Hello everyone,
I've been looking for the answer to this question for some time now, haven't come up with anything good.
I'm currently simulating an amplifier before I breadboard the prototype. Using LTSpice to get an idea on component values to optimize bias and minimize distortion throughout the audio band. I also simulate the power supply, which is just a simple transformer/bridge rectifier/filter circuit.
Since the power supply takes a while to come up to full voltage due to the big capacitors, I use the delay function of the signal voltage source to apply the signal once the power supply stabilizes. This results in rather long simulation times, in comparison with 50 cycles at 20kHz.
For good distortion results, I keep the max timestep very small for high frequency analyses. But this results in excessive simulation times when you include the power supply stabilization times.
So here's the big question: is there a way to specify different max timesteps for different time ranges in the transient analysis? That way, I could use a rather long timestep for the first few seconds of the analysis, when no signal is applied, to simulate the power supply stabilization period, and a much smaller timestep when I apply the signal.
Anyone ever done that?
I've been looking for the answer to this question for some time now, haven't come up with anything good.
I'm currently simulating an amplifier before I breadboard the prototype. Using LTSpice to get an idea on component values to optimize bias and minimize distortion throughout the audio band. I also simulate the power supply, which is just a simple transformer/bridge rectifier/filter circuit.
Since the power supply takes a while to come up to full voltage due to the big capacitors, I use the delay function of the signal voltage source to apply the signal once the power supply stabilizes. This results in rather long simulation times, in comparison with 50 cycles at 20kHz.
For good distortion results, I keep the max timestep very small for high frequency analyses. But this results in excessive simulation times when you include the power supply stabilization times.
So here's the big question: is there a way to specify different max timesteps for different time ranges in the transient analysis? That way, I could use a rather long timestep for the first few seconds of the analysis, when no signal is applied, to simulate the power supply stabilization period, and a much smaller timestep when I apply the signal.
Anyone ever done that?
spice doesn't do literal "warm up" - the device temperature is a static entry in the model
power on transients can be an issue too - probably what you mean
one possiblity is .loadbias/savebias files - you run a sim at a longer step time, for as long as needed to reach steady state
with the
.savebias ".\mybias" internal time=10s
file statement set to run at the end of the long sim and save the operating point at that time
then you add/enable the .loadbias ".\mybias" read of that saved operating point for the start of your high time resolution sim
power on transients can be an issue too - probably what you mean
one possiblity is .loadbias/savebias files - you run a sim at a longer step time, for as long as needed to reach steady state
with the
.savebias ".\mybias" internal time=10s
file statement set to run at the end of the long sim and save the operating point at that time
then you add/enable the .loadbias ".\mybias" read of that saved operating point for the start of your high time resolution sim
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.