The applet simulates the operation of a generic digital FFT receiver at a chosen clock frequency (1024-MHz default), followed by a complex (based on I/Q sampling), or real, fast digital Fourier transform of digitized amplitude data. Complex or real processing is selected using the C R radio buttons at the bottom of the applet.
The model operates on both signal and noise, has many data windowing options, and copes with a range of DFT lengths and practical circuit limitations. Several common windowing functions are available from the Window Function drop-down menu for comparison.
The detection (SNR) performance is realized for a number of signal duration, rise-time, and offset conditions in both fundamental and sub-Nyquist band modes.
For sub-Nyquist operation, the RF noise bandwidth increases in 1.024-GHz band steps ( for 1024 MHz clock default) tracking the signal frequency.
The upper display represents the real signal input component and demonstrates the chosen signal position and duration on the set window scale, together with the selected window shape.
The lower display shows the power in each DFT output bin/channel.
Two independent signals are available; both may be programmed with different parameters and are selected for adjustment using the Modify Signal radio buttons at the bottom of the applet. The two signals can be operated either singly or overlapped, with adjustable amplitudes and selected from fixed frequency, FMOP or PMOP signal types.
Clicking and dragging the mouse in the region of the displays indicates cursor values with respect to the mouse x-axis position. In addition, for the lower display, the graph values in the indicated cursor strobe range are processed to indicate the average power (green), rms power (magenta), and peak spike power (blue) to enable output SNR to be judged. The cursor strobe range is modified with the Strobe scrollbar at the bottom of the applet.
The DC Offset scrollbar is used to null the channel offsets arising due to the system vector noise components. Optimum setting occurs when channel noise levels are minimized across all bins.
FFT lengths between 32 and 32,000 and ADC quantisation between 1 bit and 16 bits can be examined.
The Scale scrollbar magnifies the x-axis about the maximum signal position.

User Notes:
To achieve maximum sensitivity in a real receiver the front-end gain should be sufficient for noise to exceed the ADC minimum threshold.
The 0-dB value on the Signal and Noise scrollbars is just sufficient to cause the ADCs to exceed the ADC upper range.
As well as overdriving the ADCs, the effects of ADC Linearity errors and Clock Jitter can also be viewed.
The latter is of importance for higher aliased frequencies, and this can be observed by noting the effects as the RF signal is increased beyond the Nyquist frequency (1024MHz for a complex FFT default example).
The effect of quantization, DFT points, window function, pulse width, rise-time, and pulse position on the signal output SNR can also be evaluated.
Signal-to-noise ratio is assessed using the cursor strobe, from measurements of the signal peak, and on the noisy baseline. Note that the input RF noise power is divided between the DFT channels (length). In the default example with the RF band = 1.024 GHz, the bin noise is down by 10log(1024) = –30.1 dB.
Both independent signals that are available may be programmed with different parameters and are selected for adjustment using the Modify Signal radio buttons at the bottom of the applet. It is instructive to view the effect of complex signal types on the detectability of weak signals.
The Save button saves the FFT bins and powers to the Data Console together with the parameter values selected for a particular design.