From the orginal TubeLabs site
Copyright © 2000 Alan Kimmel. All Rights Reserved
High Gain Followers
Some audiophiles and mfrs. stereotype certain things as sonically incorrigible, based on their auditions of some outmoded implementations of those things. Some SE advocates pick on push-pull and vice-versa. Some tube advocates pick on transistors and vice-versa. On and on goes the list. FYI, if handled correctly all these things can sing ever so sweetly as people are discovering. Those who won't even consider such a notion are second guessing their ears."
For example, take the push-pull controversy. There's no shortage of SE fans who denigrate PP. But there are some PP amps coming out, notably some from some Japanese designers, which leave no doubt as to the superb sonic potential of PP, and SE fans who audition these PP amps are eating their words. As improved and corrected implementations of previously disfavored technologies emerge, the same scenario is shaping up for many things which, again, some have stereotyped as sonically and politically incorrect. Another example of this is cathode followers (CF).
Most of the old-time tube hifi designers had high regard for CFs. Back then, test equipment was 100% vacuum tube and CFs were used in them a lot, for good reasons. I once had a tube HP scope which used many CFs. It was very good and would still be a pleasure to use today if not for its size. The old Tektronix scopes were capable of excellent performance also and some of that performance is not easily emulated even today. Followers and buffers always have been, and continue to be, highly regarded and often used in the electronics industry.
Some people believe CFs have no place in audio but these people will use local current feedback with their plate-loaded stages by leaving the cathode resistor unbypassed. A CF uses local feedback too. In fact, when NFB is local, it can be sonically quite good. NFB can't get more "local" than in a CF. Such local NFB is instantaneous and virtually equivalent to inherent linearity, with no time delays or loops to be concerned about. Accordingly the ear does in fact approve of well designed and properly applied CFs. [NFB by itself is not evil; it goes bad when, like anything else, it is abused or misapplied. See Myths That Plague Tube Audio.]
There are 3 basic ways to connect and operate a tube. Engineers call them: Common Cathode, Common Grid, and Common Plate. Of the 3, the Common Plate connection, a.k.a. the CF, has the most ideal characteristics: lowest distortion, widest bandwidth, lowest output Z, and lowest noise. It clearly doesn't make sense to attack the best of the 3, yet this is the one that some have chosen to attack. (Isn't that often what happens in this crazy world--that the best gets attacked?) If you must attack something it would make more sense to attack the Common Cathode since of the 3 it has the least ideal characteristics. But I say don't attack any of them. It's much better to explore ways to optimize all 3, and that's what I do.
A properly designed and applied CF does bring out the best of a tube-- as mentioned above all parameters except of course voltage gain are greatly improved: frequency response, distortion, output impedance, input impedance, etc. Best of all a properly designed and applied CF is most faithful to the MUSIC. Those who are prejudiced against followers of any kind, solid-state or tube, should visit my High Gain Followers page. (By the way most of what I'm saying about CFs applies to the other types of followers as well.)
Using a CF after the voltage gain stage permits the v.g. stage maximum sonic freedom because the v.g. stage doesn't have to drive a load when it works into a CF-- it doesn't have to supply current to a load (the CF supplies the current to the load). This is the condition of maximum freedom for the v.g. stage as discussed in the Introduction. Sonically, triodes are the ideal voltage gain device. And triodes prefer to be unloaded. For this to be possible you must follow the triode with a CF so that the triode sees no load. In the Kimmel Mu Stage not only is the CF relieving the triode of the burden of driving the load, the CF also enables the triode to operate in the ideal constant current, or mu, mode. But let's say you want the voltage gain stage to operate at a high enough current level that it will have a fairly low output impedance to drive the load without needing a CF. As soon as you place a load on such a stage, the load is robbing the stage of current and in the process, it also robs the stage of some of its sonic freedom. This problem is avoided when a CF is properly applied to the situation.
If you throw away CFs you are left with plate loaded tube stages. It's not hard to find bad things to say about plate loaded stages. Everyone has heard bad or mediocre plate loaded stages. Does this mean I'm against plate loaded stages? Not at all. My point is simply that you can find bad things to say about anything. And people do. There's not one device, circuit, or idea in audio that doesn't have its critics. I'm open minded I prefer to avoid criticizing and instead concentrate on exploring new and better ways to do things, like optimizing all the tools and circuit elements available to us.
It's easy to design and optimize a CF for the desired task. Like most things if a CF is abused it may not perform well so "don't abuse them". If you want a CF to drive very low-Z high-current loads then get a CF sized for the job. When OTL engineers design an OTL output stage they make it some kind of CF. There are OTLs w/ CF output stage capable of putting a lot of power into 4 ohm loads without audible distortion. How many plate loaded stages can do that? There is very little danger of abusing a CF when it is sized for the intended loads. To abuse most CFs you have to intentionally set out to abuse them.
Some point out that CFs are asymmetrical. This just means, again, we should size the CF for the job. All SE stages are asymmetrical. But there too, if the SE stage is sized for the job, with enough idle current, it's not a problem. If you want a CF which must source and sink a lot of current without consuming much idle current, a properly designed White CF can be a good choice.
Mediocre CFs can compromise the sound. Like anything else it depends on how the CF is designed AND how it's being used. Use a good CF optimally and you'll be rewarded. No one has heard every CF all CFs are not created equal. When someone denounces CFs they of course base that on the CFs they've heard which may not have been designed or used optimally. See Diego Nardi's point #6 regarding CFs in his article: "The MONOphono Preamplifier", Sound Practices issue 16, p.34.
I use CFs in my designs because they do certain things well. Pentode CFs are a notch above triode CFs, but most people haven't tried pentode CFs. For one thing, pentode CFs are especially good at providing mu-mode operation for the voltage-amplifying triode, as explained in the Introduction of this website and elsewhere.
Another thing CFs can do well is drive output tubes. If an appropriate CF is directly connected to the output tube it can drive the output tube into grid current thereby getting the most power and efficiency from the output stage.
Here's a theory which may explain why some CFs get judged unfavorably: You've heard of the fact that some SE amp designers use cancellation techniques in their amplifiers to optimize the sound of the amp: In a typical SE amp the output stage produces various sonic blunders and distortions. To make good sound, the rest of the amp consists of elements that produce complementary distortions & blunders. Ideally the blunders/distortions would be equal but opposite and thus cancel, leaving good sound to emerge. Consider that this cancellation may often be facilitated by the high output Z of a typical SE driver stage as it tries to drive the output stage. I think we all realize that the cancellation phenomenon does not dutifully confine itself only to SE amplifiers that need it. Cancellation effects can occur anywhere in a sound system: between amp and preamp, speaker and amp, speaker and preamp, etc. There can be additions and subtractions all over the system; synergy and anti-synergy can result. If all the net effects are positive enough, good sound comes forth. Here is how a CF could upset the applecart: If there is a sonically-desirable cancellation taking place because of a stage's high output Z and/or related weakness(es), for example, then obviously if you insert a CF with its high input Z, and its low output Z replacing the high output Z, you will change the cancellation that was pleasing to you. Even though the CF does nothing wrong it will of course be blamed for the resulting degradation in sound. This is another reason why it's important to have a pure and clean sound system, including every stage, from beginning to end - so you won't have unknown cancellations taking place. Or unknown complements. Otherwise, changing anything in your sound system will be a gamble. Isn't it true that top rated hi-fi components sound great in some systems but just so-so in other systems? It's the synergy/anti-synergy thing. In this way it can be easy for a hifi device or component to be perceived as mid-fi. This could well be the situation many CFs get put in-- they perform quite well but then get blamed for what goes wrong. Food for thought.
We are also seeing a resurgence in the popularity of coupling transformers. Coupling transformers are often used as impedance changing devices, as are CFs. To use a transformer to step down the output impedance of a circuit also means a step down of the output voltage (which is sometimes desirable). CFs do this without stepping down the voltage and there are no bandwidth limitations or the other limitations that can occur with transformers. I don't mean to knock audio transformers for I definitely like them and they have a well-deserved place in hifi. In this FilamentSide Chat I'm just showing some of the advantages of CFs.
The proof is in the listening-- many examples of excellent equipment using CFs can be given. Here I'll mention some I'm most familiar with:
My designs use a follower of some kind in the front end and some of my designs use CF output stages as well. Two examples of the latter are my push-pull CF power amp (The Grape Ox) and my push-pull OTL design. Both are successful attempts to address the inherent limitations of tube amps. Among other things they make use of innovative applications of CFs and both of these designs are free of listener fatigue, setting new standards for sound and performance.
Cy Brenneman of the Euphonos Audio Group has a new breed of SE amp and preamp. He has a CF-equipped preamp and CF-based amplifiers. His equipment is also startling validation of correctly designed and properly applied CFs.
As with most things there's a knack to getting the most out of CFs to realize their full potential and best applications. Three rules to remember are that CFs, like anything else, must be (A) designed correctly, (B) sized for the job, and (C) properly applied. If someone dislikes the CFs he has auditioned it's a good bet one or more of these rules were not followed OR something besides the CF was the real culprit.
The anti--CF camp could trot out data to support their view while the pro-CF camp could show equally impressive data to support CFs. After all is said and done the ear is the final judge. Some of the most highly regarded tube audio gear of today and yesterday use CFs-- it may well be that the majority of the most consistently rave-reviewed audio equipment use followers. Ultimately the listeners' opinions are the most important. Some of the places listeners can post their own opinions of equipment they audition are the newsgroups rec.audio.tubes, rec.audio.high-end, or the website www.audioreview.com/.