Hello, i am trying to build a simplified model testing a simple principle described in the end.
i am supposed to get a current switch from side to side , thus getting a square wave pattern.
but in the Tran simulation i only get flat line when i look at the differential signal ,as shown bellow.
its a very straight forward principle,where could i go wrong?
I think it's found a metastable point. That's not surprising for an oscillator, since the DC operating point calculation will remove the capacitances and short inductances and then solve the DC equations which will likely find a metastable state. Then when the transient starts and the capacitances and inductances are re-inserted, nothing changes and so everything stays flat.
Most likely you'll need to give an initial condition (Simulation->Convergence Aids->Initial Condition in ADE) on each of the out_n and out_p nodes to hold it so that one is high and one is low - if you like it's "sprung", ready to go! This is a standard technique to help oscillators to start oscillating in transient.
Hello, Thank you very much.
Since the amplitude seems to reduce a lot, you should also check which method is being used. This will be listed at the top of the tran analysis in the simulator log - check if it’s set to gear2ony. If you’ve set errpreset to conservative or explicitly set method to gear2only, I would suggest setting method (on the Options form on the tran analysis in ADE) to traponly.
Gear methods (whilst generally a good choice for accuracy) are not so good for oscillators because they introduce a small amount of numerical damping which can cause oscillation to die out or reduce in some cases. Trapezoidal integration method doesn’t have this issue.