ISA bus slave NS32016 processor board

Description

https://archive.org/details/ns32k-archive seems the internet's biggest archive of NS32K material.

There is software, and a scanned magazine with articles about building an NS32016 board called the Public Domain 32000. It includes circuit diagrams. It uses 256 kbit DRAM chips, and PAL16L8 chips with logic equations (yay!) but also uses a pair of rare DRAM controller chips (boo!).

I have OCR'd and HTML'd it, and uploaded it for people to download and read. In another language if your browser has translation abilities!

Details

The PD32, was published in Micro Cornucopia number 32, (October/November 1986), page 6.

Supports the 32016. Has 512k bytes of DRAM modules.

It is a complete processor board.

It has no bus interface.

It communicates through an 8-bit bus, but not a latch. I don't know how that works - do they wait state unti the other side has read the bus. The article talks about interfacing to the Z80 bus.
I'd prefer to get it talking to the STEbus or an FTDI USB FIFO module.

It comes with all the PAL equations and software to implement UNIX.

I could not find the DRAM controller chips on eBay, so I presume they are effectively extinct in the wild.
The design could be adapted to use modern SRAM chips. A pair of 512k x 8 bit chips would provide double the original RAM.

The byte-wide communications interface could be adapted to use Jonathan Harston's single-channel Tube API, and thus allow Acorn PANOS to run. But that's a job for people with more time than me.

I'm not that motivated to make a PD32, my hobby is mainly capturing electronic design details for posterity and people who want to make them. I don't have time to more than a few pet projects, nor do I have space to store stuff.

I had a quick go at assembling the ROM source and the PAL equations but failed. They need major syntax edits and checking.

Files

PD32 V1.0 Gerbers.zip

Gerber files for manufacturing

x-zip-compressed - 1.16 MB - 02/15/2024 at 20:23

Download

PD32-schematic.pdf

schematic

Adobe Portable Document Format - 766.96 kB - 02/15/2024 at 20:23

Preview

PD32-PCB.pdf

PCB layout

Adobe Portable Document Format - 682.59 kB - 02/15/2024 at 20:23

Preview

PD32-FRONT.jpg

3D render front PCB

JPEG Image - 194.45 kB - 02/15/2024 at 20:23

Preview

PD32-BACK.jpg

3D render back PCB

JPEG Image - 194.71 kB - 02/15/2024 at 20:23

Preview

Project Logs

Collapse

Tech data for build your own PD32
Andrew Lynch • 02/16/2024 at 14:31 • 0 comments

I've added tech data so you can build your own PD32 or at least the reconstruction of it. There are schematics, PCB layout, two BOMs, 3D rendering images, and Gerber files.

If anyone decides to make their own PD32 using the supplied Gerber files please post here with your results. Especially if any changes are necessary and/or you find improvements.

Would really like to see your results of a PD32 build. Thanks!
Cross-assembler recovery
Keith • 02/11/2024 at 01:26 • 0 comments

There is a Definicon cross-assembler but it is an MSDOS executable binary so I can't run it on my Linux PC nor can I compile it for Linux because there is no source-code.
There is C source code for one in Dr. Dobbs' Journal. The OCR text is not good because all the OCR tools I can use for free try to automatically recognise text in columns, with no way to specify single-column operation. Which mangles the order of source code scans.
Several days of tedious salvage work later, I've got the 32000 cross-assembler to the point where it compiles, but it crashes with a memory segmentation fault. Probably trying to do something that used to be okay on MSDOS but not on a modern OS.

I don't have enough spare time to get it working, but there are many capable coders out there, and I am sure one of them could get it running fairly quickly.

I notice that most of the code is doing lexical analysis, splitting the source into lines and words, turning words into symbols and numbers.

I have just started learning Python, and it has vast libraries for doing this kind of work. If I were to write one from scratch, I would start with Python to make the job easier and quicker.
PCB trace routing
Andrew Lynch • 02/07/2024 at 14:24 • 0 comments

The PD32 PCB routes in in FreeRouting as a 2-layer board, however, based on what I've read the NS32K chipset seems rather sensitive to signal quality. A 2-layer board is less expensive, but if a 4-layer design is more likely to work, then I think that's the way to go.
It seems there were some articles and/or letters in MicroCornicopia after the original PD32 article suggesting PCB quality caused a lot of problems to the initial roll out in 1986. They also used an improved 4-layer board to improve signal quality, so I think it is prudent to do the same.
The good news is the PCB easily routes as a 2-layer board so transitioning to a 4-layer PCB should be no problem. Just add two internal copper layers, one for GND and the other for VCC.

Component sourcing

Keith • 02/04/2024 at 17:01 • 0 comments

Octopart links to chips in BOM

13    U1, U3   PAL20L8 or better e.g. 22V10
14    U11,U12  27256    
15    U14,U15  74ALS573 Transparent latch, 8-bit
16    U2       74LS74    
17    U4       74LS32   
18    U5       74LS646 Octal bus transceivers and register
19    U6       74ALS00   
20    U7       74LS175   
21    U8       NS32202 Interrupt Control Unit
22    U9       74LS30   
23    U10      74LS04   
24    U13      NS32081 Floating-Point Co-processor
25    U16      NS32201 Timing Control Unit
26    U17      NS32016 Processor
27    U18      DP84412 Replace with PAL16R6 or better e.g. 16V8
28    U19      NS32082 
29    U20      DP8419  DRAM address multiplexer and refresh row counter
30    U21      74S32   
31    U22      74LS08

The PAL chips are no problem, I have original PAL chips and the logic can be ported to modern GAL chips.

UT Source have the DP8419 and DP8409

Unix System V software
Keith • 02/03/2024 at 22:03 • 0 comments

I feel completely content if a hardware project boots up and says "Hello world!".
After that, I hand over to software hobbyists. Hardware tends to have a completion point, software does not.
If you have the time and skills, it is possible to get this board running Unix, though I wonder what you could do to impress someone through a text-only interface.
Host PC software
Keith • 02/03/2024 at 21:28 • 0 comments

The board and Host PC communicate through an 8-bit bus, although the board does not have any 8-bit latches to store data. Maybe those would be on the Z80 board?
I'm pretty sure this is mainly the user interface terminal with the ability to transfer something like intel hex files. This should not be too hard to create.

I feel it could be replaced by a USB-FIFO type interface.
Toolchain
Keith • 02/03/2024 at 21:19 • 0 comments

The design includes the source code, but no binary image. We need a 32000 assembler for that. Which I thought would take some finding because the 32000 family never took off commercially. Thankfully, I was wrong!

http://cpu-ns32k.net/Crossass.html

I tried using this to assemble the PD32 firmware - without success. It seemed to want carriage returns with linefeed, and not have colons before equates. Conditional assembly didn't work. I think the syntax is significantly different, and I don't have a manual to tell me what it is. If it can't do conditional assembly it may not be up to modern standards.
Hex numbers are not the usual $FF, or C style 0xFF, but x'FF which is very strange. So the h'FF needs changing to x'FF.

I don't have time to investigate this at the moment.

It isn't hard to split the firmware into odd and even bytes when required.

Later, we will need a C compiler. Also hard to find. Here seems a good place to start:

http://cpu-ns32k.net/Gnutool.html
Checking the KiCAD PCB
Keith • 02/03/2024 at 21:13 • 0 comments

Start by asking what are the worst mistakes that could happen, and work downwards from there.
For example, check if a hole is in the right place before you check it is the right size.
I've seen boards with the wrong size, the components back to front or upside-down, so I know what to look for!
Checking the KiCAD schematic
Keith • 02/03/2024 at 21:04 • 0 comments

This involves printing it and the original, then marking checked tracks with a highlighter pen. Not difficult but you do have to be patient and meticulous. I can do this.

2024-02-05

I printed the old and new schematics on A3 paper.

First job will be to check the pin numbers of each part match.

2024-02-16

Checked final sheet today. All sheets now checked.
No discrepancies, although the original numbers header pins in an unconventional way.
The capacitor for BBG looked wrong (+ve to ground) until noting that it is the decoupler for a negative bias generator.
Original circuit omits power and ground pins, so I have not been able to correlate them.
Checking the KiCAD library parts
Keith • 02/03/2024 at 20:59 • 0 comments

I used to do CAD librarian work so I can certainly do this.

View all 13 project logs

Discussions

Steve Jones wrote 02/14/2024 at 10:54

Interested! I see most of the things I was going to mention over on the RetroBrew forum, except for Andrew Voropay's wiki page (http://wiki.sensi.org/dokuwiki/doku.php?id=ns32ktoolchain) about using old versions of GCC and binutils for cross-compiling.

Are you sure? yes | no

Andrew Lynch wrote 02/05/2024 at 11:18

Great mailing list for NS32K enthusiasts

http://mailman.berklix.org/mailman/listinfo/pc532/

Are you sure? yes | no

Andrew Lynch wrote 02/03/2024 at 17:25

I would like to build a reproduction PD32 board. However, this is an enormous project and will require multiple hands to make it work. I would like to assemble that team to do it.

I've captured the design as close to "as is" to the design described in the original MicroCornicopia article as I can make it. I think it would be easy to make some updates to the design, however, making modifications to a design which has not been through initial build and test is quite hazardous. Premature optimization is the root of all evil in my experience.

The PD32 hardware and software information exists (thanks to the generousity and wisdom of Dave Rand and George Scolaro), all we need are PCBs (easy to manufacture) and at least a couple of people willing to build and test. Preferably with NS32016 experience but at least the capability to bring up a new design to working condition and work around any errata from the original article design and/or introduced during the translation to KiCAD.

Also, a skilled software person would be needed to reconstruct the software available for the Host PC side and the Unix System V software. Please note, the Unix System V software is likely IP encumbered so that will affect any future distribution. Long term, that can be mitigated by using Minix or other free/open-source operating system. Plus, we'd need some documentation once we had working hardware and software.

In short, this is a huge project and way more than one person can reasonably do. I have brought the PD32 PCB hardware to the table and willing to share in build and test. However, I do not have actual experience with bringing up NS32016 hardware, so we need some help if this is to work.

If you want to help out, I have a few tasks that can be done before we even have hardware.

1. pin audit of the KiCAD (PDF) schematic against the original article. I have done this already, but another set of eyes would help ensure the best capture possible. Involves literally printing out the article schematic and the KiCAD PDF and going through connection by connection, pin by pin, checking them off to ensure they match. There are some discrepancies by the nature of the translation (intentional), but we are looking for the unintentional differences (errors) which would hamper build and test and successful bring up of the board.

2. datasheet audit of the components against the KiCAD footprints. The KiCAD design relies on an NS32K footprint library, and it likely has errors. Literally go through the datasheet for each part and compare to the parts in the KiCAD (PDF) schematic to ensure the pins match. I have seen this before where the datasheet and/or KiCAD footprint library have errors that get propagated into designs and turn up at the worst possible times. Major risk mitigation step.

3. Review and analysis of the 3D rendering of the PCB. Step back and look at the PCB. Are there things about it that don't make sense? Like the bypass capacitors, crystal oscillators, jumpers, missing labels, etc. All of those can have glaring errors obvious after build and test but maybe with a close inspection we can capture beforehand and increase the likelihood of a successful first time bring up of the PD32 prototype.

Further areas where expertise is needed:

1. Reconstruction of the toolchain. The design includes source code (ROM.A32) for the onboard ROMs. Find what toolchain the A32 source code is associated with and post toolchain step by step instructions. Probably something obvious but I have no clue. We need even and odd binaries to burn to EPROM (or EEPROM or Flash) to boot the system. We go nowhere without the ROM images. Super important!

2. Reconstruction of the PALs into GALs using PALASM (or GALASM). We need JEDEC files to convert the PLDs into something we can use. PALs are difficult to obtain these days (not impossible), but the design should be converted to use more common Lattice or Atmel GAL22V10 and/or GAL16V8 components. Especially the NS DRAM controller uses a component which is unobtainium (DP84412) but can be reconstructed with a fast GAL22V10. (I am guessing a 10ns part will work) Super important step. Like the ROMs, we are going nowhere without GAL JEDEC files.

3. Reconstruction of the Host PC software. Source is included but for what compiler? C of some kind presumably. Maybe GCC? The code would have to be translated to a modern C compiler (preferably SDCC or some open-source equivalent) and the binaries tested on target hardware. Again, super important step.

4. Reconstruction of the Unix System V software. This speaks for itself. It is a huge step and really the whole point of the exercise.

If you would like a challenging and historically relevant project to work, this is your opportunity!

Post here for replies and comments, questions, etc.

Are you sure? yes | no

ISA bus slave NS32016 processor board

Description

Details