|Max dB with Doug Blackburn
Back Issue Article
The Amplifier: Whats Inside? What Makes a Difference? Whats Overlooked? Part Two
Last month I began this look at the audio amplifier by discussing the critical components and concepts in the AC power and DC power subsystems of the amplifier. This month we continue with low-level gain stages and the output stage. Here is a repeat listing if the components which make up each of these subsystems:
AC power (covered in Part One, September 1998)
DC power (covered in Part One, September 1998)
Low-level gain stages
Mechanical package (covered in Part 3 Nov. 1998)
Last month I indicated this would be a two-part article with the conclusion this month. A nice goal, but in the process of composing this article, it became obvious that this would be a three-parter. Part three will appear in November and cover all the issues in the mechanical package as well as tie together some amplifier design and sound issues.
SoundStage! reader feedback on "The Amplifier Part One"
Ken Stevens of Convergent Audio Technology, makers of the renowned CAT SL-1 tube preamps and the very expensive JL-1 tube monoblock amps, wrote a nice e-mail message about "The Amplifier Part One" that went something like this: " Yup. Yup. Yup. Yup. Nope. Yup. Yup." An amazing amount of agreement from this highly respected high-end manufacturer. The one "nope" took exception with my suggestion that using a set of film & foil capacitors with values like 4.0uF, 0.47uF, 0.01uF as bypasses for the electrolytic power-supply filter capacitors would be of sonic benefit. In Mr. Stevens experience, using a single capacitor as a bypass is sonically superior to multiple bypass capacitors -- provided you have selected the best-sounding capacitor. Mr. Stevens further mentioned that he finds most film & foil capacitors sonically problematic due to mechanical resonances resulting from the manufacturing process. Where he requires film and foil capacitors, he manufactures them himself to avoid the resonance problems. If he had to use commercially available caps, he would use metalized caps with properly terminated leads, which means not the commonly available "flame spray and solder" method. Mr. Stevens has extensive experience with parts selection for his tube preamp and amps, so Im not about to disagree.
However, in power supplies for solid-state amplifiers and CD players, CD transports, and DACs, which operate at much lower voltages than tube circuits, I have gotten excellent results with multiple film & foil power-supply-filter bypasses. Either the lower operating voltages make a difference, or the capacitors themselves make a difference, or I missed something or some combination of those. Under the circumstances, my advice would be to try a single high-quality bypass and experiment with multiple values if you want to be sure you are doing the right thing with the single bypass capacitor. I would also take very seriously Mr. Stevens recommendation for using properly terminated metalized caps rather than the film and foil caps which are all the rage for manufacturers and hobbyists these days. How do you know if you are getting a good capacitor? This is the makings of a whole different article. All I can say now is that the brands you may have heard of the most which are marketed to high-end manufacturers and hobbyists are not necessarily the most neutral-sounding and best-performing film caps that are available.
My thanks to Ken Stevens for taking the time to write to SoundStage! about "The Amplifier Part One."
Finally, lets begin "The Amplifier Part Two."
Low-level gain stages
Here the designer has a major decision. Should he use multiple gain stages with easy-to-manage amounts of gain in each stage or fewer gain stages which will be much more highly stressed by providing much more gain than would be needed if multiple gain stages were used? Think of a staircase trying to raise you six feet or so. Its easy to negotiate if there are four steps, but try using a six-foot-high staircase with only two steps -- not so easy. There is no possible way to guarantee one approach is better that the other. Too much depends on the designers circuit, his implementation of it and his budget constraints. In an ideal world, fewer gain stages would definitely be the way to go. However, the world is not ideal and there are very real barriers to putting all the low-level gain into a single stage.
I want to reiterate something Ive said about the nature of amplifiers and other audio components before, but which people still arent absorbing. An amplifier does not make the input signal bigger, as most people assume. The amplifier creates a brand-new audio signal from nothing more than the raw AC you supply to it. The original input signal is used only as a template for the brand-new audio signal that is created. The original input signal is discarded after being used as a template.
Furthermore, in each gain stage, this process of using the input signal as a template for a new larger audio signal is repeated. If an amplifier has two low-level gain stages and an output stage, the amplifier creates a brand-new audio signal three different times. You can see why fewer gain stages rather than more gain stages might be desirable -- but only if these gain stages could have lots of gain without other problems. One of the designers hardest jobs is to select just how many gain stages to use, then make his choice work as well as possible.
Input RCA jacks In most cases, you are going to get RCA jacks of reasonable quality in your amplifier once you get beyond the "moderate" price range. Ive seen perfectly acceptable RCA jacks in amplifiers priced as low as $800. The very best RCA jacks will cost a little more and sound a little better, but generally there are not revolutionary sound differences in the RCAs. The more expensive RCAs will do nice little worthwhile sonic improvements. The more expensive RCAs are likely to employ better-sounding materials and more durable plating.
The method of attaching the RCAs to the circuit is quite important. Most RCAs used today are chassis mounted. This means you make a hole in the chassis and hold the RCA in place with a large nut that squeezes the RCA tight against the edges of the hole in the chassis. To make the RCA secure, you have to tighten these nuts. Unfortunately, the tighter you make these nuts, the worse the RCA sounds. To get from the RCA to the circuit board, you carefully select a good-quality wire and solder two wires to each RCA then to the circuit board. There is an alternative. More and more companies, even high-end companies like Cardas, are offering circuit-board-mounted RCAs. These have leads that are soldered directly to the circuit board -- no wire is needed to carry the signal from the RCA to the circuit board. You eliminate two sonic bottlenecks by using circuit-board-mount RCA jacks: The wire and the tight nut. Not all circuit-board-mount (PC-mount) RCAs are good, however. Mass-market equipment uses molded plastic PC-mount jacks with steel leads and only a single round wire inside to contact the male RCA pin. These cheap RCAs are sonically quite bad. They are used almost universally in mass-market equipment as well as by some unenlightened manufacturers making fairly expensive components -- especially those involved in multi-channel home-theater equipment.
Input XLR jacks Nothing new or specific to mention about the XLRs. Same things apply to XLRs as apply to RCAs. Do be careful if you mix manufacturers equipment with XLR connections though. There are a few manufacturers who wire their XLR sockets differently from everyone else. Dont plug in until you know whether the two manufacturers use compatible XLR wiring schemes or not. If you do end up with two components with non-compatible wiring to the XLRs, virtually every cable manufacturer can set you up with properly wired interconnects.
Feedback in low-level gain stages It is becoming increasingly clear that less feedback rather than more is musically desirable. However, there is a point where you have to stop removing feedback due to some element(s) of the whole presentation falling apart, like the quality of bass. Too little negative feedback is big trouble for bass quality. How much is too little and how much is too much? Ive never been impressed with zero negative feedback in an amplifier. A little bit of feedback, even as small an amount as 2dB to 4dB in one gain stage (out of three) is enough to keep the bass in line while not enough to do harm to the loveliness of the mids and highs. Too much is easy to identify. Think sibilants. You hear too much emphasis on sibilants? Chances are the amplifier you are listening to has way too much negative feedback. In fact, a good portion of what at one time was known in high-end land as "solid-state sound" was strictly an artifact of silly-large amounts of negative feedback. It is not entirely inaccurate to think of amps using a lot of negative feedback as the uptight conservative businessman of the audiophile world -- too inhibited to ever have a really good time. They are too controlled, too unforgiving, too dry, too emotionless, too uptight. They are certainly lower in total harmonic distortion than low-negative-feedback amps, but what does it matter if the music is as crisp and colorless as fall leaves in winter?
While zero negative feedback is too little, 10dB of global or even local (within a single gain stage) is often too much. Think 2dB to 6dB of local feedback for most amplifiers to do their thing with high levels of musicality across the entire musical spectrum. You can do "stupid amplifier tricks" playing with less or more feedback, but in the end, they all eventually sound "tricky" while the very moderate amount of feedback sounds like music. Its typical for solid-state amplifiers to have 20dB to 60dB of negative feedback. Are those guys actually listening to what they are building?
There may very well be some wailing about my pronouncement that 2dB to 6dB of local negative feedback with no global negative feedback is "enough" coming from the amplifier manufacturer/designer community. Dont trust them. If they think they need more feedback than that to get good sound, they just arent listening, or they dont know how to resolve the problems that will crop up (sonic problems) when they eliminate most of the feedback they are used to using. That doesnt mean my concept of the right amount of feedback is incorrect. It just means that some manufacturers/designers arent going to know how to incorporate that small of an amount of feedback into their products. The manufacturers that can build great-sounding solid-state (or tube) amps with low amounts of negative feedback without suffering muddiness of sound and woolly bass have a major sonic advantage that you will/can recognize once you have heard it.
Is having adjustable negative feedback a good idea? I have seen several tube amplifiers which have user adjustable negative feedback. On the surface, this seems like quite a worthwhile feature. It certainly is educational to hear what happens as you change the amount of feedback. However, once you begin trying different discrete resistors that are soldered into a negative-feedback loop -- well.... Lets just say that the resistors in the negative-feedback loop are impressively obvious when changed, either in value, manufacturer or material. I cannot imagine any potentiometer added to the feedback loop would do anything but sound really bad compared to a single properly selected high-quality resistor. In fact, having heard how critical the resistor in the negative-feedback loop is, I cant imagine a worse place to locate a potentiometer than in the negative-feedback loop. Compared to a good resistor, the best potentiometers sound quite bad. Use of a stepped attenuator might eliminate a lot of problems caused by sonic limitations of potentiometers. However, the negative-feedback loop is a very sensitive area of the circuit, Im not sure if mechanical contacts in the stepped attenuator could ever be good enough to be sonically inconsequential in this location.
Capacitors, resistors, inductors These are the components (other than transistors or tubes) most often found in low-level gain stages. The quality of these components is critical. Youll hear the sound of the amplifier change when you change any one of these components for something of a different value or with the same value but made with different materials or different processes. Even the material and attachment method for the leads on the ends of the components affects sound enough to worry about. Youd be amazed to hear just how much you can affect the sound of an amplifier with a handful of capacitors of different types and values (or resistors). These kinds of component changes often obsess the audiophile-tweak hobbyist who sits at home listening to ten different capacitors in a single location. Unfortunately, these tweakers dont often have two identical versions of the amp they are tweaking, so there is no easy way to directly compare the tweak to stock or tweak A to tweak B. Beside that, it takes a day to two for resistors to break in and a couple of days to a week for a single replaced capacitor to settle in to its final sound quality. How is the tweaker going to be able to remember exactly how the previous configuration sounded after a day or a week and no reference to go back to?
Then there is the problem of accurately determining if a difference in sound (which is guaranteed when you change anything) is an improvement in sound. Tweakers are notoriously bad at this. I found myself fooled many times over the years, even after I thought I was burned often enough to finally know what I was hearing. Careful component selection for a high-quality high-end product like an amplifier is very painstaking. The sheer number of choices is staggering, and the time it takes to break in every component one at a time stretches the process out to what seems like forever. Then a single change to some mechanical aspect of the amplifier can change the sound of the entire product, undoing all the component selection that had been done up to that point. This is not a project for the faint of heart or those who are not obsessively detail-oriented.
Wiring to output stage Another critical yet overlooked element in many amp designs. What wire is used, where it is routed, how it is dressed, and how it is terminated will affect the sound of the amplifier to a significant degree. This wiring is an essential link, just as important as the interconnect between your preamp and amp. It just happens to be inside your amplifier. Most of us go to a lot of pains to select the right interconnect for preamp to amp. We should expect at least as much pain from our amp builder/designer.
However, the builder/designer has to select his "perfect" wire a little differently. He has to use wires which do the least damage to the audio signal. He should not be selecting wires which impart a certain sound to the amplifier. If he does, audiophiles who buy his amplifier and begin hanging all manner of different sounding wires off of it will end up having a very difficult time finding a wire they can get happy with. The most neutral (i.e. least audible) wire will not leave a sonic signature. It might be a bad-selling wire if it were made into an interconnect -- no character, no identity. But for this job, in this location, thats exactly what you want.
Tubes or solid state Pick your poison. Try not to be influenced by "amp cliques." You dont have to have a tube amp to be cool, and you dont have to have solid state to have a bunch of power. The goal is music reproduction. Select the one that makes the music you like sound like your personal "ideal" for the quality of that sound.
Tubes are great at amplifying voltage. Unfortunately, loudspeakers need fair amounts of current to drive them adequately. How you get current out of a tube amplifier is one of the on-going design challenges. Most often this is done with an output transformer that converts a lot of voltage to current. However, it is possible to make tube amplifiers without output transformers (OTL for Output TransformerLess). There are other design problems with OTLs that make them less all-purpose than your typical solid-state or push-pull tube amplifier. But OTLs can sure sound different.
Solid-state output stages are fraught with misunderstanding. Everything matters, even the insulator between the transistor and chassis. There are people who would tell you that you have to decide if you are "bipolar" or "MOSFET" -- two different types of transistor. You dont have to decide. Heck, you dont even have to know which kind of output device is used in the amplifier.
As long as MOSFET has been brought up, let me debunk something. I dont know where it originated, some tube amp dealer, some rabid tube audiophiles, not important. What is important is that certain non-technical high-end reviewers have used the term "MOSFET mist" in their writing, and now people think there is such a thing. Its a bunch of baloney, nothing but made-up cute words drawn from mistaken conclusions. MOSFET mist is no more useful or descriptive of anything real than Triode Tarnish or Pentode Pallor. I dont doubt that some people may have listened to X number of MOSFET amps and found them all to be sonically lacking due to hazy sound. Thats fine. But what they heard was X number of not particularly good-sounding MOSFET amplifiers, and there are plenty of those, including some made by well-known manufacturers. They were not hearing something -- specifically a trait of MOSFETs.
How do I know this? Because I have heard amps employing MOSFETs which dont contain an iota of "mist" in their sound. In fact, these amps are so "mist-free" that your average tube amp (from the middle of the tube-amp bell curve) sounds cloudy and hazy in comparison. I have no problem describing certain (numerous, even) MOSFET amps as sounding hazy or cloudy. But this is not a characteristic of the MOSFET -- youre just hearing an amp that doesnt sound particularly good. So promise to help me banish the phrase "MOSFET mist" from the high-end lexicon and I promise I wont popularize Triode Tarnish or Pentode Pallor. We dont need a bunch of meaningless obfuscating descriptive terms in wide use. Its hard enough to describe the way things sound in words without confusing things further.
Resistors In the output stage, particularly in solid-state amps, resistors are going to get big and are probably hot-running. In tube amps, the resistors may not run hot from current dissipation, but from the heat generated by the tubes. So what you want here is excellent temperature stability -- meaning the value of the resistor does not change over a wide range of temperatures. You want resistors that dont change value over time (from heat exposure). You also want resistors with no inductance -- usually. Additionally, you want very low internal noise from the resistors and you want that noise not to increase when the resistor gets hot. Today you can pick from ceramics, high-current metal film, and inductive or non-inductive wire-wound resistors. Selecting these resistors (most often the emitter resistor in solid-state amps) is another critical step in finalizing the sound of the amplifier. There have been major gains in the sound quality of high-current resistors in the last 10 years or so. Amp manufacturers who arent up to speed with these newer resistors are going to remain mired in mediocre sound.
If high current is not an issue, as in tube amps, then noise properties of the resistor become ever more important due to the higher operating voltages. They may initially sound great, but as they age, the voltage causes problems in the resistive material itself or sometimes at the junctions where the wire leads are connected to the body of the resistor.
Binding posts and wiring to the binding posts Once you have created the new version of the original input signal for the last time, you need to get that final signal from the output drivers or output transformers to the binding posts on the amplifier. The wiring here is just as critical as wiring elsewhere, only here you must use care to select wire heavy enough to not limit the current generated by the amplifier. Total elimination of wiring to the binding posts is not often possible due to physical layout requirements. Output devices are often buried in heat sinks -- not a favorable location for the binding posts. In tube amps, there is no easy way to connect binding posts directly to the output transformers. So the wiring to get from the output device to the binding post is every bit as important as the speaker cable the owner will ultimately select. However, again, the designer has to select wire for neutrality and not for a specific sound quality.
Most binding posts are connected to the chassis through holes with nuts tightened down to cause the post to grip the chassis tightly (through plastic insulators of course). If you dont tighten the nuts enough, the binding posts loosen up as they are used which leads to all kinds of problems including broken wires. If these binding-post nuts are tight, however, the posts sound worse than if the nuts are just barely tight enough to do the job. Furthermore, binding posts are usually massive, and all that extra metal around the speaker-cable connections doesnt help sonics.
There are five different types of binding posts: (1) very cheap and very bad-sounding posts with red and black plastic hex nuts on them often found in amps costing less than $1000; (2) slightly more expensive posts with red and black plastic nuts that dont sound nearly as bad as the cheaper plastic-nut types and most often found in amps costing $800 to $2000; 3) all-metal "generic" posts, a little more expensive than the good-sounding plastic-nut posts but as a group better sounding than any plastic-nut binding post, usually gold-plated and most often found in amps priced from $1,800 to $2,500; (4) audiophile "designer" binding posts, all metal, durable plating, additional design features to aid tightening, sonically as a group significantly better than groups 1-3 and obviously better made, a quality product worthy of amps costing, say, $2,500 and up; (5) "the new binding post."
I have to give George Cardas credit for this new-fangled post. He addressed every "problem" conventional binding posts have and eliminated all of them. Imagine a single non-conductive block you mount to the amp chassis. There are two holes in this block for the two non-threaded "posts." You insert the posts into the hole and solder the wires from the output stage to these "posts." The "posts" remain floating with no tension or other materials in contact with them. You hold the spade lugs for the speaker cables in place with one hand while you place a second removable block over the ends of the "posts." To tighten this second non-conductive block, there is a thumbwheel that threads into an insert in the fixed non-conductive block mounted on the chassis. With this thumbwheel, you lock down a pair of loudspeaker cables terminated with spade lugs (round lugs would work too). These stressless binding posts sound the best by a significant margin. They are a little costly, but no more so than some of the posts in group 4.
There is at least one manufacturer who sells amplifiers that cost into five figures who uses plastic-nut binding posts from category 2 in his amplifiers. I dont care if this designer lucked into the best-sounding plastic-nut binding posts in the universe. For amplifiers costing over $10,000, the only binding posts that are fair to use are those from category 4 or 5 -- if for no other reason than that they are made better than category 2 binding posts. Even if there were no sonic benefit to the more expensive posts, there is a certain standard of execution and quality that people buying $10,000+ audiophile product have a right to expect. People who buy Porsches or Ferraris would scream bloody murder if the wheels came from a Mustang or Camaro, and theyd be right. Amplifier buyers should expect only the very best when paying $10,000 or more for an amplifier, and plastic-nut binding posts are nobodys definition of "best." They have a place in audio products, but not on $10,000+ amplifiers.
A word on output transformers If they were easy to make for audio purposes, they would be a non-issue. Transformers are NOT easy to make though. In fact, tube amplifier manufacturers can spend more time on the output transformer than any other part of a tube amplifier. Everything about the output transformer matters. The most serious tube amplifiers all have hand-wound output transformers because the amplifier manufacturer cannot locate a transformer manufacturer who will build the transformers carefully enough. Maintaining perfect and uniform wire tension during the winding is critical. Spacing the windings is critical. The exact number of turns of wire in each winding is critical. I know one amplifier manufacturer who makes his transformers himself. Every amp hes ever sold has had transformers wound by him and nobody else. His transformers are so hard to do correctly that he has to be in a certain frame of mind before starting or every transformer he makes has to be re-wound. If he is in the middle of winding and loses his mindset, he has to stop making them or he just wastes time because none are usable.
This is not the kind of thing that leads to high production volumes. This is the kind of thing that gives certain high-end products a sonic edge over other products which settle for less. You see the difference in its most obvious form when you look at and listen to the transformers in a tube amplifier selling for less than $2,000 and compare them to the transformers in a $5,000+ tube amplifier. The output transformers are significant contributors to the sound and performance of tube amplifiers. The transformers also contribute considerably to the cost of tube amplifiers. The bottom line on transformers is: they are an area of specialty that is poorly understood to a great extent -- poorly understood even by people who claim high degrees of transformer expertise. The best-sounding transformers are going to be the ones handmade by a few iconoclastic individuals who were obsessive enough to at some point actually figure out all the minutia and arcane tidbits of significant information to actually make sense out of the Wonderful World of Transformers. The others doing handmade transformers without really understanding what they are doing are poseurs. Of course, you are going to be up into $3,000+ products before you begin to see transformers (power, interstage or output) that are handmade by someone who knows what they are doing. Below that (approximate) threshold, the designer is going to be selecting from manufactured transformers. There are differences there as well -- very significant differences. So even though the designer is limited in the ultimate performance that will be able to be achieved at lower price points, there will still be plenty of reasons to be very careful about selecting manufactured transformers.
Next month The Amplifier Part Three
Copyright © 1998 SoundStage!
All Rights Reserved