schoenecker

FFT Analysis: And Now For Something Completely Different

Blog Post created by schoenecker Employee on Jun 6, 2017

Oscilloscope users are constantly reviewing the signals of their design in the “normal” time vs voltage display of the scope.  It is easy to overlook the FFT (Fast Fourier Analysis) view of the same signal. It is a completely different way of reviewing the signal characteristics that often reveals clues to some very difficult to solve problems. FFT can be an invaluable tool for identifying noise, crosstalk, and other common problem in many designs that can stall prototype development. In digital designs, it is often used to highlight and pinpoint the source by the frequency content on power rails.

FFT measurement with an input sine wave

By applying the FFT algorithms to the sampled data, you convert the time domain operation of the oscilloscope into a frequency view of the signal. This results in two primary benefits. One, you can easily identify each of the frequency components. Two, it reveals the magnitude of each contributing signal.

Identifying the frequency

By identifying the frequency components, it reveals if there are any signals on that are not expected. For example, a digital signal should only have frequencies that are harmonics of the base signal. If you have a 10 MHz data, there should be only frequencies at 10 MHz (the primary harmonic), 30 MHz (the third harmonic), 50 MHz (the fifth harmonic), and the continuing odd harmonics up to the bandwidth of the source.

Any other frequency is a result of noise, or crosstalk, or some type of coupling on to the signal.

Frequency components of 10MHz clock

Figure 1: In this capture of a 10 MHz clock, we can easily identify the frequency components related to the fundamental frequency, but we can also see a 20 MHz signal that is -55 dB from the fundamental.

 

It’s important to understand how the oscilloscope sampling characteristics play into the quality of this FFT measurement. The oscilloscope analog bandwidth, sample rate, memory depth, and related time capture period all can have a profound effect on the measurement result. The math that is utilized for the calculation is using the data that was sampled at 5 GSa/sec, and it makes it possible to calculate a 10 GHz FFT. However, the front end of this scope is 1 GHz, so the FFT is only valid up to the bandwidth of the oscilloscope.

Identifying the Magnitude

The other key component of the signal is the power of each signal component. When looking at a signal in the time domain, it is only possible to see the very large signal power components. In the spectrum view (or FFT display) the horizontal axis is changed from a linear voltage scale to a logarithmic voltage scale (or dB for decibel).

The display on the right side of the display is listing the power level in dBV (decibel volts, or power relative to 1volt) of each frequency in order with the respective power level.  The first, or fundamental frequency of our signal is at just less than 10 MHz, and a power level of -13.9775 dBV, which is about 200mV rms. Looking at the time display of the signal (in green), you can see that it looks about right. We can also see that the next highest power signal is at 30 MHz and a power level of -30dBV, or about 3 mVrms-- something that cannot be seen in the time display that we are used to looking at.

FFT is just a button away

On Keysight oscilloscopes, the FFT operation is often enabled by simply pressing a button on the front panel. The new 1000X low cost oscilloscopes include this feature standard. The FFT view is a great way to examine a signal to find the frequency and power that you could not normally see any other way. Make sure to take advantage of this powerful tool that next time you are trying to find elusive signals in your design.

Learn other time-saving tips to get more out of your oscilloscope with this new eBook!

Outcomes