DC offset LTSpice

[Neutrality]

While simulating a slightly modified amplifier circuit from Bob Cordell's book with LTSpiceI can easily get a DC offset of only 3.6 mV if going by the DC Op. Pnt. analysis. Although not as easily obtained in the real world, I am hoping it wont be too much of a difference from the simulation.

However doing a transient analysis with some .four THD analysis thrown in I get what you see on the attached picture.

Fourier components of V(output)
DC component:-0.049008

Is that something to worry about? What eaxctly is the DC component and can it just be ignored? You usually measure and if possible adjust DC Offset without a signal applied so whatever DC compoent you get with a signal should not matter.

[sawreyrw]

When LTspice performs a .four analysis it uses the last cycle of the waveform. It is possible that the DC component is due to the fact that not all start up transients have been comleted. Try running the circuit for a longer time to ensure all transient are complete.

[Neutrality]

Simulation transient analysis for 320ms and using the last 160ms instead of simulating for 16ms and using the last 8ms as in my first example, I now get this :

Fourier components of V(output)
DC component:-0.00082819

Which is alot better and below the 3.6mV of the DC Op. Pnt. analysis.

Guess I dont have to worry so much about it after all.

Thanks.

[AndrewT]

The sim will take identical components and predict what the DC offset is, using these matched components.

Your components will not be identical. Some errors will increase +ve output offset, while other errors will increase -ve output offset. Some of these +ve and -ve offsets will cancel leaving a small or large residual offset.
This residual will change with time and with operating temperature. Expect the output offset to change significantly during start up and possibly during shut down.
The offset will also change while each of the components is warming up to operating temperature. The slowest to warm up are those that have a tiny dissipation and no heatsink, i.e your To92 dissipating 5mW will be at ambient at turn on and still near or at ambient after a minute or so. While the temperature inside the case increases due to all the other dissipations the To92 will very slowly warm up trying to track the internal ambient temperature.

This is where a servo shows it's power.

[sawreyrw]

LTspice will use the last msec of the simulation to determine the Fourier components for .four of 1 KHz. It doesn't matter when you start saving data so long as you save the last msec.