I am trying to simulate a varactor frequency divider (which I have build and tested in the lab) in ADS. It contains nonlinear capacitances (varactors) as well as balanced microstrip coupled lines. I have tried both Transient and HB as well as TAHB - but I cannot successully simulate the circuit to see the half frequency.
From the documentation, I understand that in order to use HB, or in this case TAHB, we must have a Transient simulation that works and reaches steady-state, so that we can use it in TAHB. But I have convergence problems even in the Transient simulator. Whatever I change or edit, I keep getting the same error message that "internal timestep too small at time ....". I have tried changing MAxTimeStep, Time Step Control Method, Integration, Convolution etc. ... but still get the same error message. Occasionally this disappears and the simulation finishes without errors but it's not consistent.
Can you please guide me as to what I can do to eliminate these convergence errors and to get a close enough simulation to the measured data?
[NOTE: The divider circuit has a threshold level on the input power and very dependant on the nonlinear elements - could this have anything to do with the problem?]
From the documentation, I understand that in order to use HB, or in this case TAHB, we must have a Transient simulation that works and reaches steady-state, so that we can use it in TAHB. But I have convergence problems even in the Transient simulator. Whatever I change or edit, I keep getting the same error message that "internal timestep too small at time ....". I have tried changing MAxTimeStep, Time Step Control Method, Integration, Convolution etc. ... but still get the same error message. Occasionally this disappears and the simulation finishes without errors but it's not consistent.
Can you please guide me as to what I can do to eliminate these convergence errors and to get a close enough simulation to the measured data?
[NOTE: The divider circuit has a threshold level on the input power and very dependant on the nonlinear elements - could this have anything to do with the problem?]
http://eesof.tm.agilent.com/support/supp101.html
The following comments are general in nature. In transient make sure that you are sampling the input signal using at least 1/(5* period). This might be enough for the divided signal but if you have a very strong nonlinear circuit you may have to increase the sampling frequency. If you are using digital circuits (e.g. flip flop circuits) you may even have to set the initial conditions for the circuit to converge, since the initial voltage may not be defined from the circuit topology.
In HB you need to make the analysis frequency equal to the divided frequency and the order will have to be high enough to cover the input fundamental frequency plus some harmonics. For example a divide factor=4 and an input of 1 GHz, you can set Freq[1]=250 MHz and Order[1]=12. This will cover up to the 3rd harmonic of the input frequency.
Your convergence issue is difficult to troubleshoot without more circuit details. If you still encounter issues after trying some of these pointers please contact technical support in your region.