• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Anyone seen this amp topology before?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I've never seen anything like that before. Though I like the idea of using grid drivers for the finals, SEPP drivers are a bit of overkill. He also bases the design on a false premise.

It is well known that even though signal current is common to both the cathode and plate load resistor of the split load phase inverter, because of the differing incremental impedances at the cathode and plate the circuit can become unbalanced at high audio frequencies. A circuit has been devised, figure (1), to provide low impedance drive to the output tubes as well as correct the high-frequency unbalance of the phase inverter. In this circuit the voltage output is controlled through the upper (cathode follower) grids and the current in the load by the drive to the lower tubes' grids.

While it is true that the plate and cathode of the cathodyne phase splitter have different impedances, this doesn't present any problems if the stage that the cathodyne drives is not driven into grid current. If that doesn't happen, the only current path is from the plate to cathode and their associated resistors. The resulting voltages must balance, as per the KVL. (Which is why the Williamson moved the cathodyne splitter back from the finals: to make sure it didn't run into grid current.) As for the different impedances, with reasonable designs, the first turn-over doesn't occur until about 250KHz. There's a second turn-over at approximately 1.8MHz. These frequencies are too far removed from the top of the audio band to have any significant effect on sonic performance. Also, forcing a SEPP into balance by unbalancing the inputs really isn't an optimum solution either.

A conventional Williamson will work just as well without all the extras hanging off it. You could also just use an LTP phase splitter too.

Too much complication to "solve" a problem that doesn't exist.
 
While it is true that the plate and cathode of the cathodyne phase splitter have different impedances, this doesn't present any problems if the stage that the cathodyne drives is not driven into grid current. If that doesn't happen, the only current path is from the plate to cathode and their associated resistors. The resulting voltages must balance, as per the KVL.

I'll agree with the latter half of this, and STRONGLY agree with your conclusion. But with symmetrical loads, the source impedance from plate and cathode is identical.

Another guy trying to be clever and solve a non-existent problem. I didn't look at his perpetual motion stuff.
 
Re: interesting circuits

aardvarkash10 said:
as for the "philosophy" and "science" in the rest of the site... froot loops anyone?

Apparently, massive amounts of LSD. Profound, man.

From Wikipedia:

While he was thought to be quite brilliant by the many students he recruited to assist him, his addictions to hashish and LSD colored everything he wrote and conceived, and most invariably left within a few years when it became clear that despite his most sincere efforts, nothing he ever postulated could be scientifically verified. Undaunted, he recruited more as needed, invariably assisted by his willingness to share his psychedelics with the newcomers.
 
BlackUnikorn said:
So the unbalanced split load tube doesn't really suffer from distortion at audio fequencies? How do you know if the driven tube is pulling grid current, voltage swing?

Whatever distortion the cathodyne has is minimal since, like a cathode follower, it has a large amount of negative feedback. As for grid current, this flows whenever the grid is more positive than the cathode, turning on the parasitic diode formed by the control grid and cathode. This is mainly a headroom problem that Williamson solved by getting the cathodyne away from the finals.
 
I can't comment on the tube circuit, but the 3rd (last) circuit in the paper is not that far away from, say, a F5 of N.P.

But what do you think about his claim that feedback should only be applied to the same active device (and *not* in the inverting input of a diff amp) when the signal goes in the non-inv. input?

Rüdiger
 
Putting signal in one input and neg. feedback thru the other diffl. input, causes a well known common mode type distortion, due to both inputs swinging around in common mode. Any capacitance or non-linearities in the tail of the diffl. input pair causes errors at the most sensitive point. Usually the two diffl. transistors will be pretty well matched in ICs, but for discrete designs, any mismatch here also causes errors at the most sensitive point. Any errors in that diffl. pair's operation, corrupts the input signal reference.

(Yes, I know, most SS audio amps do this wrong. And tube ones with a diffl. input pair do also. All for the sake of a non-inverting amplifier, input to output. Tube ones have less excuse, since they can swap the secondary wires.)

Don
 
Ex-Moderator
Joined 2004
Whatever distortion the cathodyne has is minimal since, like a cathode follower, it has a large amount of negative feedback. As for grid current, this flows whenever the grid is more positive than the cathode, turning on the parasitic diode formed by the control grid and cathode. This is mainly a headroom problem that Williamson solved by getting the cathodyne away from the finals.
I saw that Da Palma website years ago and formed the conclusion that one reason for the bizarre buffer circuitry following the cathodyne may be because he used a step network to the grids of the EL84s. This would place a DC load on whatever is feeding the step network, almost as if the EL34s were taking constant grid current. Maybe this would upset the cathodyne if it were not for that buffer.

Putting signal in one input and neg. feedback thru the other diffl. input, causes a well known common mode type distortion, due to both inputs swinging around in common mode.
According to Max Robinson, use of a CCS in the tail of a tube-based LTP allows NFB to be returned to the second grid without problems.
 
"According to Max Robinson, use of a CCS in the tail of a tube-based LTP allows NFB to be returned to the second grid without problems."

With the lowish loop gain used in tube amps, the input diffl. pair has to work over a larger region of it's characteristic curvature (than SS amps.), so matching of the tubes will be important so they both operate with equal currents and equal transconductances.

The CCS will need to be a cascoded one to avoid variable junction capacitance causing a "transistor sound" insertion at the tail. Even with ideally matched tubes and CCS, the diffl. pair still generates a little odd order distortion at signal peaks as its composite gain drops off a little.

Resistor differencing to the same input (ie, inverting node summing) avoids this odd order problem. But input Z goes down of course.

Neg. feedback to the cathode also has some subtle distortion concerns, due to the input grid/cathode transfer function not being perfect. In SS designs, neg. feedback to an input emitter or source (so called "current feedback" amps) are known for having poor distortion characteristics. But tubes have a more gentle non-linearity characteristic.

Don
 
Whatever distortion the cathodyne has is minimal since, like a cathode follower, it has a large amount of negative feedback.

I'll 95% agree. But if one makes the total load too small and the mu is too low, it can have appreciable distortion. It's easy to get it right, though; to a good approximation, one can estimate the distortion by looking at the same tube in common cathode, with the total load in the plate circuit. That distortion is then divided by mu + 1.
 
SY said:
I'll 95% agree. But if one makes the total load too small and the mu is too low, it can have appreciable distortion. It's easy to get it right, though; to a good approximation, one can estimate the distortion by looking at the same tube in common cathode, with the total load in the plate circuit. That distortion is then divided by mu + 1.

You have the same problem with cathode followers. The CF makes a good Lo-Z source for a Hi-Z load. The more heavily you load it, the less NFB you have, hence more distortion. With a cathodyne, it shouldn't be all that difficult to ensure it looks into a Hi-Z load. That's what the Williamson does.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.