KL10-A/KL10-B differences

[Eric Smith]

On Mon, Jul 8, 2019 at 1:30 PM Noel Chiappa via cctalk <
cctalk at classiccmp.org> wrote:

> So early KL10's (KL10-A's, to be precise) only support a single DTE20, and
> no RH20's. Later ones supported up to 4 of the former, and up to 8 of the
> latter.
>

That's because the 1080 has different I/O backplanes than the
1090/1091/1095/2040/2050/2060.

The 1080 was intended to replace a KA10 or KI10, so it would be used with
external memory and external channels. It only needed a single DTE20 for
the internal console PDP-11, and it didn't need an RH20 because the disk
would be attached via an RH10 on the external I/O bus and external memory
bus.

I always supposed this to be part and parcel of the 'Model A/Model B' CPU
> difference, but no... EK-0KL10-02 Part 1 (no title, seems to be notes for
> F/S) pg. 9, says both KL10-A and KL10-B's are PA (DEC jargon for the Model
> A CPU - below), but the former has no RH20's, the latter does.
>

The -PA and -PV designations (and much later -PW for -1095 and -2065) are
for the "arithmetic processor" (APR), which is the main CPU portion of the
KL10. That's independent of the number of DTE and RH20 slots, though AFAIK
DEC never sold a configuration with a -PV CPU and the small I/O backplane.

>
> (A note at the bottom of the page says that a PA is a 'Model A', and
> describes it as having "internal channels". The PV is a 'Model B' -
> extended addressing, larger ucode, faster clock.)
>

A system with a -PA APR could have internal RH20 channels or not, and it
could have cache or not, depending on the system configuration
For example:
* DECsystem-1080: cache, external memory bus adapter (DMA20), and external
I/O bus adapter (DIA20), but no internal channels
* DECSYSTEM-2040: internal channels (usually two) but no cache or external
memory bus adapter.

So my new theory is that it's the MBox (either the backplane, the boards,
> or the wiring from it to connectors, etc) that is the difference between
> the KL10-A and the KL10-B.
>

It's not just the MBOX; there are significant EBOX differences as well.
Various modules from the entire CPU are different, and the backplane wiring
is slightly different. It was possible to upgrade a -PA to -PV by swapping
modules and adding some wraps to the backplane, but if you had the small
I/O backplane, upgrading to -PV didn't make it possible to add any more
channels or DTE20s.

Part of the motivation for upgrading a -PA to -PV, even if you weren't
planning to use extended addressing, is that the -PV is faster than the -PA.

Eric