Spreadsheets do not account for complex, frequency-dependent terminations, for example (table 1). Therefore, gain, isolation, and terminations to harmonics are misrepresented and can result in gross errors. Making topology changes to optimise performance is time consuming, and interfaces to other synthesis or analysis tools are non-existent.The Genesys software from Agilent Technologies has a unique simulator called Spectrasys. It bridges the gap between Excel and numeric system engines, allowing easy estimation of realistic RF system performance early in the "architectural" phase of a design. It accounts for analogue effects, like spurs leaking back into the band of interest and overloading an IF amplifier, that are missed by the numeric tools, and provides an elegant path to implementation-in-place, unlike Excel spreadsheets. Spectrasys also is inexpensive and simple to use. Our signal monitor requires a simple single conversion receiver and sufficient gain to provide a measurement signal for spectral and power measurements.Input power: -90dBm- -40dBm (50-dB dynamic range)Bandwidth: 2350-2450 MHzNoise Figure: < 5dBSpurious free range: 30dBThe process starts with a top level design. Here we define the topology and initial specifications for each of the components that make up the RF band monitor (Figure1).

We set the filter type, number of sections, insertion loss, out-of-band attenuation, etc. The amplifier parameters dictate the noise figure 1dB compression point, saturation, 2nd, and 3rd order levels. Additional amplifier models allow up to 20th order or offer variable gain. We can also incorporate the linear S-parameter data with our amplifier in a hybrid model, accounting for the effects of complex terminations and the interactions they pose with filters and mixer components. Spectrasys can also substitute actual measured data for filters, amplifiers, and mixer table data. Once we have insinuated the component parameters we can view gain, noise figure, and spectra at different nodes and along the primary receiver path.Figure 2 represents a path measurement, gain and cascaded noise figure, from input to output nodes. The gain and noise figure contribution is shown for each stage along the path, which is helpful for troubleshooting and optimization. Any of the path measurements can be optimised using Spectrasys to further help meet design goals. The optimisation feature allows us to better specify the final parameters of the system component such as amplifier gain, IP2, IP3, etc.Quickly changing the topology, impossible with spreadsheet methods, allows us to view the effects almost instantaneously. For example, changing the position of the input filter to follow the first amplifier brings us within the original goal of 4.5dB noise figure. The change is due to the increased noise power produced by our amplifier and attenuated by the filter. Other design process steps include selection and specification of the mixer. Instead of using a generic behavioral model with specified non-linear parameters we can use accurate, measured, table-based mixer data, available from numerous manufacturers. An analysis/synthesis tool called WhatIF also was used to help select a manufacturer's part by comparing the spectral content derived from the mixer's tabular data. WhatIF helelps accurately predict the level and location of spurious products around the IF frequency from local mixing as well as products from adjacent radiators such as GSM or PCS bands. Optimisation of component parameters in the final topology allows us to meet the power output requirements and detection sensitivity of the WiFi band power monitor. Monte Carlo analysis also is used to predict the behavior changes due to variations in the component parameters. The optimised parameters are now the component specifications for vendors or designers of the modules. Component synthesis was provided through the Spectrasys interface. The oscillator, filters, and directional couplers were all synthesised and built from within the environment. Results are shown in Figure 3. After synthesis, the oscillator data is inserted into the oscillator behavioral model, providing accurate spectral and phase noise characterisation in the overall system simulation. The models used for simulation are selectable. After synthesizing the filter and coupler, the actual S-parameters were inserted and used in the final simulation. Due to additional loss in the RF filter, the noise figure and output power were degraded by another 0.3 dB and 1.5dB respectively. The results give us assurance that what we have simulated will match the measurement because of the improved accuracy of spectral analysis. They show that we have met all the stated goals for this receiver. After fabrication and assembly the final monitor-receiver is measured to ensure compliance to specifications using the Testlink tool, linking the simulation environment to instrumentation. Spectral data, noise figure, and time waveforms can be imported for comparison to simulated results. What previously took weeks to design and verify can be done in hours.