08 May 2007

Turmoil

It seems that I have caused turmoil in tube land; the angry towns people are taking up pitchforks and torches, and soon they will break down my door. What was my offense? Surely it would have to be something spectacular, like describing a D-getter as a halo or confusing the 12AV7 for the 5965 or by gratuitously quoting Shakespeare. No, much worse than any of these, my sin was not being sarcastic enough.

I began my last blog entry with

“I am a big fan of the common-cathode amplifier (also known as the “cathode-coupled” or “grounded-grid” amplifier—thanks Bruce).”

The “—thanks Bruce” portion was my subtle way of rapping Bruce Rozenblit on the knuckles for misnaming the cathode-coupled amplifier the “grounded-grid amplifier.” I should have written, “—no thanks to Bruce Rozenblit, whose taxonomic howler has made already murky waters murkier still.” Because I didn’t write that last fragment, I am seen as thanking Bruce for a job well done. I am not. In fact, when I reviewed his book, Audio Reality: Myths Debunked...Truths Revealed, I wrote that he had misnamed his line stage amplifier:

Grounded Grid Preamp
This circuit is a truly minimalist line stage amplifier. It derives its name from the fact that the triode that provides the gain for the circuit receives its input from its cathode being coupled to a cathode follower rather than its grid. Still, I prefer to use the name "common cathode amplifier" as that is how the circuit is described in many of my beloved electronic books from the 40's and 50's. (Two articles on this circuit topology and it modification are available at the GlassWare site.) Furthermore, "grounded grid amplifier" is usually used to describe an amplifier with only one triode that receives its signal directly at its cathode and hence has both a very low input impedance and low input capacitance.

(By the way, I received some criticism for my having endorsed his book so highly. A few readers who bought his book on my recommendation told me that they were disappointed that his book wasn’t more like the Tube CAD Journal. Well, it isn’t, but that doesn’t mean it isn’t a good starting point from many beginners. In other words, I still stand by my recommendation; besides, in a world where equipment stands and power cords can cost $5,000, at $29 it’s a bargain. Yes, I do not think his OTL-zener-trick patent is worth any of the commotion that his fans announce so loudly, but I do think that the rest of his circuits are otherwise solid, if not a bit stodgy—in a good way, having had to repair many uncultivated and too-imaginative circuits for friends, I prefer stodgy. And I truly enjoyed reading his engineering philosophy, particularly the part where he writes about talking to two Afro-American audiophiles at a high-end show.)

So, are we done? No. The topic of circuit designation is dispiriting. The SRPP goes by four or five names; and most single-tube topologies have at least two names—even the cathode follower is also known as the “grounded-plate amplifier.” I wish it wasn’t so. But can you ever put a genie back into a bottle? On NPR radio shows, NPR for God’s sake, I hear the word “enormity” used to describe huge things, even huge good things like lotto winnings or number of adoring fans, but never moral outrages—for monstrous offenses the word “enormous” is often used. Just today, I read of the “fulsome” praise heaped on Queen Elisabeth; did the reporter mean “offensively flattering” or just copious or, possibly, pompous?

The sad reality, speaking of audio reality, is that I get at least one e-mail a week where the reader calls the cathode-coupled amplifier a “grounded-grid” amplifier. And it easy to see why he does: Bruce has named his line stage amplifier such and it is popular, besides the reader is quite unlikely to have read many 60-yearold electrical engineer textbooks (very few share my fetish) and on the DIY sites he reads that is what the circuit is called.

Do not get me wrong; I wish it were otherwise. In fact, here is the irony of the situation, in these pages, I have decried sloppy nomenclature—the procedure of assigning names to the kinds and groups listed in a taxonomic classification. For example, from the Cars, Planes, Circlotron article:

Nor will gurus restrict their terminology’s meanings to those definitions found in a dictionary; instead, the word “drive” will mean voltage or current or gain or power or whatever they want it to mean at the time. Of course, when a word can mean anything, it means nothing (something the universal pantheist never figured out).

And here is what I wrote on Pawels circuit:

Once again the poor naming convention or lack of a naming convention arises. The actual Circlotron circuit used more tricks than two power supplies. A better name for this sort of topology might be “Figure Eight,” as the current circle has been twisted or folded on itself, like a rubber band made into a figure eight. Or, perhaps, “Horizontal” would be a better name, as it would find a compliment in “Vertical,” what we now call “Totem Pole.”

And on hybrid topologies:

"A second regret I have is that I did not provide a name for each of the six common output topologies, as names would prove handy. But I cannot think of a naming system that would cleanly and definitively label each topology. And the usual audio practice of fluff names, such as "mega-ultra-supra-linear" sickens me."

And from the Partial Feedback Amplifiers article.

Like so many paradoxes, this one crumbles when precision of speech is brought to bear on it. “Impedance” is similar to “resistance,” but not identical to it. That the two words are often used interchangeably is due to sloppy thinking and that erosion of meaning that all words must bear from use. (“Transpired” once meant more than just “happened” and “nice” once wasn’t nice and, while are at it, “Class-A” once meant something other than “Class-AB”. When Confucius was asked what he would wish for if given only one wish, replied “I would restore the original meanings to words.)”

Speaking of “nice,” it once meant wanton, profligate, as in Shakespeare’s Anthony and Cleopatra:

MARK ANTONY

I will be treble-sinew'd, hearted, breathed,
And fight maliciously: for when mine hours
Were nice and lucky, men did ransom lives
Of me for jests; but now I'll set my teeth,
And send to darkness all that stop me. Come,
Let's have one other gaudy night: call to me
All my sad captains; fill our bowls once more;
Let's mock the midnight bell.

(In all honesty, I expected some flack for quoting Shakespeare in the last blog, as dead white guys go, William Shakespeare is about as dead and white as they come. And quoting him or any other dead white guy is minor sin these days, wherein we bow to the multicultural imperative, the view that the all the world’s cultures merit equal respect and scholarly interest, except the Western culture.)

Six years ago, I even offered my own cure to the circuit-naming problem:

In past issues, I have bemoaned the lack of a intelligent taxonomy of the basic tube circuits (and I have decried the general naming practice of tube circuits, i.e. no system). I proposed that the triode, MOSFET, FET, and transistor elemental circuits should be organized and named according to the following order: input, reference, output. Thus the grounded-cathode amplifier would be described as a grid-cathode-plate amplifier or GKP for short; a grounded-source amplifier, as a gate-source-drain amplifier or GSD for short. The advantage to such a system is that, like the periodic table of elements, the holes are readily apparent, which prompts us to fill in the blanks.

Cathode-coupled amplifiers
Posting a blog entry without a schematic would be almost unthinkable, so let’s have some more fun with this topology. The last variation on this circuit looked like this:

Let's ignore the active load in place of a plate resistor and let's deconstruct the circuit.

Three simple, single-tube topologies make up the more complex circuit. (I have used the same triode in each and have used typical voltage values.) The first is a simple cathode follower that receives the input signal; next, the grounded-grid amplifier amplifies the signal, without inverting its phase; then the cathode follower buffers the output. Note how many of the voltages match and since they do, we can eliminate four coupling capacitors, which brings us back to the complex circuit.

Now, what if we didn't need any gain, but we did want the widest bandwidth possible. The great feature of a cathode follower is its low-input capacitance. This makes sense, as the grid-to-plate capacitance is small and the larger grid-to-cathode capacitance is mitigated by the cathode following the grid in phase. Well, what if the plate also followed the grid? Such an arrangement would effectively nullify the grid-to-plate capacitance. The following circuit shows how this can be realized.

100% of the output signal is fed back to the grounded-grid amplifier’s grid (remember the deconstructing, I am referring to the rightmost bottom triode’s grid), which will force the amplifier to produce a near-unity-gain output. this arrangement will also supercharge the output cathode follower’s performance, as its distortion and output impedance will be greatly reduced by the 100% feedback ratio. The 100% feedback also causes the coupled cathodes to more closely track the input signal (had the right triode’s grid been grounded, the cathode will only move 50% of the swing that the input grid sees), which greatly reduces the effective grid-to-cathode capacitance, which, in turn, greatly extends the high-frequency bandwidth.

The input cathode follower is now augmented in that its plate sees the same signal as its input grid, as the leftmost top cathode follower follows the output signal. All in all, not a bad circuit, but of course there is always room for improvement.

Now the circuit is fleshed out, with the missing parts restored to their rightful places. For example, grid-stopper resistors—those unsung and under appreciated circuit savers, abound. The Tube CAD Journal signature protective diodes stand ready to keep huge hundred-volt voltage differentials from developing between grid and cathodes at start up, when the tubes are cold and not conducting (just imagine the grid being 400 volts more positive than the cathode. And, finally, the Aikido-like noise reduction trick of bypassing the bottommost shared cathode resistor, which injects just enough power supply noise-induced current to flow through the 20k plate resistor to counter the power supply noise at the output.

So is this circuit a modified cathode follower or a hotrodded cathode-coupled amplifier? Or, was Bruce right after all and the circuit is a super-mega grounded-grid amplifier? Or, how about, using my own system, the GPK-KGP-GPK amplifier? To be frank, I like the “TCJ cathode-coupled follower” best, although I must admit that GPK-KGP-GPK amplifier does have a ring to it.

A less complex, but equally interesting, circuit is shown below.

The inductors work as constant current sources, but with out displacing much voltage, assuming a DCR of 100 ohms. The diode can be removed, as it only serves to make up for the voltage drop across the top choke. Note the low B+ voltage. Note the paucity of parts. Imagine a chassis with two tubes and four chokes. If only we didn't have to deal with the coupling capacitor:

But then, who say that we have to put up with it. (Now, what if we lost the 1M grid and the 300-ohm grid-stopper resistors, replacing the first with a choke and the second with a ferrite bead...a resistor and capacitor free line stage amplifier—in marketing terms, that has got to be worth at least $5,000. Or, am I not being sarcastic enough ;}

//JRB