A little HP9877 info

From: joe <rigdonj_at_intellistar.net>
Date: Tue Jun 26 17:00:31 2001

     At 09:26 PM 6/26/01 +0100, you wrote:
>I rmemeber that somebody here was looking for info on the HP9877
>'External Tape Memory'

      I was one of them.



>This is a rackmount box with up to 4 tape
>cartridge drives on the front, which can be connected to a 9825
>calculator (and maybe to other machines too). It can be used (I believe)
>to copy just about any kind of file between tapes, make distribution
>tapes, and so on.


      Yes, that's what it was used for within HP. One HP engineer that I
talked to said that it was used at HP to produce all of the tapes that HP
sold. He thought it was used for tapes for machines other than just the
9825 but he wasn't positive. I think that's likely since no other tape
duplication device was made at HP AFIK.


>Inside the box are 4 main assemblies :
>
>1) The PSU (transformer + low voltage switching regulator for +5V +
>linear regulators for other voltages)
>
>2) The tape drive(s) (1-4 units can be fitted)


      Correct. One unit is standard and the others are optional so you can
find 9877s with oen, two, three or four tape units installed.


>3) The controller board(s). One board per tape drive installed. This
>handles motor speed control, and data conversion between a bitstream and
>the signal to the read/write electornics. And other misc functions like
>end-of-tape detection
>
>4) An interface board that links to the controller boards and to the 50
>pin Blue Ribbon connector on the back.
>
>Here's what I've discovered so far :
>
>Much of the insides came from the 9825. The tape drive and PSU are
>straight out of the 9825. The controller board nearly is -- there are a
>few very minor changes. In particular, the tape controller is the same to
>software as the one in the 9825. There are the same 8 'registers' (4 for
>input, 4 for output), with the bits in the same places. I would be
>suprised if 9825 and 9877 controller boards can't be interchanged, actually.
>
>That leaves the interface. Much of the elctronics on that is to provide
>the right sequence of signals to clear a couple of status flip-flops on
>the controller boards. It's a chain of monostables to provide a serquence
>of pulses that first selects all 4 controllers at the same time, and then
>accesses a couple of the 'registers' which have the single effect of
>clearing said flip-flops.
>
>The external connector is pretty much what you'd expect. A lot of the
>pins aren't used. Of the ones that are, you have :
>
>8 bidirectional data lines, to talk to the controller registers (==IODn
>on the 9825 bus)
>
>2 register select lines (== ICn on the 9825 bus)
>
>2 drive select lines
>
>Assorted control lines (== IOSB, DOUT, DMAR, etc on the 9825 bus)
>
>Unit Enable (disables access to all controllers when low, so you could,
>in theory, have multiple 9877s on the same interface)
>
>Initialise (clears just about all controller registers)
>
>The interface is very low level. If you want to read a tape, you set the
>tape going and every so often you get a data request from the 9877. You
>then read one of the registers. Bit 0 is the data bit just read from the
>tape. THe 9877 does not pack the bits into bytes or words.
>
>I am told this unit links to a 98032 parallel interface. Well, it's
>possible, but it would be a bit of a kludge. That doesn't suprise me too
>much, as the whole thing is a bit of a kludge (it could have been made a
>lot more elegant if the controller board had been slightly redesigned).


     One place that has a "working" 9877 send me a picture of it. It
showed a 98032 interface connected. It was hard to read but it looked like
it said "opt 077".

>For example, the chain-of-monostables on the interface board could have been
>replaced by a clear input to the flip-flops on the controller boards.
>
>There also appears to be no circuitry to directly copy between tapes
>without going via the 9825 host. This has been rumoured about, but my
>guess is that copying a tape uses a DMA buffer in the 9025. The 9825 will
>do a lot of things 'in the background', so it's quite possible that you
>could carry on using the machine for normal calculations while it was
>copying a tape, giving the impression that the 9825 wasn't really involved.
>
>If anyone has the details on how to link this to a 9825 (I guess it's the
>Option 077 for the 98032), then I would be interested in finding out more


     So would I!

      Joe


>-tony
Received on Tue Jun 26 2001 - 17:00:31 BST