Noise Lab Details

Noise Analysis Lab (pdf) Note, the lab will consist of filling out the noise table on page 2 of the lab. One thing not listed is that you will need to find the transfer function vout/vni which will be of the form 1 + Zf/Zi, where Zf is the feedback impedance (capacitor parallel to resistor) and Zi is the impedance to the input (a resistor).

In doing the noise integral, you may separate the two terms, e.g., the constant "1" and the "Zf/Zi" part. This isn't quite correct, but results in an easy analysis and only a small error. When calculating the total integrated noise due to vni, use the unity gain bandwidth of the opamp as the upper frequency limit.

Now some definitions:

Signal Size Largest possible signal at the ouput of the opamp, often simply related to the power supply rails. Signal size is calculated in volts_rms, or squared(volts_rms) so it can be compared to noise integrated rms voltage or noise in integrated volts squared.
Dynamic range - is largest signal (in this case 0.5-2.5V) to smallest possible signal just detectable over the noise level. For example, if the smallest possible signal is at the noise level then dynamic range is equal to SNR. In this assignment, assume SNR=Dynamic Range.
SNR - The ratio of the largest possible signal size to integrated noise level. Often expressed in dB, so if signal and noise are both in terms of power, then it is 10 x log_10 of the above ratio.
Noise Floor - Integrated noise over the bandwidth, in this case the sum of three noise sources, one for the opamp and one for each of the two resistors.

Question So why do we have to integrate the noise? Can't we see a much smaller signal if we use a spectrum analyzer with a narrow resolution bandwidth?

Answer What you have done here is change the problem by defining a much narrower bandwidth. In this case, the bandwidth is the resolution bandwidth of the spectrum analyzer, and the noise here is automatically integrated over this bandwidth (the Spectum Analyzer does this). In real life, the bandwidth will typically be set by the bandwidth of the IF or baseband filter filter in a communications system which is selected according to the bandwidth of the modulated signal. At the output of this filter may be a detector which tells us if there are bits coming through the system. This detector will see the signal and the total integrated noise, and does not have the option to narrow the bandwidth to see the signal better. Even if it could do this, it would be chopping off part of the modulated signal. We may test the system with a single tone (sine wave) which could allow us to narrow the bandwidth (i.e. use a spectrum analyzer) but to give a realistic number we then integrate the noise over the desired bandwidth.

Question Why don't we include other sources of noise such as power supply noise and input noise?

Answer We are trying to find out about the circuit itself to see how much noise it adds. If we know the gain of the circuit from input to output, it is straightforward to calculate how much noise there is at the output due to the input. As well, if we know the gain from the power supply (often in the form of a power supply rejection ratio) we can calculate the noise at the output due to the power supply.