Editor's note: Our guest author this month is Loren Siebert, applications engineer, Texas Instruments
Operational amplifiers (op amps) have proliferated from specialized instruments made from large tubes to tiny monolithic circuits composed of a sliver of silicon, germanium or gallium arsenide. Dozens of manufacturers produce hundreds of op amps with a dizzying array of features and specifications. When selecting an op amp, one of the first specifications many engineers look for is bandwidth. Manufacturers know this and feather device bandwidth prominently on the data sheets. It is rare to find an op amp data sheet without a bandwidth number on the front page. Because small signal bandwidth is the largest number, this is usually the number featured most prominently. A good question, though, is how important is this number and how does it relate to other device performance metrics? This article attempts to answer that question.
A typical voltage feedback op amp bandwidth versus gain plot is shown in Figure 1 . It is clear that gain plays a huge part in small signal bandwidth. In fact, the relationship is well-established such that for most gains, bandwidth (BW) multiplied by gain (G) is a constant. Amplifier A, shown in Figure 1 , has a gain bandwidth constant of approximately 280 MHz. The glaring exception to this is the unity gain, or G=1 case. For a gain of 1 the small signal bandwidth is 800 MHz. Figure 2 shows much the same thing, except that amplifier B has a gain bandwidth product of approximately 450 MHz. Since these amplifiers have similar headline small signal bandwidth numbers. Let us see how closely they perform on other measures.
As shown in Figures 3 – 4 , we can see that amplifier B has much better distortion performance. Continuing on to Table 1 we can compare a wide variety of performance metrics. While at first glance it appears that amplifier B is clearly superior to amplifier A, the overall picture is not so clear. Amplifier B is better on gain bandwidth product as well as distortion and output current. Amplifier A, however, has lower noise voltage and consumes significantly less power.
It seems clear from the information so far that small signal unity gain bandwidth is not a very useful predictor of amplifier performance. Would gain bandwidth product be better? With amplifiers A and B the gain bandwidth product does seem to correlate fairly well to the amplifier performance with regards to distortion or linearity. However, this relationship is not guaranteed either because amplifiers C and D have the opposite relationship between gain bandwidth product and linearity. In other words, there are no easy shortcuts to determining amplifier performance. If distortion is important to your design, make sure to verify the distortion specifications in the product data sheet.
Large signal bandwidth does not correlate with small signal bandwidth, nor does it correlate with distortion performance. It does, however, correlate highly with slew rate. In fact, this is one case where two parameters are closely linked. The ability of an amplifier to create a large amplitude signal is dependent on the amplifier’s slew rate capability.
Like large signal bandwidth, other parameters such as input noise and supply current do not correlate with small signal bandwidth. In fact, you could see why someone looking for a low-power, low-noise amplifier might actually prefer amplifier A over amplifier B. Between amplifier C and amplifier D the choice is even easier. With the exception of a very slight noise difference, amplifier C is clearly superior.
While a small signal bandwidth is often the leading specification for amplifier marketing, it does not seem quite as useful when actually selecting an amplifier for a particular application. When I look for an amplifier to recommend to customers, I do look at the small signal bandwidth number – but I very quickly look past that for other key specifications such as distortion, slew rate, input noise and supply current and supply voltage.
Here’s more information about TI amplifiers.