Method Analysis of Operational Amplifier Output Drive Capability (I)

When selecting an operational amplifier (op amp) to implement a specific function in a circuit, one of the most challenging selection criteria is the output current or load driving capability. Most of the performance parameters of an op amp are usually clearly stated in the data sheet, performance graphs or application guide. Designers must consider the output current and other parameters of the op amp to meet the product performance specified in the data sheet. The output current varies greatly between devices provided by different semiconductor manufacturers, and even between different devices provided by the same manufacturer, which makes the design and application of op amps more complicated. This article will explain through some examples how to evaluate the driving capability of the circuit to be designed based on the performance parameters of the op amp, so as to help designers ensure that the product they choose has sufficient load driving capability in all cases.

　　What factors affect driving ability?

　　Output drive capability is a function of a number of internal and external set values ​​or conditions. The bias current, driver level, structure and process of the output stage are all internal factors. Once a device is selected to implement a specific function, the designer cannot change these internal conditions that affect the output drive capability. Most low-power op amps have poor output drive capability, one of the reasons is that their output stage bias current is small. On the other hand, high-speed op amps usually have higher drive capability to meet the low resistance requirements of high-speed circuits. High-speed op amps usually have higher power supply operating current, which also improves the output drive capability.

　　Traditionally, integrated PNP stages have had poorer performance than NPN transistors. In such processes, the lower beta values ​​of PNP output transistors compared to NPNs mean that the output drive capability will be unbalanced. Full-rail output op amps usually use the collector of the transistor as the output pin, and the poor performance of PNP transistors will result in poorer ability to provide source current than sink current. For non-full-rail devices, the situation is exactly the opposite. Since most devices use the emitter output of the PNP transistor, which greatly affects the sink current characteristics, their ability to output sink current is poor. Moreover, when estimating the output current capability of a device, the performance variation between devices should also be taken into account. Therefore, while designers select devices based on the "typical" data sheet specifications, they must also consider the "limit" and "minimum" specifications to ensure that each device used has sufficient drive capability when it is produced.

　　In addition to the internal factors listed above, some external factors also affect the drive capability. Some of them can be controlled to optimize the output drive capability, while others are difficult to control. The following are the external factors that affect the output drive capability: output voltage margin relative to the corresponding power supply voltage (difference relative to the power supply voltage); input overdrive voltage; total power supply voltage; DC and AC coupled loads; junction temperature.

　　Output drive capability is usually given in terms of output short-circuit current. Here, the manufacturer specifies the current that can be supplied when the output is grounded (or half the supply voltage in the case of single-supply operation, called "Vs/2"). The manufacturer may provide two values, one for source current (usually preceded by a "+") and one for sink current (usually preceded by a "-"). In applications where the voltage swing at the load is small, the output stage driver will have a large voltage difference with respect to the supply voltage (source current is V+, sink current is V-), and the user can use this data to effectively predict the performance of the op amp. Imagine an op amp with a large load and the load is driven by a voltage close to ground (or Vs/2 in the case of a single supply). If the load of the amplifier stage changes stepwise, the current that can be supplied to the load will be consistent with the current value given in the op amp data sheet as "output short-circuit current". Once the output begins to change accordingly, two things will happen: the op amp's output voltage margin will decrease, and the op amp's input overdrive voltage will decrease.

　　The output current that can be provided will be reduced due to the former reason, which is also related to the design of the op amp. As described in the latter, a reduction in overdrive voltage will also cause a reduction in output current.

　　Another, more useful way to determine current capability is to use a graph of output current versus output voltage. Figure 1 shows a graph of output current versus output voltage for the LMH6642 from National Semiconductor. For most devices, there is usually one graph for both source current (Figure 1a) and sink current (Figure 1b).

　　Figure 1: Output characteristics of the LMH6642.

　　Using this type of graph, it is possible to estimate the current that an op amp can deliver for a given output swing. These graphs are provided by semiconductor manufacturers to show the relationship between the output current capability of an amplifier and the output voltage.

　　Note that in Figure 1, "Vout from V+" is plotted against the output source current, and "Vout from V-" is plotted against the output sink current. One reason for presenting the data this way is that it can be more easily applied to single-supply or dual-supply operation than the output voltage relative to ground. Another reason is that voltage margins have a much greater impact on output current than total supply voltage, so for any supply voltage, even if the exact conditions are not found in the data sheet, this data sheet presentation allows the designer to make a rough calculation using a set of closest curves.

　　Figure 1 can be used to predict the voltage swing for a given load. If the axes are linear, the designer only needs to add a load curve to the characteristic curve of Figure 1, and the voltage swing can be determined by the intersection of the two curves. However, as shown in the figure, in many cases, especially when the op amp is full swing output, both axes use logarithmic coordinates so that the curve can have better resolution when the output current is small and the output is only a few millivolts. Under logarithmic coordinates, the load curve is no longer a simple straight line and will not be easy to draw. So how can we predict the output swing for a given load?

　　If the designer is willing to take the time to iterate the swing prediction between the device performance and the external circuit requirements, a very accurate result can be obtained. Here, I will use some examples to illustrate how to make such predictions.

　　Figure 2: Example of predicting the output voltage swing at a given load.

　　Consider an application such as Figure 2a, where the LMH6642 is used to drive a load with RL = 100Ω connected to Vs/2 (1/2 supply voltage). Assume that the output of the LMH6642 is biased at Vs/2 or 5V in this case:

　　The question is, can a designer use the data for the LMH6642 shown in Figure 1 to estimate the maximum possible output swing? The answer is yes.

　　To estimate the swing, a table is created (Table 1) that starts with an initial guess for the output swing (column 2) followed by a series of revisions to the guess (compare columns 3 and 5, with the results shown in column 6).

　　Table 1: Using iteration to predict the output swing of Figure 2a (LMH6642).

　　Repeating this process until the device characteristics match the load requirements under the given conditions, the final result is obtained at the bottom of the second column, thus completing the estimation of the swing. Therefore, the repeated results in Table 1 show that the circuit in Figure 2a can produce a maximum voltage of 8.75V on a 100Ω load. Converted to a peak-to-peak value is 7.5VPP {= (8.75-5)V x 2 = 7.5VPP}.

　　Here are some notes about the approach used in Table 1: For the circuit in Figure 2a, only sourcing current is available. Therefore, only Figure 1a is used. In each case, the values ​​in column 5 are calculated in Figure 1 assuming the worst-case temperature condition. The values ​​in column 5 are read from the graph in Figure 1a using column 4 as the y-axis. The final result in column 2, the value for the fourth iteration, is still an approximate solution because the value in column 3 (87.5mA) is still lower than that in column 5 (90mA). However, the resolution of the graph no longer allows fine-tuning of this result.

　　Now let's change the example just discussed slightly and assume that the output load of the LMH6642 remains unchanged, but the signal is AC coupled, as shown in Figure 2b. The method for predicting the output swing is the same as before, except that some entries in the table (column 3) need to be modified because the AC-coupled load only "sees" the swing of the signal, and the DC component (bias) of the output voltage is blocked by the AC coupling capacitor. In addition, it is also important to note that the AC-coupled load requires the output of the LMH6642 to be able to both accept and source current (unlike the application in Figure 2a, which only requires the output to source current). Therefore, the smaller value of the source current and sink current characteristics is selected and filled in column 5 of Table 2.

　　Table 2: Output swing predicted using iterations for Figure 2b.

　　The final result in column 2 (9.6V) corresponds to an output swing of 9.2VPP {=(9.6-5)V*2=9.2Vpp} on an AC-coupled load, which is, as expected, larger than the value (7.5VPP) in the DC-coupled load example discussed earlier because there is no DC load.

　　The process of estimating swing using these optional output capability plots is very similar to the previous examples, using an iterative approach to fine-tune the initial guess.