Many people occasionally use op amps as comparators. In general, this works when all you need is a simple comparator, and you have a “redundant” op amp in a quad op amp package. The phase compensation required for stable op amp operation means that the op amp will be very slow when used as a comparator, but if speed is not critical, then an op amp can suffice. We are occasionally asked about this use of our op amps. This approach sometimes works, and sometimes it doesn’t work as well as one might expect. Why does this happen?

Many op amps have voltage clamps between the inputs, most of which are typically implemented using back-to-back diodes (sometimes two or more diodes in series). These diodes protect the input transistor from reverse breakdown at its base junction. Many IC process breakdowns occur with a differential input of about 6V, which can greatly alter or damage the transistors. The figure below shows the NPN input stage, D1 and D2 provide this protection.

In most common op amp applications, the input voltage is around zero volts, which simply cannot turn on these diodes. But obviously, for the operation of the comparator, this protection becomes a problem. The differential voltage range (about 0.7V) is limited before one input drags the other (pulling its voltage in a nasty way). Still, we can use op amps as comparators. However, we must be careful in doing so. In some circuits, this practice may be completely unacceptable.

The problem is that we (including other op amp manufacturers) don’t always account for the existence of these clamps. Even if it is stated, we may not give a detailed explanation or elaboration. Maybe we should say, “Be careful when using it as a comparator!” Datasheet writers also usually just assume you’re definitely going to use an op amp as an op amp. We recently had a meeting in the Tucson, Arizona product division. It was decided that we will make this situation clearer in the future. But what about the op amps that are already in production? The following guidelines may help you:

In general, bipolar NPN transistor op amps have input clamps, such as: OP07, OPA227, and OPA277. An exception is the uA741, which has NPN input transistors and some additional series lateral PNPs that provide inherent protection for the NPN.

General-purpose op amps using lateral PNP input transistors generally do not have input clamps, for example: LM324″ target=”_blank” title=”Link to Product Folder”>LM324, LM358″ target=”_blank” title=”Link to Product Folder” >LM358, OPA234″ target=”_blank” title=”Link to Product Folder”>OPA234, OPA2251″ target=”_blank” title=”Link to Product Folder”>OPA2251 and OPA244″ target=”_blank” title=”Link to Product Folder”>OPA244. These op amps are generally of the “single-supply” type, which means they have a common-mode range that extends to the negative supply (or slightly lower). A negative input bias current indicates that the input Bias current flows from the input pins. At this point, we can usually identify them as this type of op amp. However, it should be noted that high-speed op amps that use PNP inputs generally have input clamps, and these PNPs are some with more Vertical PNP with low breakdown voltage.

JFET and CMOS amplifiers operating at higher voltages (generally greater than 20V) may or may not be clamped. This uncertainty requires you to do more careful inspection. The characteristics of the process and transistor type used determine whether or not there is a clamp inside.

Most low voltage CMOS op amps do not have a clamp. Auto-zero or chopper types are a special case, which may have clamp-like behavior.

Bottom line is… if you’re considering using an op amp as a comparator, be careful. Read the product manual carefully, do not miss a bit of information, including some notes in the application section. Verify its performance on a breadboard or prototype to see the effect of one input voltage on another. Do not rely on SPICE macromodels. Some macromodels may not include some additional components to model the clamp. Also, there may be other phenomena that can arise when you fumblingly move the op amp from one rail to another, which we may not be able to model accurately. If you have any questions, please visit the Texas Instruments online technical support community at www.deyisupport.com.

Does differential clamping affect standard op amp circuits? The input voltage to the op amp should be about zero, right? We will discuss this in detail next time.