I'm repairing a Stellar Wars that has a System 6 board in it (shipped with a 4, I believe, but was replaced). I fixed all the issues that I can find and the machine is running well. On my list of things to improve it, I planned to replace the 3 3624s at IC21, IC22, and IC26 with one 2716 at IC14. I was going to do this because the old PROMs are old and heat up when the machine is played. Also, I thought if I just ditched those I would not bother to replace those last 3 sockets.

In the manual, it says:
Screen Shot 2021-05-12 at 9.20.25 AM (resized).png

No problem - PinWiki says the same thing:
For all other System 3-6 games, the ROMs that might be located at U26, U22, and U21 (left to right on the top ROM row) can be replaced with a single 2716 2kb ROM (or EPROM) at IC14 that contains the combined images. When doing so, jumper J3 should be IN, jumper J4 OUT. These jumpers are located just above the ROMs and to the right of IC15.

The only problem is that my board already has a jumper at J3 and nothing at J4:
IMG_2811 (resized).jpeg

In fact, my jumpers match the PinWiki photo for what it's supposed to look like with a combined IC14:
644px-WMSSystem6ROMConfig (resized).jpeg

I don't understand why this machine is running with the J3-IN / J4-OUT and 3 PROMs at 21, 22, and 26. That seems to be the config they want for a single 2716 at IC14. Are the jumpers just the same for both ways, and you can either populate (21, 22, and 26) or 14?
This is my first System 6 - I guess I don't know what I don't know.
Any help?

Also, not to derail my real question, but these machine are so inexpensive compared to other SS of the same year and I really don't understand why. Great sounds, tons of lights / solenoids, good flippers. I'm not sure why Stellar Wars goes for $600-$800 (or used to before).

Awesome - thank you both for the replies.
I will try a 2716 without changing anything and hopefully that makes the board a little more solid.

This one was fun! I think it's a good architecture. More components than they needed (and that connector is a big WTF) but really solid architecture.
I chased a really fun 6810 issue for a while. It appeared as a short (2.5 ohms) on the data bus between D0 & D1. But as soon as I touched a soldering iron to a pad, the "short" would go away. I would remove a socket, think the short was gone, and then after measuring it for a minute or two, the resistance would drop back down again.

Grounding the pins was temporarily fixing the issue until I measured it again for a minute (literally about 60 seconds).
Turns out that the right 6810 (soldered into the board) was bad. It charged up from the meter's trickle current (I guess!?) and turned into a short between pins 2 & 3 on IC16. I was going to put a socket there anyway, so no big deal, but it was really interesting when I was watching the poor thing try to boot with D0 & D1 making weird intermittent shorts. It would get down into the part of the code where it sets up the PIA chips (around M7018), and then get super confused.

... hijacking my own thread ...

Am I reading this right? It looks like they set the stack pointer to 7FD9 so they can do PULAs and setup the PIAs with bytes stored in a chunk at the end of the file? Once they're done with setting up PIAs, they LDS #M00FF, where it belongs.
I haven't seen the stack pointer used in that way before. Thought it was clever.

Screen Shot 2021-05-12 at 10.04.22 AM (resized).png

So cool - thanks for the link.
Disassembled code is a real window into the personality of the person who programmed. Fascinating.

I'm not yet - we'll see how this goes. I have a Stellar Wars up and running now and I think it's a really nice game. It's fast & flashy and has interesting goals spread around.