On Mon, Nov 13, 2023 at 02:07 AM, Dave McGuire wrote:
Ooohh! Please post pics of that when it's done.
Nothing fancy, Dave, except it employs a few different HP LED display units that I have waiting for employment: they are the (I think so, at least) HP 5082-7340 single-digit hex displays with the nice dots and rounded edges (HP 5082-7300-like, not seven-segment) for seconds, minutes, hours, date and year, plus one three-digit alphanumeric (5082-7100 or 1990-0476), showing the month in three alpha characters (Jan, Feb, Mar, etc.). Unfortunately, my '7100 is slightly damaged, which is why I'm looking for a replacement.
All this is not at all OT for this group, as I'm sure you'll agree.
So, that's 10 ea. '7340 plus one '7100. There are two rows of characters: The top row shows seconds, minutes, and hours, six digits total, the bottom row shows date, month and year. All that in beautiful red LED dots. The two rows are stacked with single-in-line IC connector pins and sockets, holding the two halves together. Case is two small boxes, the bottom one upside down.
Currently, things are half-finished. I'll post a few pictures some time in the near future.
I find it amazing that these days, a very accurate clock may be built using twelve of these '7300's, one Arduino and one half-a-postage stamp-sized GPS receiving module. No glue logic or the like needed, just a 5V/1A power supply. Actually, I'm using a buck regulator, allowing me to use a wall wart outputting anything from 6V to 12V @ 0.5 to 1A.
The 3 * 5 x 7 display has its own Arduino nano plus buffer/drivers for row scanning, character forming and column filling (very old hat, I love it). Using the clock's Arduino would cause visible interruptions in the row scanning, even when using interrupts, because of the amount of processing needed to prepare digits from the GPS' filtered NMEA telegrams, taking DST and UTC deviation into account and the processing needed to prepare column information for the alphanumeric part. Still, it needs some C++ programming since the Arduino compiler produces code which is too slow. I'm pretty sure that optimizing and doing just the row-enabling and -disabling under interrupt would be feasible but I decided to make life a bit easier still... As it is, it's two independent units, communicating a one-hex-digit month number.
TBC!
Raymond
