Writing a binary (bcd) clock in Forth for RC2014 and Minstrel 4th

 A

nother idea I had for using the 8x8 LED Matrix is a binary clock.  It fits very neatly into 8x8 pixels 

(The 8x8 matrix looks way better in real life than it does in these videos. It's very difficult to photograph or film.)

Eventually I want to learn to use interrupts on the RC2014. An interrupt routine would update a counter very accurately and independently of the main program, which could be divided for accurate seconds, minutes and hours.

In the absence of that and no counter (as far as I'm aware) in CP/M (the OS I use on my RC2014) it was necessary to write a 'jiffy clock' that you'd call from your main loop. Here it is:

The idea is that you call clinc every 1/60 second and it gives you hrs mins and secs words for access to those numbers. If you can't call it exactly every 1/60s, then it allows for calibration and by coincidence, the main loop of my program happens to loop almost at that rate, hence the number 62 at line 27. 

I was trying to be clever here and use a recursive routine. That's a very neat way to update a string of bytes (eg for a score) where each counts up to a certain value before tripping the next. In this case we want to count each up to 60. Except that we don't. The first may not be 60 as it's used for calibration, and the last needs to cycle back to zero at 24. Hence the 'wrapper', clinc, which handles those things and uses the recursive word for seconds and minutes. For the Minstrel version, discussed a little later, I dropped the recursion idea, thus lengthening the code a bit making it more readable.

The full RC2014 code is available here: (see BINCL-RC.F and jiffy.f)  

https://github.com/shieladixon/8x8_matrix_toolbox/tree/main/forth

(NB I use DXForth, the code may need to be altered for your favourite Forth.)

Porting to the Minstrel 4th

I was looking forward to taking advantage of the counter (FRAMES) within the Jupiter Ace system. This is a 4-byte counter which is updated by the interrupt routine, so it counts steadily, independently of your program.

My code for 'reading' FRAMES, adapted from an example in the manual, is this:

You simply call time at least once per second and you can read the secs, mins and hrs variables. That works, and works really well if you're just printing the time to the screen, say.  But when used within the matrix refresh loop, causes quite a blink, because of all the double-length maths involved. 

The problem is that the frame counter is simply a counter, so calculating the seconds, minutes and hours from it is quite an overhead. 

I spent some time trying tricks such as dividing the work up or inserting matrix refreshes between the calculations. That ended up with messy code and still a flickery display. 

In the end I've worked in the same jiffy clock as the RC2014 version. The advantage there is that the counter maintains 4 bytes as jiffies, seconds, minutes and hours.  So those can be accessed immediately without causing a delay. 

My latest code for this is also at https://github.com/shieladixon/8x8_matrix_toolbox/tree/main/forth - see BINCL_ACE.F

[update] Both of these versions now allow the user to input the time when the program starts. The standalone executable for RC2014 is also there.