According to two schematics I've seen (The 5160 and a Turbo XT Clone), pin 4 on that one 138 on my schematic isn't supposed to be tied to ground.
Honestly, I think you're just making your life harder by cross-comparing the schematics of a computer that could be completely different from this one to the partial fragmentary bits you know about this one. If you really want to solve this you need to get a complete and accurate schematic of what the select circuitry of *this computer* is so you can understand what needs to change.
Here's about all I can manage here to try to help:
Comment the First:
Read the datasheet for the 74138 and try to wrap your head around what it does, I think it'll help you a lot. The '138 is a "3-8 decoder/demultiplexer"; what that means in TTL-speak is it's a thing that takes a three bit binary number on its inputs (pins 1-3, A0-2 on your kicad symbol), and
if it's selected(* see below) sets the "O0-O7" pins so all but the one corresponding to that number are high. (Most TTL logic is "active low", and that includes the ouputs of this decoder.) That's all it does. If you slap one on a breadboard and wire 8 LEDs to the outputs and three switches to the inputs you can walk the switches through all eight possible positions and an LED corresponding to that binary number will light up; you will never have two LEDs lit up at the same time. If the chip is unselected
all will be off.
(* assuming here you wired the LEDs with the cathode towards the decoder)
... So what do the those E1-3 lines do? They're the select lines. Look at the kicad symbol: two of them have that little triangle on the pin, the other doesn't. The triangle indicates "active low". So in other words, to "select" this chip you have to have two of these lines pulled down to TTL low (or ground) and one pulled up to TTL high (towards VCC). Why the world does this chip have three select lines instead of just one? Because this was a trick a lot of chips did so you could cut down on the total number of decoders. For instance, let's say you want to decode 16 lines, not just 8. (IE, instead of getting 8 unique states out of 3 input/address lines you want 16 out of four.) If these chips just had one enable line you'd need a separate 2:1 decoder upstream from them to switch which chip was active based on that fourth input line. But because of these select options you don't need that; instead you can connect the fourth input to the "high" enable line on one of the chips and one of the "low" enable lines on the other (while tying the remainder to ground or VCC as needed), and which chip is active will flip back and forth based on the status of that pin. Again, this is a trick common among decoders and small memory chips/ROMs in particular, things that you might want more than one of without having to stack multiple layers of decode.
... Anyway, long and short of it, when you see a '138 with one of the select lines tied to ground or pulled up to VCC then all that is telling you is that line wasn't needed to decide whether or not to enable this decoder. It is extremely reasonable and common to see '138s wired like this.
Comment the Second:
If you're right about ROM select lines being what comes out of this chip then I highly doubt it has anything to do with the RAM circuitry. At all.
Comment the Third:
Again, let's talk about the goal of reverse engineering the schematics, and for that I'm going to focus on the labels you put on U42. You labeled all three A0-A2 inputs and half the E-lines as coming from "To_BIOS_ROM". Knowing now that those are all *inputs* then, well,
what exactly are those lines likely to be connected to on that BIOS ROM? Especially considering that the outputs of that '138 are largely connected to
I/O chips? Maybe I should leave this one hanging, because finding the answer will arm you as to what you actually need to know to understand what you're looking at.
(Hint: This is going to be related to the answer for what signals run between U23 and U64B, which then contribute to the decision about whether or not those ROM sockets get turned on by U44...)
