I found that when I run parametric sweep with fine voltage step, netlists are sometimes generated with wrong value. For example, when I run parametric analysis with the voltage changing from 0 to 1V with 1/2560V(=390.625uV) step, 62.5mV is changed to 62.4999999999998V in the netlist.
Is there some kind of precision issue during parametric sweep? Is there a way to fix this? I am currrently using ic-6.1.5.-64b.500.9
Why does this really matter? It's accurate to 3 parts in a million billion (1e15), so that ought to be good enough for all practical purposes... (that's the limit of floating point precision, so you're seeing a floating point rounding error).
In reply to Andrew Beckett:
No, It doesn't matter as you said. I just found this while running an ADC with ideal components. I was not getting ideal codes, so was curious why.
Thanks for the answer.