Has anyone ever had trouble driving a G201 w/ an unbuffered parallel port?

Hello!
Has anyone ever had trouble driving the G201's opto inputs with an
unbuffered parallel port? the minimum input current requirements of
the gecko is 16mA, and from p.108 of Jan Axelson's "Parallel Port
Complete" it says printer ports can sink 12ma when the outputs are
0.4V (logic low). It also says that printer ports have an output
impedance of 45 to 55 ohms. On the few printer ports I've tried,
it's always worked, but I'm concerned that it is 'out of spec'.

I have no problem adding a 74ACT245 buffer to drive the LEDs, but
I'm afraid that this will un-do the majority of noise immunity that
the opto-isolators provide. It's my understanding that optos work on
current, not voltage. And that most noise sources are high
impedance; too high to change the state of an opto. But the inputs
of logic gates are high impedance and operate on voltage (not
current). So a high impedance will have no problem changing the
state of an input of a logic gate.

If I buffer the printer port, should i low-pass filter the lines
and then square them up with a schemidt trigger? This won't prevent
noise like optos will, but will filter out spikes shorter than a
certian amount.

What do 'commercial' break-out boards do? Do they buffer, and
noise filter at all?

Has anyone tried any of the line termination techniques like what
is shown on page 112 of the "Parallel Port Complete"? One of them
shows each line on the reciving end with a 100 ohm resistor in series
with a 100 pF cap, with the other end of the cap to ground.

Sorry for all the questions, but I've been plagued by noise before,
and want to make this box 'bullet proof'.

Thanks!!

- Steven Ciciora

P.S. As they say, in Engineering, there are no answers, only
compromises. The 'best' answer is the one that minimizes the
compromises.