CMOS vs. bipolar op amps

Which type is best for my application?

BY KEVIN TRETTER
Microchip Technology
Chandler, AZ
http://www.microchip.com

Today’s system designer has many choices when selecting operational amplifiers (op amps). The three largest op-amp manufacturers collectively have over 1,600 products from which to choose, and that doesn’t include specialty amplifiers!

How do you sort through this overwhelming number of devices? One way is to start by selecting the optimal process technology. Most manufacturers clearly label an op amp as CMOS, bipolar, or even BiCMOS, but what does this mean with regard to the actual application?

Process technologies

Before digging into actual op-amp specifications, it is worthwhile to take a brief look at each of these process technologies. Bipolar transistors, invented in the late 1940s at Bell Laboratories, are so named because they use both electrons and holes in their operation. These transistors were used extensively in the design of ICs for decades, and are still in use today.

Complementary metal-oxide semiconductors (CMOS) came in the 1960s and became popular as lower-power digital-logic alternatives to transistor-transistor logic (TTL). Unlike other transistor types, CMOS transistors only draw current when switching states — hence, their association with low power.

As the name implies, BiCMOS combines both bipolar and CMOS technologies on the same device. This combination allows the IC designer to take advantage of the inherent benefits of each technology. Until the last couple of decades, combining bipolar and CMOS into one seamless process was not very reliable, and was very costly. Today, BiCMOS is used for certain applications (including op amps) where the inherent benefits outweigh the additional costs associated with this process.

In the end, each of these process technologies has advantages and disadvantages when it comes to op-amp design. The following sections will take a closer look at these tradeoffs and how they relate to actual amplifier specifications.

Fig. 1. CMOS op amps, like the MCP6041 from Microchip Technology, provide low quiescent current for low-power applications.

Power consumption

As mentioned above, CMOS is known for lower power, as the transistors only draw current when switching states. However, this power advantage is only true for slower amplifiers.

As bandwidth increases, a CMOS amplifier’s current increases dramatically, and soon draws more current than a comparable bipolar amplifier. Because of the exponentially increasing current required for CMOS to achieve high speeds, bipolars are typically better suited for high-bandwidth, high-slewing applications.

For lower-bandwidth applications, CMOS amplifiers can still provide power advantages. For example, the MCP6041 low-power CMOS op amp from Microchip Technology has a typical quiescent current of only 600 nA and provides a gain-bandwidth-product of 14 kHz.

Noise performance

In terms of flicker or 1/f noise, CMOS transistors have worse low-frequency noise than bipolar transistors. At low frequencies, this noise is dominated by irregularities in the conduction path and noise due to the bias currents within the transistors. In a bipolar transistor, the conduction path is buried down inside the silicon.

On a CMOS transistor, the current flow is near the surface, making it susceptible to defects in the surface of the silicon, which increases the low-frequency noise. At higher frequencies, 1/f noise is negligible as the white noise from other sources begins to dominate.

CMOS transistors have a lower transconductance, relative to similarly sized bipolar transistors, which results in higher broadband noise. In general, bipolar op amps hold an inherent advantage over CMOS when it comes to noise performance.

Voltage offset

Another important amplifier specification is input offset voltage. As the name implies, this specification is the amplifier’s voltage difference between the inverting and non-inverting inputs. This error voltage can vary from microvolts up to millivolts, and is highly dependent on how well matched the input transistors are.

Bipolar transistors inherently offer better matching, resulting in lower offset voltages for a given architecture. Some manufacturers compensate for this inherent mismatch by using laser trimming, fuses, or even EPROM.

These techniques can improve an amplifier’s performance significantly, regardless of the process technology. Better matching also results in less voltage-offset drift over temperature, which is also an important consideration in many applications.

Price/packaging

The reputation CMOS has as a more cost-effective technology is mainly due to traditionally lower wafer costs, driven by the high volume of CMOS logic chips. Despite the lower wafer costs, for a given current capability, CMOS transistors occupy more silicon area than bipolar transistors, resulting in a larger silicon dice. So, even though the wafer costs may be lower, there are fewer die per wafer, thus negating the cost benefit. In the end, the cost structure of these two process technologies is very similar.

A larger silicon solution also limits a manufacturer’s packaging options. This can be a significant limitation, as system designers are constantly tasked with placing more performance and functionality into smaller and smaller form factors. Several packages, such as ball grid arrays and leadless types, address this situation.

Input bias current

All amplifiers have a specification called input bias current, the amount of current flow into the inputs of an amplifier to bias the input transistors. This current can be thought of as leakage current, but is referred to as bias current when on the inputs of an amplifier. This bias current can range from picoamperes to hundreds of nanoamperes.

Op amps with a CMOS input stage generally have less bias current than those with bipolar input transistors, typically around 1 pA, while bipolar transistors can be orders of magnitude higher. This bias current is converted into a voltage through the circuitry’s input resistance, and ends up resulting in an error voltage at the output of the amplifier. The less bias current the better, and in this regard CMOS has a distinct advantage.

Which process is best for amplifiers?

Which process is best for amplifiers has been debated in the past and is expected to be a point of discussion for years to come. Bipolar amplifiers are grounded in history, but CMOS amplifiers offer some inherent advantages. BiCMOS processes are relative newcomers to the field, but this hybrid technology takes the best of both worlds and provides superior performance at a price point that is becoming more and more competitive.

So in the end, the answer to the question is “it depends,” which is why this topic continues to be debated. System designers must evaluate the function of the amplifier in their system and determine which specifications are most critical.

If the op amp interfaces with a high-impedance sensor, such as a thermocouple with some passive filtering, then keeping bias currents to a minimum will be important. In this case, an amplifier with a CMOS input stage is the best choice. On the other hand, if the application requires a high-speed, high-slewing amplifier, then a bipolar amplifier may offer the best performance at the lowest quiescent current.

No universal amplifier or process technology addresses all the applications in which op amps are found. This is why manufacturers continue to provide a multitude of op amps on a variety of process technologies. ■

For more on op amps, visit http://www2.electronicproducts.com/AnalogMixICs.aspx.