recreating the PDP-11

From: pat_at_cart-server.purdueriots.com <(pat_at_cart-server.purdueriots.com)>
Date: Sat Apr 6 13:59:05 2002

On Sat, 6 Apr 2002, Ben Franchuk wrote:

> pat_at_cart-server.purdueriots.com wrote:
> >
> > In a project I'm thinking about starting this summer, I want to try and
> > re-create a PDP-11 (or maybe a -8 or something) using either SSI logic
> > and/or PAL/GALs. I would like to do the entire thing in SSI, but I fear
> > that I'd quickly eat up a lot of money on the project, so where necessary
> > I'd replace sections with PALs. Anyone have any good ideas?
> (snip)
> > 1) CPU card
> > 2) Memory - SRAM
> > 3) Front-panel switches
> > 4) Console serial port
> > 5) A few digital I/O ports
> > 6) Floppy interface (anyone suggest an easy to interface controller?)
> > 7) IDE controller
>
> Barring #7 can I sell you the nice 12/24 bit FPGA cpu I homebrewing. :)

I'm trying to avoid any more integration than a PAL or EPROM would
provide. Thannks for the offer anyhow. :)

> The PDP-11 is way too complex for SSI logic. A reduced PDP-8 style cpu
> would be possible with SSI logic. While not true SSI logic modern
> EEPROMS would be very useful for control logic. Byte magazine had a few

Yeah that might be a good idea.

> TTL designed cpu's. Stack based languages like Forth are easy to do in
> TTL. I would have done my FPGA computer in TTL but nowadays you can't
> find the 74LS chips I needed anymore like 74ls382 4-bit alu. One other
> option is 2901 bit slice parts. B.G.Micro seems to have a few in stock.
> If I were to redesign I would use 2901 bit slice parts and EEPROMS for
> control now that I know where to find 2901's.

I know it's not a conventional way to do it, but if I design my own CPU I
might just use a few EPROMS - at least for multiplicaion and divison, and
perhaps for addition and subraction. If I only do an 8-bit proc (very
likely), that's doable in a 256k*8 and a 128k*8 EPROM, running me only
less than $12 for 120ns parts and only 64pins in two chips to implement
all of the difficult ALU functions. AND, OR, etc should be easy and
cheaper to do with 74* TTL or CMOS IC's. Either battery-backed SRAM or
(E)EPROM for the state controller should be cheap, easy, and effective.

Mind you, I haven't bothered looking around for prices yet, so things may
actually be cheaper.

> A floppy disk controller interface is fairly simple but you need a fast
> CPU for the disk if some sort of DMA is not used. That is why my FPGA
> computer is not quite finished , I am squeezing in DMA for HD floppies.
> Also the front panel logic needs a bit of work too.

I might think about using an 80xx series DMA controller and a
semi-standard WDC floppy controller like those used in PeeCees so I can
get easy acces to docs for them. However, I might not bother adding in
the fdd stuff or hdd stuff - as long as I can get the code to fit onto an
SRAM or something I should be OK. I just have to figure out what I want
to do before I go any further.

I just

> If you really want to be authentic don't forget to add Refresh logic on
> the CPU for DRAM.

That's OK, I'll deal with the non-authentic feel of battery-backed SRAMs.
Anyhow, the SRAM will act more like some core memory. :)

-- Pat
Received on Sat Apr 06 2002 - 13:59:05 BST