Andre Jute explains what "Zero Negative Feedback" really means to the electronically-aware aficionado of ultrafi tube amplifiers

 

ZFNB: Religious imperative or manner of speaking?

Is a zero negative feedback amp even possible?

by Andre Jute

There is negative feedback in every audiophile's amp, even those that are correctly described as having "zero negative feedback" or as being ZNFB amps. "Zero Negative Feedback" is an accepted form of technospeak for an amplifier without any added loops of global feedback from the output to the input, and without any deliberately added stage or local feedback outside a set of accepted topologies which also have feedback built in. For instance, cathode followers and SRPP voltage gain stages are commonly seen in amps or preamps which are uncontroversially accepted as ZNFB, yet are clearly dependent on a degenerative action (aka Negative Feedback) for their effect. Even more common choices, such as bypassing or not bypassing the cathode resistor which generates the negative bias on the grid has an outcome in negative feedback. We have accepted those as forms of "good negative feedback" in contrast with added loop (global) negative feedback, which is anathema to the most refined audiophiles, the ultrafidelista.

This apparent imprecision is one of the most difficult things for the non-technical, who can sometimes take Zero Negative Feedback as a religious imperative, to grasp and accept emotionally. That "zero negative feedback" is merely a manner of speaking hallowed by decades of practice rather than a precise technical term seems to them small consolation for being forced to accept something they have come to consider a hateful evil.


The linguistic imprecision makes negative feedback one of the most abrasive interfaces between the audiophile's emotional passion for his music and the necessary technicalities.

Those who do not wish their delusions shattered may wish to content themselves with reading KISS 123: The Ultrafidelist View of Negative Feedback and to give the rest of this chapter a miss. There is no harm in considering your expensive triode tubes as black boxes delivering the right effect, but into whose machiavellian machinations one doesn't want to enquire too closely!



Measuring a triode's internal feedback

Putting our hand on the feedback inside a triode isn't as easy as analysing a pentode circuit and determining, "Those wires from the output to the input, and those resistors, are the negative feedback network".

Here is an experiment anyone capable of building his own tube amp such as The KISS Amp 300B can set up which will make visible the degenerative force of internal triode negative feedback at work.

An experiment to isolate and measure triode internal feedback

Set up a plate-follower triode gain stage.

Apply and increase the grid voltage. Note that the plate current increases. And note the effect, that the plate voltage drops. So we have output change somehow related to input change. However, we have two interdependent effects from the change in grid voltage. That the *net* effect of this latter interaction is invisible inside the tube is a cause of confusion for some who still claim that there is no negative feedback inside a triode. This experiment will make constituent parts of the net effect calculable and prove that the visible product is less than the total of the internal reactions.

Now hold the voltage to the plate steady while you increase the grid voltage. Observe that the current increase at the plate is greater for any increment of grid voltage than in the case where two variables (plate voltage and current) were permitted freely to interact from a single change at the other terminal (grid voltage).

The difference in measureable current at the plate between the two parts of the experiment, which I hereby christen The Missing Current or the Jute Triode Effect, is clearly caused by a degenerative force. It is what we normally call negative feedback.

QED.

How a pentode is created from a triode: the multigrid anology

The degenerative force aka negative feedback can be eliminated from the triode by making the plate current independent of the plate voltage, as we have seen above. One way of doing this (if you have a big lab with lots of facilities) is by adding one more grid between the control (or input) grid and the plate, and putting a high steady voltage on it. A third grid is required for the protection of the second grid but it does not do anything else and is irrelevant here, except in the naming of this new tube: pentode.

This rationale works both ways. I might as easily argue that removing the third and second grids from a pentode releases the interdependence of plate voltage and plate current consequent upon varying input grid voltage, and thereby causes current increase to be less than expected, a clear sign of a degenerative interaction, also known as negative feedback.

Put the screens back into the triode to make it into a pentode and observe how much higher its gain is, how much higher its output impedance is, and how much its linearity is degraded by comparison with the original triode. That is the difference caused by the removal of the native negative feedback inherent in the pentode's original triode state.

A light dawns

This rational development explains why triode tube amps generally sound better than multigrid tube amps: the multigrid amp has to be kludged right by adding something the triode naturally possesses. It is like the difference between a woman born beautiful and one stuffed with silicone by a plastic surgeon.

Why are so many classical writers quiet on this subject?
I can understand why non-technical ultrafidelista who have read this far may refuse to believe in something they cannot see: they have a large emotional investment in their faith in zero negative feedback amps. But there are quite a few technically qualified people who claim that internal negative feedback in a triode does not exist because if it did the classical texts would have mentioned it. That's an argument only a fool will wear. The Radiotron Designer's Handbook, the indispensible RDH, doesn't mention SRPP either, an absolutely indispensible audiophile topology. Examples of that nature can be multiplied almost indefinitely.


I don't believe that the classical writers on electronics were unaware of the negative feedback within a triode, merely that it was of little importance for them, usually because of the time when they wrote: the RDH was heavily revised for its fourth edition when pentodes had already replaced triodes in quality commercial amps and were indeed themselves under threat from transistors; the other key text, the MIT Research Laboratories Series (which in Volume 18 does have the SRPP though it doesn't call it that) was written in and immediately after the greatest war the world had ever seen, with knowledge codified under immense pressure to be immediately useful. Finally, it has often been claimed that Professor Child referred to the degeneration inside the triode and that he meant negative feedback.

Today none of the reputable modern thinkers doubt that there is negative feedback inside a triode. One of the notably productive deep thinkers of audio electronics, John Broskie, on his TubeCAD netsite gives this pointed analysis of the negative feedback inside a triode:

"The triode's plate, very much like a resistor, will resist an increase in plate voltage by increasing its current conduction and will resist a decrease in plate voltage by decreasing its current conduction; thus, the plate's opposition to voltage variations defines a negative feedback mechanism. In other words, the triode already enjoys a short, fast feedback loop from its plate to its cathode."

John Broskie: www.tubecad.com/march2001/page2


So, is a ZNFB amplifier is possible?

Yes, a ZNFB amplifier is possible. Just build it altogether of pentodes and add no feedback devices of any kind. But no one in his right mind does it because it is guaranteed to sound like nothing any audiophile would want to listen to. The strict technical reality is that no desirable or even practical high-fidelity amplifier can be built without some manner of negative feedback. In triodes (or trioded multigrids) the NFB merely acts internally, invisibly, without additional component networks to the outside of the tube.

That also explains why it is tradional to call any amp, regardless of its native NFB, "Zero Negative Feedback" as long as there are no added loops of (generally) global feedback.


THE VOLTAGES IN THIS AMP WILL KILL YOU.

GET EXPERIENCED SUPERVISION IF IT IS YOUR FIRST TUBE AMP


All text and illustration is Copyright © Andre Jute 2006

and may not be reproduced except in the thread KISS xxx on rec.audio.tubes