Simulation With Spice, How much reliable?

[fala]

Hello, I have built a Voltage Booster for OpAmp and I simulated that with spice in Protel DXP 2004 software package. the design is taken from National application note AN-272 120V Swing Booster(**broken link removed**).
what I need is maximum stability of voltage and capability of maintaining very low voltages. for driver stage I have used different push pulls: Bipolars(as original document), bipolar darlingtons and mosfets. all of them complementary npn, pnp(n, p channels in case of mosfets). I used 1uV(1 micro volt) sinusoidal(1KHz) input signal and compared outputs(in absence of any external noise). some transistors definately had smoother outputs than others. I want to know how much these results are accurate ? if I had no noise in real world, would I get these results? can I rely on results of comparisons between different transistors from this software?
thanks.

 

[flatulent]

Simulators are only as accurate as the models used. At lower frequencies the models are acceptable. You should also no a .noise analysis. This is part of the .ac analysis.

 

[fala]

thanks flatulent, I want to build a constant current source with voltage booster that I've explained. the resolution of current should be better than 1nA(current noise less than 1nA). with spice simulator I've found out darlington BJTs are usually better than MOSFETs. but in theory I think noise current of MOSFETs are less than BJTs in general. so this made me a little confuse. can someone explain this?

 

[flatulent]

It all depends on the frequency range. At RF and microwave the MOS is quieter. At audio the BJT is generally quieter. JFET are even quieter but not high power.