Editor: Being both a degreed electrical engineer and an audiophile designer is walking the razor's edge. Too many "engineers" design poor-sounding products when they become slaves to the test equipment. On the other hand, too many "seat-of-the-pants" designers commit faults in the mundane engineering aspects of serviceability, stability, and knowing the limitations of the parts. With the quality of today's parts, there is no excuse for poor reliability. When I hear a manufacturer blaming a part for his reliability problem, it is all too often that the designer chose a grid resistor that was much too large to hold bias stability, a pass transistor in a regulator that was taken beyond its safe area, a screen voltage that was too high, or is simply running the product too damn hot.

Music Reference amplifiers have a failure rate of less than 4% in the first five years and less than 5% in the next five years. Usually what fails is something simple, like a power switch or a rectifier. There are no catastrophic power-supply failures or rooms filled with smoke from exploding electrolytics or resistors that fry when a tube fails. We do not achieve reliability through a modification program; we start with reliability.

I tend to find good correlation between what the ear hears and what test equipment reveals. I use both methods, along with a collection of difficult loads, to design for "real world" conditions. I can't speak to what is going on in the industry with amps that measure poorly, why some people build them, or why some people like them. I simply want people to know that they need not fear embarrassment when measurements are taken on my amplifiers.

John Atkinson's comments were reminiscent of those made by Dick Olsher when he reviewed the RM-9 in December 1989—"Too many choices"—but I can't see making an amplifier with just one set of output taps or limited to a single type of tube. Years ago, I discovered that the 6550, KT88, and even the EL34 were more the same than different. (This was the subject of the debate in the Manley-Modjeski letters of 1990-91.) Though EL34s result in about 5% less power in the RM-200 and 3dB more THD, they work well, making the RM-200 a fine test bed to compare their sound. As in 1989, I am not suggesting rolling these tubes into other makers' amps; they have to be designed with that in mind.

Through my research, I determined that the much-maligned Chinese 6550 was capable of the highest peak-cathode currents of any 6550 available, and about 1.5 times higher than the venerated Genelex KT88. Once the grid-leakage problem was solved through the driver design, it was full speed ahead with the Chinese tubes. Granted, I have to toss quite a few, but far fewer now that my driver circuit can tolerate grid leakage. As to rolling the driver tube, if you like the 6922 or its higher-current 6H30 version, go ahead, the circuit won't mind—though in this application I prefer the 6BQ7 for its ruggedness, less delicate grid, and higher current. Just be sure the DC is matched in the two sections, as it is directly coupled to the output grids. We want to keep the bias balance within 10% in the output tubes.

I always read the amp reviews first and go right to JA's test section. As a designer, what I am looking for is a few tests to size up the competition. Although the families of curves that show THD vs frequency at low powers and various loads are interesting, what I really want to see is the same curves at full power or just a bit below clipping. Those will tell almost all you need to know about an amplifier's linearity at various frequencies and its ability to produce power at those frequencies. I noted JA's concern about full-power testing on p.102 of the January 2002 issue: in his experience (and mine), "that is when some amplifiers will break." If an amplifier blows up doing this, so be it. Send it back to the manufacturer and inform the public. When are we going to start testing for reliability? How are we going to separate the manufacturers that make sturdy, useful products from those that use consumers as product testers?

I believe that much of an amplifier's sound is in the final tuning of the feedback network, the internal compensation, and, finally, its interaction with the load. Rather than tune my amplifiers to fly through the HF IM test, I trade off some linearity above the audioband, where it won't do you any good, for something that will do you some good—namely, stability into all reactive loads, smooth clipping and overload recovery, and (at the low end) full-power output (-1dB) to 25Hz. Just because an amplifier is stable into the standard 2µF IHF load, there is no assurance that it will be stable into much smaller capacitances—like the 0.01-0.1µF present in many high-end cables.

If you want to see some ringing squarewaves, give those values a try with a 1kHz squarewave from zero to full level. At best, you will see oscillations that top the squarewave by 50%, or obliterate the test signal with full-level oscillation. At worst, you will see smoke, the test will be rapidly terminated, and there will be gnashing of teeth followed by much sadness over the dead amplifier. There are all too many of these "conditionally stable" amplifiers out there, and this is my explanation for many cable/amplifier interaction problems. How about testing with simple capacitive loads from 0.01 to 10µF (that's what I do)? Sure, some amplifiers would go back to the manufacturer broken, but in the long run, we might get more stable amplifiers.

I encourage JA to add these tests to his arsenal, and let my amp be the first. The RM-200 will take any load you want to throw at it and never break into oscillation. Conditionally stable amplifiers are an accident waiting to happen, and the bigger the amp, the bigger the accident. I think of them as the Tasmanian Devil, just waiting for the right conditions to oscillate and then watch out—there go your tweeters, there go your woofers, smoke billows, and you are left with a smoldering heap.

I thank Michael for taking the time to read the manual and for his generous quoting from it. One part he did not mention, which I think is of all the more importance in today's world, is the subject of power consumption. Other than using the argument that if you can afford the fancy sports car you ought not to complain about the gas or maintenance costs, we should consider the proper use of energy. Are we justified in consuming a kilowatt to listen to a few watts of music? Not to mention the cost of the additional load on the home AC system. All MR amps draw less than their rated power at idle (which is where you do most of your listening). The RM-200 draws just under 140W at idle, and at loud levels will budge only another 30W. (Remember, the average power of unclipped music is between 0.1 and 0.01 the peak power.)

Living in the "shadows" for the last 10 years, I have had time to think more about what amplifiers should be, to explore architecture while teaching at Taliesin, to throw pots, study the Tao and right livelihood, consult for two local guitar-pickup companies, and finish an electrostatic speaker that has been gestating since before I joined Harold Beveridge. Frankly, I'd rather consult than produce. I'd rather further the industry by consulting on designs for major manufacturers than by naming names and admonishing them for their engineering oversights. My passion is invention and design—solving problems that have long been awaiting solutions. As always, my services are available to any manufacturer, consumer, or those who want to get started building fine audio equipment and know that good design is the product of a lifetime's study.

If you live in Santa Barbara or care to travel here on Tuesday nights, you can attend my classes in "Consumer Electronic Equipment Repair" or "Basic Electronics from a Historical Perspective" at the local Adult Education Center. If you come next semester, you can learn (AE administration willing) how to build your own tube amp and voice it to your pleasure. Hey, if you're really dedicated to audio, come out here and start a company to build my designs so I can do more research.—Roger A. Modjeski, Music Reference

Company Info Music Reference PO Box 40807 Santa Barbara, CA 93140 (805) 687-2236 www.ramlabs-musicreference.com		Article Contents Page 1 Page 2 Page 3 Page 4 Specifications Associated Equipment Measurements Measurements part 2 Manufacturer's Comment