18 Sep
2013
18 Sep
'13
10:28 a.m.
I have PDP8 boards that used 7439 as well as 8881 and 97401 ICs.
--Bill
________________________________
From: David Riley <fraveydank at gmail.com>
To: General Discussion: On-Topic and Off-Topic Posts <cctalk at classiccmp.org>
Sent: Wednesday, September 18, 2013 9:34 AM
Subject: Re: substituting 7401 for 8881 in a PDP-8/E
On Sep 18, 2013, at 8:45 AM, Pete Turnbull <peter.turnbull at york.ac.uk> wrote:
My PDP-8/E stopped working at the weekend, which is a
bit of a nuisance as I need it working for the DEC Legacy event at Windermere in October
-- and I'm going on holiday on Friday so won't have much real time to deal with it
and prepare my talk before then.
In the meantime I've "fixed" the problem by replacing a dead 8881 bus
driver (E23 on the display board, drives D0-D3 from the switches) with an SN7401.? It
works with M8300+M8310 CPU, M8330 timing board and M8320 bus loads, but probably
doesn't have enough drive capability for a full system.? IIRC an 8881 should be
capable of sinking 70mA with Vol=0.8V, while a 7401 is only guaranteed for 16mA.? I've
calculated the terminators alone need 23mA plus around 1.6mA per load (20 loads in a
medium system?), and DEC sometimes used selected 7401s marked 97401 presumably for this
reason.
Anyone got practical experience of substituting for an 8881?
Practical, not yet.? Theoretical, I've got tons.? I'm working on designing
a QBus board (with plans for Unibus in the future) which would have used
the same drivers originally, and I'm having a hard time figuring out what
drivers to use that are currently available.? My best solution has been
to just use discrete SMD FETs on the output side and fast (sadly, expensive)
comparators on the input.
If you're desperate to get something up and running quickly and it needs to
fit an 8881/7400/7401 socket, you could always use an AND gate with a simple
BJT (in the US, I'd use a 2N3904) wired up in an unholy mess.? The problem
is that the 7408 has a different pinout than the 7401/8881 (the gates are
reversed).
Your other option would be making NAND gates with individual transistors:
stack them up so one emitter feeds to the other collector.? Obviously, you
need to put a resistor on the base.? If you have discrete enhancement-mode
FETs available, so much the better, but those aren't in most parts bins I
know of.
I'm assuming you don't have the time to spin a PCB, so this would require a
big mess of wires sitting in a socket.? If you have time to spin a simple
PCB, you could do the whole thing with small-outline parts (dual FETs with
the appropriate leakage/capacitance characteristics) are quite easy to come
by online) and make it fit in the original socket.? I suspect you'd be
running quite close to the margin, though.
I'm getting close to finding the time to close the loop on my prototype
Qbus board.? If anyone wants in on that, please let me know; getting more
parts in a go reduces costs drastically.
- Dave