Looking at the PiPD 1 front panel, the one I have have on my 18 bit CPU
is horrid,
MIne
The Other guy
MIne
The Other guy
5 Likes
Don’t let other projects shame you. That’s a good, functional front panel on a machine you designed and built. Yourself. Like, one person. And you did it because you wanted to.
The Pi-DP1 is a copy of a commercial machine made by a commercial company, manufactured in a tax shelter. I don’t want to know how much it costs, but I imagine a lot more than yours. And I wonder how many people will buy it and never build it: roughly half the people I know who have these kits have never built them.
A laser-cut acrylic front panel, if you want to, could add a lot to your project. But it’s just perfect as it is.
2 Likes
I am kinda hoping for a case instead.
Most of the design here, was trial and error, a cold start panel
was a must just to see if the ALU and mother board works.
Later then I added a memory/io card with eeprom and I toggled a
bootstrap loader in. Later I got wise, and burned the bootstrap loader
into the CPLD used for data bus IO.
Most of the design was to explore the concept that 16 or 8 bit computers
are too small to handle byte and word data both in instruction space and address
size and 18 bits are ‘just right’.
The current CPU only has the carry flag, and I need to explore a version
with real flags.
I was hoping to port 16 bit software, but there really is nothing out there
in source, and what there is needs a full featured C compiler and libraries;
like PDP 11, or 6809.
1 Like
While the PDP-1 may be my all-time favorite computer, please don’t worry: these are entirely different things. A project like the PiDP-1 works only at scale, and even then, there are features, for which – as they put it –, “we decided against such things, they can all be done in software and the kit is already a package of 15 PCBs as it stands. And though PCBs are cheap, it does add up.” Yours, on the other hand, is unique, it’s your brain child, it’s truly yours, a product of your very ingenuity. And, as for looks, it has some going for it. There’s quite a number of prototypes of historically important machines that didn’t look that great.
(Even the PDP-1 console started with rather modest lab-like looks.)
Hi Ben, great to see a design from you ! Do you have schematics and/or a description somewhere ?
Doc’s that is the slow part.
The schematic is mostly memory and ATF1508 cpld’s, with cmos 22v10 's thrown in.
The HEX displays sure make life easy to display information.
Most issues have been with PCB and mother board layout, it works or it does not.
I can’t find the bug with the tools I have, so it is a trial and error design
.
Kicad I find is almost impossible to use since it went metric. I can’t get imperial
measurements to line up on a grid when placing parts or snapping traces.
I am still using two layer PCB’s, but 10 mill traces and spacing.
Snap grid 25 mils. Routing buses takes up most of my space.
Right now I am writing software to see, what improvements I can make.
So far I have made 3 wire fixes to the ALU PCB just because I had not
planned their use when I made the PCB a few months ago.
I am finding it is taking more hardware in the CPLD’s, than I expected
to make changes. Already I am at 102% usage in the control cpld.
Software wise it is going from basic 18 bit CPU like the PDP-7 to something
like the PDP-11. Most changes now are software.
When I designed this I thought I would be happy with 18 bits and a 18 address bus.
Now I am getting greedy and wanting 36 bits. Sadly it seems 36
bits will be 4x more complex than 2x design, mostly for a larger address space.
This is too complex for the cplds I can get.
Hi Ben,
In Kicad you have, at the left side, buttons for changing to imperial.
But I also had trouble aligning my parts, it seems much more difficult than it should be.
I looked-up the pdp7 instruction set, and it’s a scaled-up version of the pdp8. I now understand why you need several PLD’s, almost every instruction is totally different from another, and there are those weird ‘microcoded’ instructions as on the pdp-8.
Don’t understand why 36-bit would be more than 2x the complexity, unless you want to make a DEC-10 with all its instructions.
Most pcb houses can make 6 mil trace and clearance, that might make your buses more compact.
I suppose you simulated your design before you made pcb’s ?
Most of my problem is that Kicad it is for old school layout with the default footprints,
and default trace widths. To do anything you need to fight the system for thru the hole parts and older chips footprints and schematic layouts. I could do 6 mil, but a 4 layer board also then is needed.
One nice thing however is that is easy to zip up and send the layout to my favorite pcb fabrication place. It takes them a few minutes to give me a quote, and check the layout. A week later DHL is at my door with the PCB’s. I give them $20 for customs, and no other hidden costs.
Since I was doing a simple design, I did not need a full a microcoded control system.
For a larger machine I need to split the layout into several pcb’s and design for a bigger system.
I could build bigger, but it is new can of worms to have it all working.
A minor change like going to 20 bits, requires no changes in control, just a larger data path by 1 bit.
I never simulate, mostly because I don’t have the tools. Now looking at the microcode
I can make few minor changes, but nothing that will save macro-cells.
Pffft, I beg to differ, your front panel (and the rest of the build) is awesome, and made with handcrafted love.