Homebrewing a classic

From: Allison J Parent <allisonp_at_world.std.com>
Date: Mon Oct 5 20:39:45 1998

< I've been sketching ideas based on a rocker that is threaded on a rod. T

Smooth rod with a return spring and seperators between teh switches.

< back of the rocker contains a 'paddle' which when the rocker is down,
< breaks an IR/detector gap. This has a couple of benefits. One the rocker

No question it's more reliable... what about dust. How about hall effect
switches?

< switch register - what size? Could have 8 to 32 bits if we wanted

Number of bits is based on CPU native word size.

< I've about half decided to build something PDP-8 like out of PIC chips.
< They are relatively cheap, easy to program, and quite fast. I hadn't
< thought about using them as base units (microcode sequencer, ALU control
< etc) until this discussion began, but it also raised the point of

Doable but not many are needed as the 8 series had a single accumularot
and the amount of random logic that is not control related is small.

< 'emulators' and I began to wonder what if you build a Z80 system that
< emulated a PDP-8 with one of these custom front panels. That turns out t
< be more of a software effort than a hardware effort.

It was done on an 8080 in the mid 70s using a terminal as a text graphics
display.

< You can get PC boards made fairly cheaply, if you wanted to do a transis
< based CPU I would definitely consider having a few hundred 'flip chip'
< equivalents made. It wouldn't have to be 1:1 DEC replacements but they
< showed it could be done. Wiring up 12 flip flops by hand on a piece of p
< board (4 transistors each!) would be extremely painful.

Yes it would also the shear number even at a penny(US) each would scare
most people as something like an straight-8 used a few thousand.
Resistors aren't cheap nor caps so that scales up real fast. Transistors
if used would be a good thing for doing a serial machine as they have
lower parts counts and few hardware "registers". Not modern silicone
transistors do not have close to similar characteristics to the old
grown junction or MADT germainium devices so circuit chages would be
required.


< First, find a 50Amp 5 volt supply. :-) The thing about TTL logic that
< always amazes me is how much power it draws. I built a digital PLL out o
< TTL chips once (about 40 chips) it was easily drawing 3-4 amps at speed
< 74HC logic might be a good compromise here.

I've done some heafty TTL designs (one was 200 chips SSI and some MSI)
and it didn't need near 20 amps never mind 50. I don't thing the 8E PS
can do more than 35A.

Besides a simple design is possible without meltdown for heat.

< computers called the 'beer budget graphics display' and was featured in
< BYTE magazine in the late 70's. Basically this was two R2R D/A ladders
< driving the X/Y inputs to an oscilloscope. If you make a 12 bit machine
< could build a 4K x 4K resolution display fairly easily. (You will be cor

I built it using a 8bit R2R ladder and that is the limit of that design
for a lot of reasons. It settles slow (more than 10uS settling time)
and the high order drivers to the R2R ladder have to supply some power.
A later published design used 8bit moto1408, also slow.

Or scale back to 1024x1024. This is known as vector graphics and the
real limitation is how many vectors can you draw before you have to go
back and redraw (flicker/refresh) which is a fuction of DAC speed, CPU
speed and crt driver speed.

It was the display common to the PDP-1, and seen on 8s and even some 11s
for early graphics.

Allison
Received on Mon Oct 05 1998 - 20:39:45 BST