I see that junction is not connected to anything else in the circuit, so
voltage cannot be determined.
But I always measure (in real life) voltage inversely proportional to
capacitance of capacitor at such junctions.
How comes than that it is indeterminate? (Maybe because we don't know prior
state (charged) of capacitor?)
(I don't have any practical expirience in building circuits yet).
If two caps are in series, the voltage at the junction can have any
value, even if the end voltages are known. This is the
initial-conditions problem. Any mathematical integral (the c voltage
is the integral of i/c) has a 'constant of integration' which just
means that we can't know the history of a system backwards infinitely
in time.
In real life, if caps are discharged and then placed in series, and a
signal applied, the junction voltage *is* known, at least until
leakage makes it drift away (which could be years for good caps.) But
Spice is picky enough to not allow the constant-of-integration thing
to stand. I think that maybe SPice does DC analysis with all caps
open-circuited, which leaves such nodes hanging; maybe Kevin can
elaborate.
So just plonk a 1 Tohm resistance from all such nodes to ground, or
shunt the caps with suitably scaled resistances; either makes Spice
happy.
John