies that there are reasons for these limits other than prevention of interference between services and stations. The power limits exist to protect quality, so does the Dolby system. Fortunately for listeners, one does not exclude the other.

-Robert Berkovitz; Dolby Laboratories, Inc.

London Viva Audio ETC!

Dear Sir:

Welcome back to "Audio ETC"! Every reader of AUDIO has missed Canby's incisive monthly commentary, and every writer in the field has yearned once again to savor his inimitable style. Canby can say more with his incomplete sentences and anomalous word use than do most of us who strive to polish our prose to the nth degree of grammatical rectitude.

-Craig Stark, New York, N.Y.

Quadraphonics Issue

Dear Sir:

In the Editor's Review for December, 1971, you comment on some reader's reaction to October's quadraphonics coverage. That issue has to be the most informative and thorough coverage yet given to this subject. Although I am not terribly interested in quadraphonics, I certainly feel much better informed.

D. Ewart; Vancouver, B.C.

(Source: Audio magazine.)

= = = =