01 November 2008

October has passed without a single blog posting from me. Sad. But it was a tough month. First, my son attends a year-round school and his track had the entire month of October off. As much as I love my son, have you ever noticed how hard it is to get anything done when you have virtually unlimited "help" from a five-year-old? To complicate matters, a week into October, my copy of MS Outlook gave up the ghost. I have since learned that Outlook simply stops working, without warning, when the PST database file exceeds 2 gig—which mine did last month. (Amazingly enough, I had started fresh PST file this January and, in less than a year, I hit the 2G limit. Amazing, just amazing. I am now, quite sensibly, a Thunderbird user.) Some French spammer is spoofing my domain name, so I got about a bazillion "Notification d'état de la distribution" messages from around the world. And then our fourteen-year-old cat also died. Extremely sad. Meanwhile, my computer grew ever more senile (its complete end approaches quickly, probably due to failing capacitors on the motherboard). I have bought a replacement computer, but it must be outfitted with all my special programs and backed-up files. Also during October, I ran out of just about all the PCBs and many parts, so I had to order more. This led to my revising several of the old PCBs (which in turn entailed redoing the user guides for the revised boards) and my creating a few new PCBs. And, finally, there is the election, when we Americans enjoy, as Ambrose Bierce said, the sacred privilege of voting for the man of another man’s choice. Of course, I and America were not alone in enduring a rough October, as the world has seen trillions of dollars melt away and with it the promise of economic perpetual motion.

On a much happier note, almost all the Aikido PCBs are back in stock (although I did not order that many boards to be made, as I haven’t figured out if mine is a counter-cyclical business or not). And I have a few new, truly cool boards to offer the DIY tube-loving community.

 

Janus Regulator PCB, Rev A
The revised Janus regulator PCB features a one inch increase in height and now all four high-voltage electrolytic capacitors hold bypass capacitors.

Moreover, the PCB now offers more rectifier options, as it can run a single tube rectifier (5Y3/5R4/5AR4) or hybrid setup which uses two solid-state rectifiers with the tube rectifier or four solid-state rectifiers for an entirely solid-state rectification. In addition, a bigger series resistor (5W to 12W instead of 3W) can be housed and the heater regulator can now fit a bigger heatsink. Below are the four possible rectifier configurations:

Above, we see the two setups that use a tube rectifier; below, the two all-solid-state versions.

The new Janus regulator PCB is shown below.

Because the new PCB is larger and now uses two more large, high-voltage film capacitors, the price went up tad, $12 USD. To purchase, please visit the GlassWare Yahoo Store at http://store.yahoo.com/glass-ware

 

UltraPath Push-Pull
Blog number 147 gave an overview and critique of the Ultrapath circuit. Here is a quick recap: the Ultrapath circuit is a small variation on the grounded-cathode amplifier, differing in the B+ termination of the cathode bypass capacitor, rather than the usual ground termination. Depending on the rest of the circuit, this may or may not prove to be a good idea.

               

All the examples from blog 147 were single-ended, but shifting the cathode bypass capacitor's termination to the B+ connection can also be performed in a push-pull amplifier, as we can see below.

Well, is such an amplifier worth trying? Sure it looks a bit silly, but would prove noisier than the conventional push-pull amplifier, as the output transformer’s primary would see a common-mode, power-supply noise signal across its primary's leads. Of course, a perfect null is only posible with perfectly-matched output tubes and a strict class-A mode of operation, for if one output tube cuts off, the power-supply noise cannot cancel in the output transformer.

Obviously, the output stage does not exist in a vacuum, as it must receive its input signal from the amplifier’s front-end circuitry. Now, let’s imagine a simple differential amplifier serving as the gain stage and the phase splitter at once. With a differential amplifier, if the common cathode resistor’s value is large or if a constant-current source is used, then the differential amplifier’s output signal will effectively be “Ultrapathed,” as all the power-supply noise will be superimposed on the output signal because of the differential amplifier’s supremely poor PSRR. A calamity? Not so, if the output stage holds an Ultrapathed output stage. The differential amplifier and the Ultrapathed output stage effectively use the B+ connection as the “ground” reference.

The signal cascades from the circuit above to the one below.

Note that adding an RC power-supply filter to the input stage would incease the noise delivered to the speaker.

So is this the ultimate answer to push-pull amplifier design? No, but it is a clever way to make a fairly good and cheap push-pull power amplifier. Ideally, the B+ connection should be, in AC terms, indistinguishable from ground and we would never have to worry about having to switch from ground reference to B+ reference. But then if this were an ideal world, last October would have been much easier on the soul.

 

Ultra White Cathode Follower
Not all tube-based push-pull amplifiers hold an output transformer. The SRPP and White cathode follower need no center-tapped output transformer to work. The White cathode follower, which I have covered many times before, holds some promising results when its cathode resistor’s bypass capacitor is terminated into ground, but not in the way you might envisage. Put differently: if the White cathode follower is used with positive power supply rail, then having the cathode resistor’s bypass capacitor terminate into the B+ connection is a bad idea, as the bypass capacitor will relay the power-supply noise into the output signal. On the other hand, if a negative power supply rail is used with the White cathode follower, the bypass capacitor’s upward termination into ground is a great idea.

Note how the input, the output, and the bypass capacitor’s termination are referenced to ground. Also note how the two cathode resistors in series work to smooth away the negative power supply rail’s noise at the bottommost cathode. Now, just for conceptual kicks, let’s look at the same White cathode follower with a positive power supply.

Note that the same cathode-to-plate voltages and idle current obtain.

 

Aikido White Cathode Follower
About a decade ago, I set about trying to apply an Aikido-like elimination of power supply from a standard White cathode follower. The result was the following circuit.

Aikido White Cathode Follower

Capacitors C1 and C2 inject the correct amount of power-supply noise to achieve a power-supply-noise null at the output. Using 6DJ8/6922/ECC88 tubes, the ratio was 8 to 1 in favor of capacitor C1 over C2, for example 100µF for capacitor C2 and 800µF for capacitor C2. Why? I could not mathematically determine why that ratio proved effective, so I have not mentioned it here. No doubt, once the formula is found, it will seem embarrassingly obvious.

Next Time
I have quite a bit more to say about the new PCBs and a little bit more to say about Ultrapath and push-pull amplifiers.

// JRB