From: Johnny Billquist
Early DMA controllers were all multiple cards, so
having their own
dedicated backplanes were a pretty sane idea.
Well, there was also that large intermediate generation which still had their
own backplanes, but they were 'system unit' (for lack of a defined term for
this form factor) backplanes (e.g. RK11-D, RH11, etc), which fit into e.g. a
BA11-K, along with other 'generic' UNIBUS backplanes (i.e. DD11-C, etc).
That first generation I spoke of (the one with backplanes that mounted
directly in H960's) are marked by the use of lots of small Flip Chip cards,
not the larger quad/hex boards that one finds in the 'second generation'
(above). I guess the larger board versions were cheaper to manufacture, which
is why they got rid of the first generation ones in favour of the second
generation (which were often functionally identical to the first-generation
ones they replaced, e.g. RK11-C and -D).
Those first-generation one used very similar construction technique to the
KA10 generation of machines, which also used that size Flip Chip (although a
different series, mostly with individual transistors), and heavier wire on the
bacplanes. What did KI10's use, does anyone know? I have this bit set that
they used roughly the same kind of Flip Chips as the 'first generation' PDP-11
DMA devices, but I've never seen a KI in person.
Noel