The article on op-amp basics treated op-amps as ideal devices. While most modern op-amps do act as nearly ideal circuit building blocks, they are not perfect. Many or the errors can be attributed to op-amp itself while other are due to the design of the circuit the op-amp is used in. When selecting an op-amp for a specific application it pays to read the data sheet and select a part that best fits the intended use.
While selecting the right op-amp for the job is important, for instance if you are interested in low noise you should buy a low noise op-amp, there are things you can do in your circuit design to minimize errors. Three common problems areas, offset voltage, noise, and instabilities or oscillations, are addressed here.
An op-amp is a differential amplifier. Ideally, when both the inverting and non-inverting inputs are at the same voltage, the output of the op-amp will be 0V. This is not the case in a real world op-amp. You can think of the input offset voltage as a small internal battery (so to speak) that is in series with one of the inputs. This is shown in the following figure.
When configured as a buffer, the input offset voltage shows up as an output offset voltage of the same magnitude because the buffers gain is 1. In an amplifying configuration, the input offset voltage is amplified as well making for a larger output offset voltage.
Depending on the circuit and the particular op-amp specifications, this may or may not be an issue. For instance, in an AC coupled circuit, this can frequently be ignored. In lower precision DC coupled circuits, the offset voltage may likewise be acceptable. If this is an issue, there are several approaches to dealing with the issue.
One simple solution is to purchase an op-amp with an offset voltage that meets your needs. If this is insufficient, or the cost of the op-amp is a critical issue, some op-amps are offered with inputs to which a pot can be connected for the purpose of zeroing out the offset voltage. If the op-amp of interest doesn’t have inputs specifically for this, an external potentiometer that is part of a voltage divider summed into the op-amp can be added to the design that accomplishes the same effect.
A potentially more prevalent source of voltage offset errors is due to bias currents. Bias current is conceptually similar to input offset voltage involving current flow as opposed to a voltage offset. The following figure illustrates the point.
Op-amps have a small of current flowing into, or out of, their inputs (see the op-amps data sheets for the expected bias current range). The current is very small (typically fractions of a microamp) but can have a large effect on your circuit. Leakage currents behave something like a constant current source. They are minimally affected by the resistance or voltage of the circuit connected to them. Fortunately, for reasons that will be discussed soon, they are frequently fairly well balanced between inputs.
The bias current does not result in any direct output offset errors. However, offset voltages will occur due to voltages that appears across resistors in the op-amps feedback loop due to the current flow. Fortunately this can be easily addressed by following a couple of simple design practices. In critical circuits it is best to both choose a low leakage current op-amp and follow these practices.
The choice of resistor values in an op-amps feedback network will effect the offset voltage. Consider an inverting amplifier with a gain of two made from a 100KΩ and 200KΩ resistor. These are not unusually large values but they will have a noticeable effect on the output offset even with a relatively small leakage current value of 1 microamp. To understand why this is we need only rely on Ohm’s law.
The current that leaves (or enters) the inverting input passes through R1 and R2. It has to, there is no other place for it to go. The parallel combination of these values is about 67KΩ. Because of the bias current flow a voltage will appear at the inverting input (-). In this case it is:
67KΩ * 1 microamp = 67 mV
The circuit has a gain of two therefore twice this value, 133 mV, will appear at the output. If the circuit had a gain of 10, this bias leakage current would cause an output of about 0.67V. For all but the most crude circuits, this can be a problem. Note that there is a bias current flowing from the non-inverting input as well. It is not shown because it has no effect because there is no resistance in that leg.
There are several solutions to this problem. The use of a low leakage current op-amp seems an obvious choice but, depending on other requirements this may not be practical solution. Reducing the resistance values is another choice. Using low value resistors have other consequences however. The op-amp may not tolerate the additional output current required with small resistors well and using low valued resistors increases current flow limiting battery life as well as decreasing the circuits input resistance. A third option is to add a resistor to the circuit as shown below.
Placing a resistor between the non-inverting input and ground greatly reduces the output offset voltage. Doing this causes the same offset voltage to appear on both inputs of the op-amp negating the effect. In this case, a 66.5 KΩ resistor was used because it is the closest standard value to the 66.67 KΩ value resulting from the parallel resistance of R1 and R2.
Since an op-amp amplifies the difference between its inputs, and the inputs now have the same voltage due to leakage present at their inputs, there is no resultant output. This is not a perfect scheme as the bias currents are not perfectly balanced or the resistances perfectly balanced but the issue will be largely eliminated.
The same technique can be used in a non-inverting amplifier configuration as well. In this case the additional resistor is added to the non-inverting input. As is the case with the inverting amplifier configuration just discussed, its value should be the parallel combination of the feedback resistors.
One of the concerns encountered when amplifying a signal is the amount of noise present at the circuits output. Output noise can occur for a number of reasons. One potential source is the op-amp itself. If the noise level is important, selecting a low noise amplifier is important. That said, even if the op-amp were perfect, the output of your circuit will contain some noise and the source is somewhat surprising. It is the resistors in the fee back loop that are the source of the noise.
The noise is due to the molecular motion of the resistive material. The type of material the resistor is made out of can effect the amount of noise generated. Metal film, metal foil, and wire wound resistors produce the least amount of noise but all resistors suffer from this issue. Fortunately there are things you can do to minimize the problem.
The amount of noise produced by a resistor is proportional not only to the ambient temperature but also to the resistance value. For instance, at room temperature over a 10 KHz bandwidth, a 20 KΩ resistor will produce about 1.8µV of noise while a 2 KΩresistor will produce about 0.6 µV. The simplest solution is to use lower value resistors. There is a limit to how low you can go of course based on the output current rating of the op-amp and the amount of power you want to draw from your supply but, in general, if noise is a critical concern use a low noise op-amp and low valued resistors.
Op-amps generally don’t tolerate capacitive loads well. While some op-amps are specified for capacitive loads, most aren’t and can oscillate or become erratic. This can be a problem particularly when your circuit is intended to drive equipment that you do not have detailed specifications for, or the load your driving is at the end of a long cable (cables have capacitance).
The op-amp become unstable or oscillates with a capacitive load because the capacitor disturbs the feedback loop. Remember that an op-amp does have internal resistance (its the feedback makes it appear as if it doesn’t). The capacitor at the output essentially causes a phase shift in the feedback circuit. You can visualize this by picturing a simply RC filter at the output with the feedback taken at the capacitor. Depending on the frequencies present in the signal you are amplifying, this phase shift can cause the feedback to be out of phase with the input (one is going up while the other down) which causes oscillations and instability.
The solution to this malady is simply. Add a resistor to the output of the op-amp. Even a relatively small valued resistor (something around 50Ω) will often correct the issue. This works because the op-amps feedback circuit is now primarily resistive and the feedback stays in phase with the input eliminating the instability.