>> There is no such thing as a 10base5 <-> 10base2 transceiver, really.
>> If you mean an AUI <-> 10base2 transceiver
> I guess I mean AUI; I really don't know anything about hardware from
> that era :-)

:-)

Way back when, Ethernet was what is today often called (when it's
remembered at all) "thicknet".  The cable was co-ax, about as big
around as your (or at least my) thumb, and about as flexible as garden
hose.  This used transceivers, which were gadgets that generally
(universally, in my experience) had a DA15 connector for communication
with the host and some kind of interface to the co-ax.  (I've seen two
kinds of co-ax interfaces - vampire taps, which have a spike that
punches down to the centre conductor and much shorter spikes that
connect to the outer braid, and in-line, which require cutting the
co-ax and installing connectors on the ends.)

This was 10base5.  (As I understand it, the 10 is the bit rate,
megabits per second; the "base" stands for "baseband"; and the 5 has
something to do with the cable, but I'm not sure what - fuzzy memory
says it's the maximum cable length in hundreds of metres.)

This is also the heyday of AUI ("attachment unit interface" or some
such; I don't think I ever really knew the official expansion).  This
is the DA15F connector with the funny slide latch that always breaks.
This was typically connected (with a "drop cable") to the wiring area
where the co-ax ran and where, perforce, the transceiver was.
High-quality drop cables were almost as thick as the co-ax itself,
though fortunately they were somewhat more flexible.  Long drop cables
could be up in the 50-100 foot range.

Then came "thinnet", known for a time as "cheapernet".  This used much
thinner co-ax, somewhere around 1/8 or 3/16 inch.  Transceivers still
existed, but for machines intended to use thinnet they were normally
built into the machine.  With thinnet, rather than using a drop cable
for the machine to come to the network, instead, you brought the
network to the machine (which is why it was practical to build the
transceiver into the machine).

This was 10base2.  (Fuzzy memory says the 2 is, again, max cable length
in hundreds of metres.)

Thinnet and thicknet are compatible at the on-wire signal level, so
much so that it's possible to paste a thinnet segment and a thicknet
segment together with nothing but a passive connector adapter.

Then came 10baseT.  This was a significant change in the physical-layer
topology; it changed from a shared bus to point-to-point.  10baseT hubs
are just signal repeaters, thus producing the illusion of a shared
physical medium even though it no longer really existed.  Again, given
the large installed base of machines with AUI network connectors,
AUI<->10baseT transceivers immediately appeared.  (Switches, as opposed
to hubs, are a very different animal.  True hubs are rare these days;
the extra cost to build a switch is now trivial.)

The difference between a repeater and a transceiver is that a repeater
basically speaks the same thing out each side - a carrier signal
amplifier, usually little more than that - whereas a transceiver does
things like modulation and demodulation, collision detect, and such.

Your "10base5 switch", from your description, is something I've never
seen.  There was a device that was relatively common in the AUI world,
called a multiport, or AUI repeater, or various other such names.  This
was a device that had one host-side AUI interface (DA15F) and a bunch
of transceiver-side AUI interfaces (DA15Ms).  You plugged a bunch of
hosts into the DA15Ms, and the multiport looks like a transceiver to
each of them.  You then plugged the DA15F into a real transceiver, and
it looks like a host to that.  What you describe as multiple DA15Fs,
though, which means multiple host-side interfaces - it's a device you
would connect to multiple transceivers, and I have trouble thinking
what it could be.  It sounds like something that bridges multiple
segments, but I hadn't thought you could do that at the AUI-interface
level.  Obviously there's something I don't really understand (which
isn't surprising; my knowledge of AUI is more practical than
theoretical).

> And my DEC 10base2 transceiver/repeater/whatever only has one LED
> that I can see, too. (but it's about five times as big as this one,
> so it's probably older [...])

Sounds like mine - one of my AUI<->10base2 tranceivers is made by DEC,
and it's big and heavy and probably insanely robust, overdesigned by a
couple hundred percent.  When I next am in the same building as my
collection of such things, I can dig it out and tell you its model.

/~\ The ASCII				der Mouse
\ / Ribbon Campaign
 X  Against HTML	       mouse@rodents.montreal.qc.ca
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B