• Perry

Tube Amplifiers Explained, Part 11: Negative Feedback

Updated: Apr 18, 2020

Part of a blog series Tube Amplifier Circuits Explained

If we believe an excess of distortion is not ideal, then the real question remaining to ask is: how can we limit it?

There is a technique we will review in a moment, but first I’ll just mention that the component selection and design of the circuit is the most important starting point. The AE1 kit uses a 12AT7 driver tube and EL34 output tube, and is designed with a certain operating point, power supply and filtering, load lines, output transformer, etc. All of these choices result in a certain performance and level of distortion. There are many types of tubes, each with different characteristics for amplification factor, transconductance, grid curves. There are also many other circuit design choices more complex than this one, possibly involving multiple tubes or stages before the output. A push-pull is also another design, different from single-ended and with different implications for power and distortion. All that to say, we start with components and a circuit designed as best we can for our intended objectives, and then we can consider one more tool in our toolbelt to optimize it: negative feedback.

As with many other things, there are varying points of view about the use of negative feedback. Some may view this as an undesirable way to improve the performance of the amplifier, somehow compromising on sound quality or purity of design. Certainly if the circuit design and performance is poor, and negative feedback is used to try and put lipstick on a pig, then I can understand. But to categorically view negative feedback as something to avoid is, in my opinion, missing a very beneficial technique.

So what is negative feedback and how does it work? I’ll try to explain it in the way that makes sense to me, acknowledging there are others who can cover the theory and equations much better if you want to learn more.

Feedback as a general concept can be found in all sorts of places where the actual result is compared to a reference level. Consider a few examples:

  • A thermostat detects the room is too cold and turns on the furnace until the temperature meets an expected level and then the thermostat turns it off.

  • Your car’s cruise control measures how fast the car is going and modulates the gas to the engine if the car is going too slow or fast relative to a certain level.

  • Your toilet has a float that detects the water level and opens or closes a water valve until the tank is filled to the expected point.

In the amplification process we have looked at so far, we have an input voltage and an output voltage that is larger based on the gain of the amplifier. This can be called “open loop gain” (meaning there is no feedback loop) and we know this gain is not perfectly uniform for all input voltages. It is sort of "out of control", so to speak.

A negative feedback loop can be used to change the circuit and create a different “closed loop gain” that takes a portion of the output voltage—the “feedback fraction”—and subtracts it from the input to create a new control voltage. A general feedback structure is shown here.

Taking a portion of the output and subtracting it from the input will attenuate the control voltage, lowering overall gain. This is a small sacrifice we make to use negative feedback, and presumes we have a substantial amount of gain to begin with. We have now tied the input to the output at a certain relationship, and the actual gain of the amplified signal is now determined entirely by this feedback fraction, instead of the open loop gain on its own. In order to achieve a state of equilibrium, the control voltage will compensate higher or lower depending on the relationship the feedback loop sees between output and input. This is exactly what we need to deal with non-linearity and is why it will reduce harmonic distortion.

Before we look at our actual circuit again, I want to address one more mental trap that is easy to fall into. It is easy to trace a circuit at our slow and methodic human pace: “Ok, our input signal is coming in here… [finger pointing on schematic] …and then the tube amplifies the signal and it comes out over here… and then the feedback loop sends it down here where it comes back over to the input… and then, um, it goes through the tube again a second time? And over and over?” This is not the right way to think about it. There is not an iterative process happening repeatedly over time, it is an almost instantaneous influence on the control voltage when we add in the feedback loop[1]. If there were an imbalance based on the feedback, the voltages would compensate faster than any audio frequency that we care about.

Negative feedback can be implemented in various ways: local feedback around one tube, global feedback around an entire circuit, etc. Let’s take another look at our AE1 schematic and see how it is using negative feedback. There is one resistor we have not discussed yet, the 200k Ohm resistor that comes off the anode of the output tube and connects back to the other side of the coupling capacitor. This is one type of feedback, essentially connecting the two tube stages plate to plate. The anode voltage of the EL34 is the output we are taking a fraction of (using the 200k resistor) and connecting back and subtracting from the input that is coming from the driver stage.

How do we know this is negative feedback, subtracting instead of adding? If you look back at a previous post showing the load line, a positive change in the grid causes a negative change in the anode voltage. This means the output voltage of the anode is inverted relative to the input grid. So in our feedback example here, the output is already of inverse polarity compared to the input and connecting them will subtract the output.

I know I’ve only barely described this, and poorly! Although we could look at other types of feedback or use some equations, I think I’m going to stop here and not try to go further than this general overview. But I hope you get the general concept: using the output as an influence on the input to achieve a controlled gain level. If you haven’t gotten it just yet but want to know more, this could be a topic you choose to study more using additional resources.

We have so far covered a lot about the amplification part of the circuit. Now let's discuss the power supply and then I think we can wrap up this series!

[1] This subject is obviously more complicated than I’m describing here, and time can play a factor in various ways, but the intent is to explain feedback for very basic understanding, not a detailed technical description.

1,485 views1 comment