Also, in HB sim, non-linear models for FETs & BJTs are to be used. So when a feedback is required to make the active device unstable, should I use the linear (small-sig) model or the non-linear one (which will be the one eventually used in HB sim)???
Also, in HB sim, non-linear models for FETs & BJTs are to be used. So when a feedback is required to make the active device unstable, should I use the linear (small-sig) model or the non-linear one (which will be the one eventually used in HB sim)???
To simulate mixers or self oscillating devices, a HB simulation must be used with Non-Linear" models of the circuit components.Because either mixers and oscillators are Non-linear devices and therefore small signal models are valid anymore.
In HB simulation, you have to define all possible frequency components either input(s) and output(s). If your mixer is a self oscillating circuit,simulation will be a little bit difficult rather than only a simple mixer due to the simulation complexity because of oscillator and mixer function are on the same system. But it's possible.
In my opinion such system like yours, Kyrylov and Direct method HB setup is more suitable to find all mixing products and hard nonlinearity of the devices.
In fact, there are some possible ways to follow in these kind of simulations, the better way "try and see".
I've tried doing HB sim for my SOM but the results I get are very different from what is expected from a mixer:
for a normal mixer, the harmonic corresponding to the RF has power similiar to the input power (right?). For my SOM, the RF harmonic is -70dBm for a -10dBm input using a p_1tone.
I guess this means that the RF is "competing" with the LO and somehow it doesn't enter the SOM for mixing?
Sorry i forgot one more thing. Ansoft designer can perform a 2-tone oscillator mixing simulation, it automaticaly find out the oscillation frequency and do the HB simulation. It is very suitable for your SOM, but you have to be sure that LO pulling is not occured.
Use HB simulation only if you know what exactly happend in your circuit. If your oscillator tends to oscillate at another freqency too, or frequency pulling is occured, you'll get a starnge output spectrum. I suggest running a transient at 1st, ensure the the mixing function is correct, then use HB to simulation and optimize your circuit.
I think the pulling pulling occurs in my SOM... I'm using a dielectric resonator to fix the freq. Maybe I'll try running a transient 1st to see if the freq is correct.
By the way, how is conversion gain calculated (without using HB to see the output spectrum)?
It seems that there is no LO pulling effect in your SOM because RF and IF frequency is not very very close. LO pulling means that, for example, if the LO freq. is 2.4GHz and the RF input freq. is 2.41GHz, with a large RF input signal, the LO freq. will be DRAG to 2.41GHz and you'll get only one large power tone on your output spectrum.
Run transient under SPICE and do a FFT(fast fourier transform) with your output waveform data. Seek for your IF tone at the designed freq. The magnitude difference in dB between IF tone ond input RF tone is the conversion gain (or loss). It takes a long long time to do transient so that's why i suggest you optimize your conversion gain using other HB simulators. But many HB based software also include transient simulation ability, just use your favorite one.
For transient simulation, someone suggests that (1/timestep) > 10* Fmax. Fmax is the highest frequency among all frequencies which you are interesting to. Namely, each cycle of a sine wave is interpolated by 10 points(as for me, I set to 40* Fmax). Some transient simulators do not use one fixed timestep, they dynamically adjust the timestep value so you can only set maximum timestep. Set timestop to the time when your oscillator seems to be under steady state.
back on the issue of SOM sim in ADS. I checked with the support staff, ADS can only do either oscillator or mixer simulation. So SOMs cannot be simulated in ADS.
Anyone knows of other software that can simulate SOM?