I would guess 95% chance your header pin solder joints are likely busted at 1j6 and/or 1j7. these sys 6 games PCBs have a terrible solder side. Otherwise 5% chance of a 74154 decoder or PIA issue.

The displays have this even when playing a game?

Redo your work on the 40 pin. If it was not there before, the work you did is the cause.

For me 0 0 0 on the displays is usually a driver board PIA problem. Seems weird but I've seen it multiple times.

Have you looked at the Pinwiki? http://www.pinwiki.com/wiki/index.php?title=Williams_System_3_-_7#Digit_Failures

Have you tried reseating the connectors on the master display? Both ends of the cables?

Logic probe the digit strobes. Look at the schematic. You will see a 74154 decoder. This decoder has 4 inputs and 16 ouputs. First check all the inputs for pulsing. Then check every relevant output for pulses.

No pulsing on an encoded input to the decoder, then lift the leg of the PIA and check to see if it is pulsing. If that PIA leg pulses when lifted from the PCB, the decoder is bad or there is a short. If the PIA leg is still not pulsing, replace the PIA.

If everything looks go, go to the master display board and check there to make sure these signals make it all the way to the display board.

Also a good idea to get the Andre test rom. Run the board and the bench and see what it says. Might save you a bunch of troubleshooting. The Andre rom has yet to fail me in identifying a bad PIA (RAM is a different story). This test ROM saves me a lot of time.

Another master display and another MPU to swap out would help a bunch too if you are not familiar poking around with a logic probe.

The fact you removed the board and it no longer works is troubling. Makes me thing you either damaged a conenctor (i really think this the most likely cause) or something got shorted when you had the iron out. Bright light, magnification, check your work!

Anyone ever figure out and document a truth table for that 4:16 decoder circuit? Might be able to narrow down to a single encoded input based on which digits light up.

I think you may be barking up the wrong tree with the 40 pin. All the display connection are on the CPU board.

Do you tried the questionable CPU and Driver board with a known working Master Display board - same problem.

Are you putting the questionalble boards in another machine that was working? If so, we can rule out the wiring.

Have you tried the CPU board with a known working Driver and Master Display board? If the same problem exists, then we can confirm the problem is not with the driver board (or 40 pin).

This leaves us to the CPU board once you confirm the above.

Start by carefully inspecting the 1J5, 1J6, 1J7 for bad solder joints. Even if they look good, reflow the solder on all these header pins or better yet, re-pin those headers.

Inspect R38 through R94 and C38 through C61 for damage (they are just below the headers mentioned above).

If these do not prove to find the solution, follow the procedure that barakandl mentioned above to determine if the 74154 (IC6) is at fault or the PIA (IC18) is at fault.

A little more detail to what Andrew provided:
The 74154 inputs are on pins 20, 21, 22, 23
The 74154 outputs are on pins 1-11 and 13-17

Time to logic probe the decoder and PIA or shotgun replace them. It is probably one of those two ICs based on your board swapping and connector replacing.

You can usually determine the 74154 decoder is bad with a DMM on diode test, board not powered. Red lead on ground, black lead probe all the inputs and ouputs. They all should read .400 - .700 ish. If any are short, open or below .200 reading the decoder is bad(make sure they are not tied to ground already, a few outputs might be that are unused, i forget). Not fool proof test though, but usually works.