The new mother board and front panel works after some issues with the sockets on the 22v10's.
Most of the re design was to use standard 8 resistor sips, 4.7K,2.2K,3.3K.
ATF1508's (15 ns) 128 micro cell CPLD's are used everywhere as well as CMOS 22v10's for programmable logic.
Historic parts like 74XX219's and 6850 ACIA's still can be found on ebay.
Now I need to box this up after I get some software written, like a file system.

The virtual computer went *Poof* into hyper space, and reality split.
The instruction set is almost stable, with just some changes in shift encoding.
Testing out the new mother board now with a hello world program.

The nice thing is I can change a mother board jumper and boot at octal 760,000 from the ROM
with no front panel.. Of course the front panel is needed to
bootstrap data into the EEPOM.
Ben.
Foot note, the shift changes are in and working. Can shift from 1..9

I have come up with a idea for a different order code for a 36 bit computer.
I plan to write a simple compiler for the order code and then emulate it on the 18 bit machine
I have now. Floating point routines need to be written, but other than that the coding looks simple.
Lion36 is the project name for now.

Got looking at the 8008 chip and thinking I could do something like that for a 36 bit machine.
Ta-da the 9090 cpu. I am having build issues with the version with IRQ's so it is likely I will not have them.
Since the CP/M and OS/8 (PDP8) did not have use them I don't that will be a issue for a simple user OS.
The next few days I will be loading programs via the front panel,and checking if the control logic works.
char (9 bits) and short (18 bits) are unsigned. long is 36 bits.
While brain dead in many ways, I do have 36 bit computer now,
PS. I found my IRQ coding had typo error,and it now compiles ok.

Had a build issue with the alu logic, not reading memory, and fixed a few bugs.
The 9090 is now the 9009 since I could not add direct addressing More testing for later in the week.
I hope I don't find more bugs, the computer design is packed to the gills already.
I will try to move halt around, so I have another instruction slot.

PS. The sun came out so I could finish the project. Set the clock to 13 mhz, for a 1.25 uS cycle time.

PPS.hmm all done, the moon is out. If no bugs I am done and have a working design with indexed and
register indirect. Slower than I liked, because of more don't care terms to make it fit.
Yes we have micro-chip design, late 1976 since it is a tad more complex than a z80.
Shades of vacuum tube computing, 200,000 additions per second.:
PPPS, Thought of clock hack to get some speed back by starting in state 2.

Saving all the old stuff.Moving from 18 bit stuff to 36 bits. Insert a photo of a room sized computer from the late 1950's. Tiny C is not a option for compiling since I have char , short and long,
Nor is regular C a option as I only have unsigned short and char.New year ,new language- Bengol 69 2026 Year of the CAT.
I have to rebuild the CPLD's since I am changing the order codes.
ps. after changing the order codes, it stopped working. Back to square 1.

New year, new bug. ADC (add with carry) never worked in any in the designs. Now fixed and backed up to CD rom.
PS: Revised the order code again, this time it builds and runs OK.

Busy working on a different file system and high level language parameter passing
Moved from the old IBM 2315 disk to a DEC RL01K. ~ 5.2 Mb for 16 bit data 4.9 Mb for 18 bit data.
Just got the basic text printing routines working, octal numbers,character or text string today.
Got the rom bootstrap working at 0760000.

.

Needed to revise I/O again. Packing register W (frame pointer) with the pc, hardware wise.
Now I have A/Z B C D PC/W X Y S.

Added lea (load effective address) needed with the frame register w,
and aea (add effective address) needed for indexing.
The kernel fprintf is now in for the console. Now I can start fleshing out the file system.
I most likely now will spend simple time writing a simple Algol-ish language.
C and Pascal are too complex.

A milestone, having the instruction set more stable.

Were you looking at writing a compiler or an interpreter? An interpreter might be simpler, many early micros came with built-in interpreters.
Something like Tiny Basic.
There is also Forth.
With ABCD and index registers maybe some 8080 code could be ported.

Grabs the book Porting the 8080 for dummies.
Downloads the book "K2FDOS a floppy disk operating system for the 8080."

There is really no file system design out there I can port out there. You had the dual floppy disc
with .25Mb per drive like cp/m written pl/m or big unix system with 64Kb/256kb ram, and 5.00Mb per drive written in C.
I need a high level language that I can bootstrap into, that handles long,short,character data types.
The 6800/6809 had flex, but there is no source code for that.

K2FDOS dos have the disadvantage of slow file deleting, but it has a small data foot print. In my case 8KB ram for
disc buffers, and 2kb? rom memory ignoring the basic BIOS routines.

Having the frame pointer, really makes writing subroutines easy.

Today I am trying a a experimental version to add signed chars and shorts.
The order code needs rearrangement, so every CPLD needs to be reprogrammed, and a wire added
on the CPU card for sign extend. Character immediate is still unsigned.
Foot Note. It compiles OK but does not work.
Gone back to the original version. The IRQ version does not work, but I don't plan to run with IRQ's in this
version of the hardware.
PS.The CPLD was defective,so I now have working version.

Had to drop K2FDOS for the simple reason I may want run a threaded FORTH, and that needs 1K blocks.
Gone to a FAT style file system, and that adds 8Kb more ram for the file index table (FIT). Since my media
is a whopping 1.95 Mb per disc, the fat table is reasonable size. Directories are 8.2 file names.
PS: Root directory now 2K, 128 files. The shell program is now 6kb.