2. Testing methodology
The linear system around a periodic solution is a Linear Time-Periodic (LTP) system. The behaviour of such a LTP system can be described perfectly by a infinite series of Frequency Response Functions (FRF) called Harmonic Transfer Function (HTF) [1]. The HTF which describes the frequency response from the excitation frequency to a translated frequency will be denoted as . is used to indicate the pulsation of the LTP system.
To test the LSSS simulator, a periodically time-varying differential equation is implemented in ADS. The HTFs of such differential equations can also be calculated in Matlab using techniques described in [1]. By comparing the Matlab results to the results from ADS, we can get an indication of the error made in the simulator. The following periodically time-varying first order differential equation will be used:
The time-variation will be a random-phase multisine with a base pulsation of . The multisine excites all tones up to . The amplitude of each tone in the multisine is set to and random phases are used. A DC offset of was added to the multisine as well. The resulting signal is shown in Figure 2.1.
Figure 2.1 Values of the parameter in the differential equation (2.1) as a function of time.
The implementation of this differential equation is done in ADS with a Symbolically Defined Device (SDD) block. The simulation set-up is shown in Figure 2.2. The SDD block in the schematic is configured such that the voltage at port 1 acts as the input signal and the resulting current flowing into port 1 is the output signal . The control of the factor is done through the second port of the SDD block. It is important to force a non-zero initial condition in that port, as setting the node to zero leads to errors in the simulation.
The whole circuit is simulated with a HB simulation with a base frequency of which is given an order of . The LSSS was run on a logarithmic frequency grid going from to with steps per decade. With these settings, ADS returns HTFs sampled at frequency points.
Figure 2.2 Simulation set-up in ADS to implement the time-varying differential equation (2.1)