Understanding the Rail-to-Rail Characteristics of Op Amps

灞波儿奔

Understanding the Rail-to-Rail Characteristics of Op Amps [Copy link]

Follower circuit:

The sampling voltage VI_AMP_IN on the front-stage sampling resistor is sent to the ADC for A/D conversion through the follower action VI_AMP_OUT of U6.
The role of U6 here is to reduce the "load effect" and improve the acquisition accuracy.
D3 and D4 are the input protection diodes of the op amp. When the input abnormal voltage is higher than the power supply voltage by VF (diode forward conduction voltage drop) or lower than the ground potential by VF, the diode will be turned on and clamped.

1 Main features of LMV831
First, the maximum input error voltage VOS of the op amp is 1mV, which is conducive to improving the overall accuracy; Second, due to the use of CMOS technology, the input bias current is as low as 0.1pA, so there is no need to spend extra effort on eliminating the bias voltage; Third, the output drive current reaches 30mA, which is very suitable for use with ADC; Fourth, the EMIRR of the op amp is as high as 120dB at a frequency of 1.8GHz. This feature is conducive to resisting the interference of the RF module on the board; Fifth, rail-to-rail output is very important under single power supply conditions.

2 Output characteristics

From the table above, we can see that the heavier the load, the worse the rail-to-rail characteristics of the op amp output. However, because the op amp in this case is connected to a low-speed ADC, the load is very light, so we can take 6mV (VOH) and 5mV (VOL) in the table as typical values.
Under single-power supply conditions, the load resistor RL will be connected to V+/2, which actually uses V+/2 as a virtual ground.

3 Simulation verification
The simulation circuit is established as follows:

As shown in Figure 1-2, LMV831 is built into a common-mode amplifier with an amplification factor of 2. At the same time, a triangle wave with an amplitude of 5V and a frequency of 10Hz is input (in order to saturate the output). The simulation results are shown in Figure 1-3.
Obviously, the output amplitude is very close to the supply voltage of LMV831, 4.5V. The measured amplitude is 4.49V (the platform part of the trapezoidal waveform), and the lower end of the waveform is also close to 0V, which confirms the rail-to-rail output characteristics of the op amp.

4 Rail-to-rail, there are still details to pay attention to:

The acceptable voltage range of ADC is 0~4.096V, and now the follower built by LMV831 can support 0~4.49V output, it seems that everything is ready. If the pre-stage sampling voltage is also in the range of 0~4.096V (that is, the op amp input voltage), the whole circuit is perfect! However, my intuition tells me that things are definitely not that simple. I suddenly remembered that when I was selecting the model, there was a Rail-to-Rail option in TI's op amp screening conditions:

The options from left to right are: input rail-to-rail, output rail-to-rail, input to positive rail, input to negative rail - wait, does LMV831 support rail-to-rail input? I was looking forward to it, but unfortunately, the LMV831 data sheet did not mention whether the input is also rail-to-rail. Further research found that when the op amp is powered by 3.3V, the common-mode input range is -0.1V~2.1V! That is to say, when powered by 3.3V, LMV831 does not support rail-to-rail input!

5 Input characteristics
The common-mode input range of the op amp is closely related to the supply voltage. The higher the voltage, the larger the input range. In order to verify the highest undistorted input voltage under the supply voltage of 4.5V, the simulation circuit shown in Figure 1-4 was built.

Perform "parameter simulation" on the circuit and test the output voltage under the supply voltage of 3.3V, 3.9V, and 4.5V, as shown in Figure 1-5. The triangle wave is the input waveform, and the three waveforms similar to isosceles trapezoids are the output of the op amp. Among them, the dark yellow is the output when the supply voltage is 4.5V, the green corresponds to the supply voltage of 3.9V, and the purple corresponds to the supply voltage of 3.3V. It is obvious that: first, the LMV831 is not rail-to-rail input; second, the common-mode input range of the op amp increases with the increase of the supply voltage. Under the supply voltage of 4.5V, the maximum output voltage of the follower (gain is 1) is about 3.39V, that is, the maximum input voltage is 3.39V. In short, under the supply voltage of 4.5V, the maximum common-mode input voltage (without distortion) of LMV831 is 3.39V.

After learning the truth, on the one hand, the supply voltage of the op amp was increased from 3.3V to 4.5V to increase the output range, and on the other hand, the sampling resistor value was reduced so that the maximum sampling voltage was less than 3.39V, thus avoiding the risk of board modification!

6 Practical verification
In the actual product, a 91:34 resistor was used to divide the input 0~10V
, and the divided voltage was then sent to the ADC for A/D conversion through a voltage follower. The resistor sampling voltage, op amp input voltage, op amp output voltage, and A/D conversion voltage were measured respectively, and Figure 1-6 was drawn.

Pay attention to the data in Table 1-2. As the op amp input voltage slowly approaches the "threshold", the transfer error increases sharply. When the input voltage is 3.9954V, the op amp saturates and outputs 4.5460V! This phenomenon can be clearly observed by drawing the chart Figure 1-6, thus confirming the conjecture. In fact, my purpose has been achieved, because I only need the linearity to meet 0.2% FS when the external input is 0~12V.