# When an Input is an Output

Editor's note: Here is another neat guest tech article from my friend and colleague, Jerry Steele, Applications Manager, ON Semiconductor. (I think these short article tips are really useful)

Recently a colleague was tasked with recommending a simple pulse width modulation circuit to a customer. A straightforward approach might implement an oscillator with a square-wave output, followed by an integrator to provide a triangle waveform for a comparator with the control input for the pulse width depicted in Figure 1. While the circuit shown would work fine, I wanted to demonstrate to my colleague how using an input as an output could save her an op amp.

Figure 1 One approach to a pulse width modulator.

You can save one of the op amps by using a classic op amp (or comparator) relaxation oscillator. The trick here is to use the signal at the capacitor of the oscillator as the source of the triangle wave. The main caveat is that for large amplitudes at the capacitor the waveform is more exponential rather than linear. Linearity improves as the swing across the capacitor is reduced, and in general if the swing at the capacitor is around an order of magnitude lower than the amplifier or comparator output it is pretty linear. The reduced level is a result of selecting values in the positive feedback that provide transition levels at the input of around one-tenth those of the output. But the main point of this blog is the concept of using what is normally an input to a circuit as an output since it provides the required waveform. The concept is shown in Figure 2.

Figure 2 Using an input as an output provides the triangle wave for the duty cycle comparator.

You can also take advantage of the technique of using an input as an output if you need a sine wave. In this case resort to the classic phase shift oscillator that is always shown using a high pass phase shift network:

Figure 3 The phase shift oscillator is shown most often with a high pass feedback network.

To get a sine wave use a low pass network. The output is still a square wave, but the input to the circuit is a pretty decent square wave. It won't be state of the art for low distortion, but it is pretty clean.

Figure 4 illustrates a schematic and simulation of a phase shift oscillator used to generate a sine wave.

Figure 4 Using a low pass network on a phase shift oscillator provides a pretty decent sine wave signal at the inverting input. Never underestimate the power of using an input as an output.

Do you find this application useful? Have you ever encountered a need like this before?

## 1 comment on “When an Input is an Output”

1. Katie O'Kew
December 27, 2017

It might puzzle those for whom this material appears to be intended

to introduce an idea that occurs frequently in both analog and digital

disciplines. The output of a logic gate is the input to a following one;

the output of some amplifier will often be the input to another circuit

which overall performs some intended function.

What was meant in the contributed examples (I think) is that a node

at which a significant state-variable of a closed-loop system appears,

can (usefully) be regarded as both input and output, a very common

situation.  For example, in a two-transistor differential LC oscillator,

it's impossible to say which of the nodes directly associated with the

tank elements is not both input and output .  Or, consider the case of

an R-S flip-flop, or a two-transistor relaxation oscillator, and so on.

Indeed, I think it is true to assert that, the existence of a node which

can be said to be simultaneously both input and output is a property

that exclusively arises in closed-loop systems .  (Please correct me, if

this is not formally correct).  Even a parallel-tuned LC resonator is a

closed-loop system.

Teaching analog design to novices is already a challenge. If they are

told that “some circuits (of an unspecified type) exhibit the odd fact

that there exists at least one node which may be regarded, equally,

as either the input or the output of a circuit”, this might well stick in

an impressionable mind, distracting him or her from the fundaments

of broader topics. I believe that a teacher (of any subject) must use

the minimum number of new, necessary topics as possible, in order

to avoid the risk that, one or two of the students may be distracted

by pondering for too long about something that sounded like a very

profound and important concept – which somehow they still did not

understand.  It happened to me, when I was being taught calculus,