I've been trying to use the serial ports on my Sun-3/260 (NetBSD/sun3,
1.2 plus my patches) to interface to external discrete electronics.
I've found the bizarre result that - at least for input to the Sun - it
seems that if I connect my circuit's reference point to pin 7 on the
connector, the nominal "signal ground" pin, it doesn't work.  But if I
leave pin 7 open and connect my circuit's reference point to earth
ground, it works!  It is as if the hardware is referencing the input
(and probably output) signal to earth ground and ignoring pin 7.  (How
did I discover this, and what is my evidence?  Well, my circuit's power
is isolated from earth.  I have a dumb terminal on hand, and was using
it to snoop the wire; when I connected it "in parallel", everything
worked; disconnect it and input stopped working.  I started connecting
one pin at a time, eventually discovering that pin 7 was the key;
further trials revealed that it worked iff my circuit reference point
was connected to the terminal's pin 7, regardless of whether the Sun's
pin 7 was connected.  Unplugging the terminal broke even this, and an
ohmmeter shows zero resistance between the terminal's pin 7 and the
ground pin on its power cord.  IMO this is severely broken of the
terminal; circuit reference ground should not be earthed - that's just
_asking_ for ground loops when someone uses something like a 24VDC
supply with one side earthed, bisecting it to get externally-advertised
signal ground.  But of course the Sun is also broken - pin _1_ could
reasonably be earthed, but connecting my circuit's reference point to
pin 1 does nothing, and the diagram in the -3/260 book shows no
connection for pin 1.)

This is so totally bizarre I just have to ask, is it just that this
particular board is fried, or do other people see similar brokenness?

					der Mouse

			       mouse@rodents.montreal.qc.ca
		     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B