Using opamps for amplification and filtering

Many uses of sensors for computer input require both amplification of the sensor output and filtering the waveform to eliminate noise.

We worked with filters last week. This week we'll consider using operational amplifiers, alone and in conjunction with filters.

We have some lm1458 opamps in the lab, and today we will use these in inverting and non-inverting amplifiers, and as a low-pass Sallen-Key filter.

The inverting amplifier looks like this:

and the non-inverting amplifier looks like this:

Neither of these diagrams shows the power supply for the op-amp. Use the +15V and -15V terminals on the power supply, but adjust the outputs down to +12V and -12V.

Build an inverting amplifier with a gain of 10. (Use moderately large resistors.) Then test it by using the frequency generator from the power supply board. Display both the input and output traces on the oscilloscope for a sine wave of 1KHz; if necessary adjust the amplitude of the frequency generator so that the amplified signal is not clipped off at the top and bottom.. (Since the maximum voltage the op-amp can produce is the power supply voltage, in this case 12V, if the input signal is greater than 1.2V, there will be a flat spot on the top of the amplified signal.)

Note the relative phase of the waveforms.

Build a non-inverting amplifier with a gain of 11 (same resistors as for the inverting amplifier should work. Make observations similar to those for the inverting amplifier.

Sallen Key low-pass filter

A common Second order low-pass filter design is this one:

As you can see, the design includes a low-pass filter and a non-inverting amplifier, as well as a more confusing positive feedback. As you might expect, computing the values of the components for R₁, R₂, C₁, C₂ is a job for a specialized program.

One such program suggests R₁ = 3.8K, R₂ = 3.9K, C₁= 22nF, C₂ = 150nF for a corner frequency of 1K. (R₃ and R₄ affect the gain, but not the filtering.)

So build the filter with the indicated component values, and try it on frequencies of 10Hz, 100Hz, 1000Hz, 10000Hz, and graph the result. (If you use more frequencies you may get a smoother graph, but we're already breaking the fundamental law of bad science: If you have have enough kinds of graph paper, you can place any three points in a straight line. Therefore, never collect more than three data points.)