This is drawing something up from the deep dark recesses of my memory. I seem to recall something about there being some delays in setting each pin, and that it might be best to set them all together (i.e. to construct an entire byte to send rather than set each bit separately).
I can’t find that thread, but check here:
http://www.psychopy.org/api/hardware/labjack.html
and here:
to make sure you are using PsychoPy’s modified U3 package and then you can access the setData() method: https://groups.google.com/d/msg/psychopy-users/4Fxoqbqt5lw/YsHfvypIP-sJ
Some discussion was here: https://labjack.com/forums/u3/bit-latency-digital-output-fio0-fio7 but I’m not sure that those sort of microsecond delays are causing your problem, but they can cause issues at the receiving end, as each bit is only available to be read very briefly before a new bit is set, wiping out the previous one.