OP

Quoted from DumbAss:

So life goes on.

The kitchen sink plumbing had a small leak that could be controlled by the stop valve but eventually the stop valve failed. Take out a day or two to deal with this.
You know you are old when you wake up one morning with a sore, inflamed and swollen toe/foot but you don't remember any trauma to the area. You end up hobbling around the house as it slows you down significantly.
You get the initial SARS-CoV-2 vaccine in the afternoon and later in the evening the injection site at the shoulder starts hurting and you start feeling cold and shivering despite it being (relatively) warm. It means take it easy for the next few days to be sure.

Yesterday a board fabrication order arrived. There's a few revisions to a few things in the order but I'll focus on the one thing that some have been waiting for. It's for numeric/alphanumeric segmented displays.
In order to better understand the rest of this post the master/slave alphanumeric display system I built needs to be better explained. The system consists of a single slave panel that has enough LED blocks to be able to display any of the alphanumeric display systems used by Williams or Data East. It is focused on alphanumeric but can be adapted to numeric only as well. The slave panel has an upper and lower row of 16 alphanumeric LED blocks. It has a left and right row of 8 numeric LED blocks that correspond to the "jackpot" and "chips" displays used in Taxi, Police Force and Riverboat Gambler. These numeric displays can also double as "credit/match" displays used in High Speed, Grand Lizard, Road Kings, Pinbot and Millionaire. The master board drives the LED blocks on the slave panel. This particular slave panel has the 8P8C (RJ-45 or ethernet jack) connectors installed.
[quoted image]
This "universal alphanumeric master display board" (below) will drive the slave for any System 11 alphanumeric display protocol as well as Data East alphanumeric display protocol. There is a "bug" in this revision 00 board when used with the "discrete" configuration. This is the High Speed configuration. The credit/match digits are mirrored in the panel as well as being displayed on the small 7-segment displays at the bottom. The mode selection is based on a binary selection tree and it not very intuitive. I built this board for me because I got sick of swapping out the master board on my bench when I needed to test different systems. I never thought I would need more than the initial board I built. Turns out I was wrong and there is a (small) demand for something like this. This is the revision 00 board.
[quoted image]
I had had this board (below) made in October 2019 (!) and it sat around unpopulated and untested for that period of time. This board is an 8P8C master display board for Williams System 3-6 (6-digit) and Williams System 6A-9 (7-digit). It requires a connection to the slave board using an ethernet patch cable. It was built for experimentation and my initial portable display testing system. Coupled with an adapter it initially showed an issue with System 9 numeric display but I figured it out and got it work. For the astute reader this suggests a System 9 CPU board is in development because I need to have a way to verify a board on the bench and I have to put in all the required testing tools first. The System 3-6 side of this board (obviously) remains untested (as the 8P8C connectors are not populated). Once I get some more experience with System 3-6 I'll probably return to this. As it stands I have zero experience with these systems.
[quoted image]
With the small demand for the universal master alphanumeric and the working 7-digit numeric board I figured I would combine the two into a board that works with System 9 numeric and any System 11 or Data East alphanumeric. I also changed the mode selection to be a little more intuitive. The original binary selection tree method did not allow for any ambiguous or indeterminate state. This new selection is purely based on a simple switch to enable the mode. It does allow multiple modes and has no error correction for or detection of these states. That's the trade-off for a more intuitive mode selection. For Chris Hibler's usage scenario the board is mounted so that the mode selection is not easily reached being at the rear of the display panel. So I provided for a remote connection and a remote selection board. This allows for a simple wiring harness to remote the selection to a board that can be mounted to face forward.
[quoted image]
Alas ... this board has problems too. It mostly works. I traded one problem for another. At power when used in discrete or quad mode, the stray credit/match strobes are brightly illuminated until the blanking signal kicks in. It's also random. Some power on cycles will show the problem. Other power on cycles do not. I left it for a while to go do other things. Then I came back to seriously look at it. Long story short ... it turns out the change that I made for original revision 00 problem left a floating signal on a bus. A buffer/transceiver IC will take that floating signal and output a fixed/correct signal (that is not floating). As the input signal is floating it is random what the output signal will be. The revision 00 board does not exhibit this problem because there is no floating intermediate signal. That was introduced in this revision. So I added two pull down resistors to fix the floating signals and ground them. Grounding them fixed the problem.
That dealt with the alphanumeric side. I then went to test the numeric side and it was completely blank. Turns out that the signal on the 26-pin ribbon cable is BLANKING but the signal on 1J3-12 is ~BLANKING. I merged the signals on the master board so the blanking signal was reversed ... hence no display. So I had to cut the signal bridge. When I did this I introduced a floating signal. Good thing I learned how to detect floating signals above. I applied a pull down resistor to the ~BLANKING signal from 1J3-12 and corrected the connection to the strobe drives. Everything works.
[quoted image]
The revision 01 board also reversed the LED indicator status. When the mode was selected the LED was off. That left all the other LEDs on and it's very hard to easily see what the mode is. So I fixed that as well. Stupid mistake. Not paying attention to detail. This typically happens with me when I am impatient and push something through prematurely without taking the time to carefully review everything. At least two, three or more times is usually required to find most problems. Even then I still make mistakes.
[quoted image]
The board will remain unbuilt and untested for a short period of time as I need some more IC sockets (waiting on delivery) and I have another higher priority board I need to get complete and out of design. I am borrowing boards that need to be returned so these boards have to go through the process NOW rather than later. I like to work in a FIFO (first in first out) queue system but priority is priority which is why a linked list structure is used instead of a linear array.

OP

Quoted from DumbAss:

Heat up the pads.
Add some fresh solder to help the solder flow.
Keep the heat applied and when ready remove the heat source and "shake" the board to remove the solder.
Marvel at the perfectly clean hole without using a vacuum desoldering tool.

I have advised others do this but I don't think people have used it as a solution.