NetNews Usenet Archive 1992 #27/NN_1992_27.iso/spool/alt/sys/pdp8/78

Text File | 1992-11-17 | 7.5 KB | 166 lines

Newsgroups: alt.sys.pdp8
Path: sparky!uunet!cis.ohio-state.edu!zaphod.mps.ohio-state.edu!sol.ctr.columbia.edu!news.columbia.edu!watsun.cc.columbia.edu!lasner
From: [e-mail] (Charles Lasner)
Subject: Re: Schematics for an M8330 board?
Message-ID: <[e-mail]>
Sender: [e-mail] (The Network News)
Nntp-Posting-Host: watsun.cc.columbia.edu
Reply-To: [e-mail] (Charles Lasner)
Organization: Columbia University
References: <[e-mail]>
Distribution: na
Date: Tue, 17 Nov 1992 04:17:33 GMT
Lines: 151

In article <[e-mail]> [e-mail] (Paul Popelka) writes:
>
>Hi,
>
>I just bought a pdp 8/m and I'm trying to get it working.
>When I got it home I discovered that someone had put the
>M8330 timing generator module in backwards and, it looks
>like they then powered it up.  This fried one 7401 chip.
>I'm going to replace it, but I assume other chips will
>be bad too.
>
>So, to find the rest of the bad chips I probably will need
>a schematic for this board, and maybe others.  Does anyone
>out there know where I can get a copy of the schematic (for
>the M8330 or better yet the whole machine)?
Lotsa readers have schematics of the card, for at least one viable revision of
the card.

I have a general question about ECO's to the CPU cards to address to the
those more familiar with newer logic families:

One of the major upgrades to the CPU cards involves a wholesale upgrade of
chips generally from 74H to pin-compatible 74S, and minimally changing some
of the bypass caps to larger values.  Apparently this tightens up some
buss timing critical to operation in long busses, etc.

My understanding of Schottky logic is that S series chips are rather gross,
in that they generate lots of switching current that can induce unwanted
signals in lead runs, that can cause unreliable results, etc.  So, is the
etch of the affected boards "worthy" of this upgrade, or is doing so sort-of
trading in one form of unreliability for another?  I have seen boards with
elaborate grounding techniques when S chips were used, and these caps are
really rather "dinky" compared to that, and even have leads longer than
necessary, as compared to the usual glass caps that resemble diodes (with
smaller capacitance values; I think .01 as compared to .25.)

An additional question:

Is perhaps a newer logic family a better alternative?  Can someone
post all of the compatible variants so we can get a feel for what is
suitable/not suitable?  I am aware of:

Plain 74xx
74L
74H
74S
74LS
74ALS
74C
74HCT
74F

Someone suggested that 74F would be the best improvement, and would be better
than 74S for the upgrade of these cards.  Any opinions, please?

Incidentally, for certain 74H chips, there is apparently no substitute!
In the specific case of 7401 and 7400, the 74H01 and 74H00 are *NOT* pin-
compatible!  Do the newer logic families tend to follow the regular 74xx
chips or the 74H chips relative to these discrepancies?

Related topics:

SP chips were widely used in these cards as bus drivers/receivers.  The SP-380
is an inverting bus receiver, which was "upgraded" to the 8640 by Signetics,
and is a common Q-bus receiver chip.  However, at least the following chips
are "orphaned" and need sources of replacement:

SP-314 is an 8-input AND gate (NAND?) to decode the bus for device selector
applications.  All inputs are conditioned as in 8640

SP-384 is pin-compatible with SP-380, but it is *non*-inverting.  (One does
an OR function of two inputs, so the other does an AND (NAND?) of them.)

Less often seen, but occasionally needed are:

SP-317 and SP-391, but I don't remember the functionality.

Additionally, most output drivers where originally selected 7401 chips noted
by a leading select digit of 9 (97401).  This was upgraded to the 8881, which
is also a popular Q-bus driver.  In fact, there is an additional chip used
as a bidirectional driver called 8641, that combines some of the functionality
of both the 8881 and 8640.  It's used is the MM-8AB memory, and many Q-bus
cards.

An additional problem:  The SP-384 is involved in a CPU upgrade: the gradeout
digit specified internal to DEC was 5, thus the chips read 5384.  But this
turns out to be insufficient if the application includes certain fast
interfaces such as the RK8E, and thus requires upgrading the spec to 7384.

How are we to get a decent supply of these chips?  I personally will fund
part of a "spares" store if someone can locate a vendor for them.  I would
assume that the other chips, which are mostly standard 74xx chips except
for the designer series (see below) and the stated SP and related chips,
are readily available or can be substituted by one of the newer types directly.

Designer series:

Virtually all of the chips that start with the digit 8 are part of what was
originally designated by Signetics as the "Designer Series" of chips.  Mostly
they are standard TTL compatible with 74xx, but the functionality is different,
often allowing for a more sensible pin arrangement relative to the actual
way a chip gets used (avoiding land runs under chips, etc. or plate-through
holes, etc), or for more functionality in the package, such as a shift
register that has buss-driver worthy outputs or inputs, etc.  Most of the
chips have numbers like DS 82xx.  Note the "DS" in the front of the chip
numbers.  This causes confusion at most electronics vendors because they
are assuming these are Intel support chips' numbers, since the 4 digits
are identical (such as DS 8272 as opposed to 8272).

So, we need an ongoing source of all of these problem chips to maintain the
Omnibus boards.

Some core memory boards use some obscure hybrid chips, that are mostly
plane terminators, etc., and are essentially a DEC custom product, so if
they break, you lose, but all of these other chips are products that were
offered to the industry in a generic way, and it was probably reasonable
for DEC to choose them for use in their designs.  Clearly the concepts
present in the chips (high-impedance input with hysteresis, low-leakage
open-collector drivers, etc.) are what is in current usage, although likely
in even more compact chip cases.  Clearly, a new design for an Omnibus
peripheral could be accomplished with fewer total chips in any design, but
what of maintaining existing boards?

>
>Has anyone had any experience with what putting a board in
>backwards can do to the rest of the machine?
>

The boards really don't fit backwards, so someone must have forced it.

I do have experience with boards put in upside down!

The middle board of the RK8E board set says on it "TOP" for a very good
reason.  It can be inserted upside down because it has 8 finger edges and
no handles.  A field circus guy failed to heed this and then turned on
the 8/e.  Sparks flew from the board, and eventually we say about 8-9
destroyed lands on the board.  For those familiar with the RK board set,
they use smaller than average land size, and it is quite a pain to remove
chips from this board, even with a vacuum desolder.  Most people elect
to dyke the chips out and just remove the pins rather than risk ruining the
plated-through holes.

I eventually repaired such a card, using #30 wire-wrap wire as replacement
runs for the destroyed wires.  I replaced about a half-dozen chips before I
got it to work.  (Possibly a few were replaced in error, but you have to
guess a little sometimes; the RK has a few operations that are hard to
get to cycle to look at on a scope.)

A power supply fuse was also blown (I think +15), but no other damage!

cjl