The Fix

Last week, in 1802 Membership Card Debugging – Part 1, I had gotten as far as determining that while the CPU was running, the latches were not latching the output port or the address. Then I got to thinking about the testing I was doing. I was following the pinouts listed in the schematic in the membership card manual, and so I did not notice that something was different between the two chips. They were not, as I had thought, two latches! One is a 74HC273 positive edge triggered octal D-Flip Flop while the other is a 74HC373 transparent D-Latch. Similar ICs to be sure, but they have different operating characteristics.

Actually, I should have thought about this when I saw the O-Scope read out. It’s clear they are not receiving the same timing signals. I read the instructions about a dozen times while assembling this, and I had completely missed the part number difference. I also did not notice it on the board or on the chips themselves. Worse, I had them swapped!

So it looks like the thing I warned myself about back in 1802 Membership Card Build has come to pass. I had to do some desoldering on this tiny board. Luckily it was on the ICs and I was able to get the solder off fairly quickly with my handy desoldering iron. I had a little mishap with both chips though. I broke a pin off of one and the other just would not let go and wound up getting broken in half. Argh!

So I put in an order to Jameco for a replacement. The pair together came to about $1.70. Jameco has a minimum order of $20.00 without a fee, so I bought a few odds and ends to bring my total above $20.00. (Hey, I can always use more wire, solder, posts, jumpers,…)

The replacement chips arrived today, I soldered them in place, and bingo! Everything works!

Exploring the 1802

So now I have a working 1802 computer and I can enter programs and view them on the front panel. So then I had to a little bit of exploring. As one does with a blinken-light machine, I set to work writing a program to sweep the lights back and forth Knight Rider style!

This took a while because I have never coded for the 1802. As I started to explore, I realized just how different this processor is from other early micro processors. It has a plethora of registers, with 16 address registers which can be accessed via their high and low bytes. These function like 16 pointers into memory, any one of which can be used as the program counter. There are two registers, X and P, which indicate which of the 16 will be used for I/o and program counting respectively.

A little bit of a read-through of the manual plus the little fold out card reference for the instruction set was all I needed to set to work. I knew that the lights lived on output port 4. I decided to use R2 as a counter to delay the motion of the lights as was done in the Lee Hart’s slow blinking example. I was originally going to store the output byte in the high part of register 3, but then I realized there is no way to do register to port output. Rather, I/O appears to mainly be register-indirect memory access. This is definitely a processor designed to sling bytes around in memory!

So I reworked my program a bit more. I decided to store the display byte at 0x00FF, and then access it via register 4. My initial attempts had weird behavior until I realized that the OUTn instruction actually increments Rn after doing output. This is so you can very easily copy bytes out. (Weirdly INn does not seem to increment!) So I added a decrement after each OUT.

When it came time to design the logic, I realized that the shift left and shift right instructions only differ by 1 bit. So I made the program perform an ANI operation against the byte 0x81 and if that was non-zero I flipped the bit in the instruction. So this is a bit of self-modifying code. Dangerous, but short!

Oh, and did I mention I did all this by hand?

After my little bit of hand assembly, I toggled the program in via the front panel, and thus was another blinkenlight gif born:

Of course, a keen observer will note that there is one segment out on the displays. Probably just has a faulty solder joint (those were pretty tight.) I will fix that on another day.

Oh, and one other thing. I did test out the ROM. I had a little bit of trouble getting my terminal connected, but I did eventually get the monitor up and was able to run some BASIC programs. But I think I will write about that in another post.

Now to add my new blinking program to my CS castle. Thank you for reading along with my little adventure. I am going to keep playing with the 1802 and I’ll post some more programs and thoughts soon!

UPDATE: I finalized my build with a nice little enclosure and a few more toggle switches. You can read about my final build here: Finishing My 1802 Membership Card.

Computers   |   Home   |   Humor   |   Links