Post 247                                                                     21 October 2012

At a much lower price point, the KEF LS50 won my heart and ears. This little speaker can dance. Pick your five favorite audio adjectives and they probably will apply to the LS50 speakers, great imaging, articulate, quick and agile—even big bass. I was convinced that a subwoofer hid behind curtains, but the LS50s provided the only bass. Amazing. The adjective that I found myself using was "fun." These are fantastically fun speakers.

Being little high-tech wonders, the LS50s are filled with innovations, such as soft ports, advanced materials, and a coincident drivers (the tweeter sits at the center of the woofer). They only cost $1500 the pair; and if they could fit in my Christmas stocking, there is nothing that I would prefer finding inside.

A Good Omen, Perhaps?
After attending a funeral or an audio show, it is easy to be pessimistic. I noted, however, something that might count as an encouraging sign of better times to come. Babes. Sexy women in big ads. Before you turn me into the federal insensitivity police, hear me out. In David McCullough's book, 1776, he points out how the British army welcomed news of rapes by its soldiers as a sign of the army's vitality, although they shot the guilty soldiers. Well, I was surprised to see several big posters displaying audio gear and babes. Think about it, the odd thing is that the dog didn't bark, that the makers of audio gear, knowing that 99.99% of their customers are male, do not adorn all their ads in sexy women. Maybe Eric Barbour is right and most audiophiles are not all that manly? Car nuts certainly suffer no compunction at emblazoning their ads with babes.

Maybe this harkens a return of animal spirits in audio.

High-Gain Tube Circuits
Some audio applications require gain, lots of it. Phono preamps and microphone preamplifiers immediately come to mind, but there are other needs for high gain. For example, if the circuit employs a negative feedback loop, it is the unused gain that powers the feedback. In other words, the more open-loop gain, the more feedback, and the lower the distortion and output impedance. Other uses might a be a hot linestage. I have long argued that line stages usually offer too much voltage gain, that most power amplifiers require only 1Vpk to be brought to full power. But there are setups that require much more than +20dB (times 10) of gain to achieve the 1Vpk of output. For example, a phono preamp that bellowed with a high-output moving magnet cartridge, might only whisper with a moving-coil pickup. In this case, a line stage with a gain of +30dB or, even, +40dB might be needed. In addition, certain hybrid power amplifiers might require a high-gain, tube based input stage, because the amplifier can swing big voltages into the speakers.

The Aikido topology delivers a gain equal to about half of the mu of the input tube used. For example, a 12AX7 input tube presents a mu of 100, so a gain of x50 (+34dB) can be realized. While this seems like plenty, more gain may be needed. We could bypass the bottom cathode resistor in the input stage, but this will only increase the gain to about 60 to 70.

(Note the 20k & 10k power-supply-noise voltage-divider resistors, which are just guess-estimates on my part.)

If we need more gain, something like the following can yield close to the mu of the triode used in the first stage. Yes, it does look a lot like an SRPP frontend, but as the first stage doesn't really encounter any load, the circuit cannot push and pull.

Do not forget that the external load is a critical part of the SRPP circuit. In this circuit, the top triode in the first stage functions as something approximating a constant-current source. Okay, why not just use the SRPP stage and forgo the modified cathode follower? Prejudice, prejudice on my part. I always strive to unburden the gain stage entirely, so that its only job is that of developing voltage gain. The SRPP stage performs two tasks: it provides gain and drives the external load. Nonetheless, the maximum gain realizable from an SRPP stage is always less than the mu of the bottom triode used. So, how do we get a gain higher than the triode's mu?

One approach is to cascade two grounded-cathode amplifiers.

The above circuit realizes a gain of about 320 (+50dB), which is plenty—even scary—high. The 17.5µF capacitor was added to achieve one of my signature Aikido-noise-reduction techniques. Of course, some tweaking would be required; if for no other reason that electrolytic capacitors are usually way off their marked values.

Aikido Pentode
Another approach is to use a pentode-based gain stage. One of the reasons that pentodes were invented was to get more gain. One of my favorite pentodes is the EF86. The following circuit is an Aikido pentode gain stage. Like its triode-based cousin, the input stage's only job is creating gain and its output stage's jobs are to provide a low output impedance and to scrub away the power-supply noise that input stage leaks profusely. In the case of the pentode input tube, close to 100% of the ripple will appear at its plate, its output.

The Aikido cathode follower (ACF) receives 100% of the power-supply noise at its bottom triode's grid. The result is that noise nulls at the ACF's output. Okay, great, can we stop here? Well, that depends... Note the number of quality signal capacitors within the circuit: three. If you are the type of audio DIYer that buys $140 boutique capacitors, then the following circuit might be more to your wallet's liking.

We have the same number of tubes, but two fewer capacitors. The pentode's 2nd grid (screen grid) is driven by a 12AU7-based cathode follower that injects an optimal amount of power-supply noise, which will achieve a power-supply noise null at the pentode's plate. The second 12AU7-based cathode follower drives the external load. Sweet. Clever. But do not think that we got something for nothing, for we must pay in two ways. The first is that this two-cathode-follower Aikido version will draw more current; second, unlike the first Aikido-pentode circuit, this circuit is tweaky, which means that we must fiddle with the resistor values. (Many audiophiles would sooner fetch moon rocks than figure out new resistor values.) A workaround is to add a potentiometer, which would allow an easy noise null, even with aging pentodes or pentodes from different makers.

Aikido Cascode
Another approach is to use the cascode topology, which was created to achieve many of the same goals as a pentode circuit, such as low-input capacitance and high gain, but without the pentode's higher noise.

The following is from the Tube CAD Circuit Guide PDF:

“CASCODE” is a contraction of “CASCaded triodes with the gain of a pentode and the low noise of a triODE.” It is built up of one triode in current series with another triode. The bottom triode’s cathode is grounded and its grid is fed the input signal, while its plate is connected to the cathode of the top tube, whose grid is grounded. This loading of the plate with a cathode prevents the bottom tube’s plate voltage from moving very much in response to signal at its grid, as the top tube cathode functions like a voltage regulator made from a cathode follower. Nonetheless, the bottom tube will experience variations in the current flowing from its cathode to its plate, because of variations in the grid to cathode voltage. The same would also hold true if the triode were connected across a regulated power supply, as a varying grid to cathode voltage would also define a varying plate current in this arrangement. In the cascode circuit, this varying current through the bottom tube must also flow through the top tube, as they are in series. And as the top tube’s plate is loaded with a plate resistor, the varying current defines a varying voltage across its plate resistor, which yields the output voltage and gain of this circuit, which can be considerable. Thus, “the gain of a pentode.”

The circuit, furthermore, functions much like a pentode in that the Miller Effect capacitance is very low, as the grid of the bottom tube is shielded from the inverted amplification at the top tube’s plate. Also like a pentode based circuit, the distortion from this circuit is higher than would have been the case with only one triode. This is because the bottom triode does not experience any of the linearizing effects of running the triode in a closer approximation of a constant current mode, as in the case of a grounded cathode amplifier with a large plate resistor. It differs, however, from the pentode by the fact that there is no grid current flowing into the top triode’s grid as there is with the grid 2 of a pentode and by the fact that the noise is lower than a pentode’s. Thus, “the low noise of a triode.”

The secret behind this circuit is that it is an elaborate attempt to conserve as much of the triode’s transconductance as possible. In a normal grounded cathode amplifier, the plate load resistance subtracts from the effective transconductance of the triode. The plate load resistance added to the plate resistance (rp) and then divided into the mu of the triode determines the transconductance of the triode.

The key word throughout was "transconductance." The cascode circuit does not rely upon the triode's amplification factor (mu or µ), but its transconductance (gm) for gain. The higher the transconductance, the higher the gain.

The problem with the cascode circuit, however, is that it, like the pentode circuit, offers a dismal PSRR. Well, if we can undo the pentode's poor PSRR with an Aikido technique, can we do the same with the cascode? Sure. In fact, the very first issue of the Tube CAD Journal (1999) held an article I wrote on Improving the Cascode's PSRR. Here is the original and ugly GIF schematic.

Since my first rule is let the gain stage develop gain without having to drive any difficult or uncertain load, the following Aikido Cascode circuit will also undo the cascode's poor PSRR and drive external loads with a low output impedance.

The cascode input stage develops a high gain and the ACF output stage scrubs away the power-supply noise.

Next Time
I will have more information on the Aikido Cascode and the new PCBs and, of course, new circuits.

//JRB